[House Hearing, 119 Congress]
[From the U.S. Government Publishing Office]
AI IN THE EVERYDAY: CURRENT APPLICATIONS AND FUTURE FRONTIERS IN
COMMUNICATIONS AND TECHNOLOGY
=======================================================================
HEARING
BEFORE THE
SUBCOMMITTEE ON COMMUNICATIONS AND
TECHNOLOGY
OF THE
COMMITTEE ON ENERGY AND COMMERCE
HOUSE OF REPRESENTATIVES
ONE HUNDRED NINETEENTH CONGRESS
FIRST SESSION
__________
JUNE 4, 2025
__________
Serial No. 119-23
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
Published for the use of the Committee on Energy and Commerce
govinfo.gov/committee/house-energy
energycommerce.house.gov
__________
U.S. GOVERNMENT PUBLISHING OFFICE
60-730 PDF WASHINGTON : 2025
-----------------------------------------------------------------------------------
COMMITTEE ON ENERGY AND COMMERCE
BRETT GUTHRIE, Kentucky
Chairman
ROBERT E. LATTA, Ohio FRANK PALLONE, Jr., New Jersey
H. MORGAN GRIFFITH, Virginia Ranking Member
GUS M. BILIRAKIS, Florida DIANA DeGETTE, Colorado
RICHARD HUDSON, North Carolina JAN SCHAKOWSKY, Illinois
EARL L. ``BUDDY'' CARTER, Georgia DORIS O. MATSUI, California
GARY J. PALMER, Alabama KATHY CASTOR, Florida
NEAL P. DUNN, Florida PAUL TONKO, New York
DAN CRENSHAW, Texas YVETTE D. CLARKE, New York
JOHN JOYCE, Pennsylvania, Vice RAUL RUIZ, California
Chairman SCOTT H. PETERS, California
RANDY K. WEBER, Sr., Texas DEBBIE DINGELL, Michigan
RICK W. ALLEN, Georgia MARC A. VEASEY, Texas
TROY BALDERSON, Ohio ROBIN L. KELLY, Illinois
RUSS FULCHER, Idaho NANETTE DIAZ BARRAGAN, California
AUGUST PFLUGER, Texas DARREN SOTO, Florida
DIANA HARSHBARGER, Tennessee KIM SCHRIER, Washington
MARIANNETTE MILLER-MEEKS, Iowa LORI TRAHAN, Massachusetts
KAT CAMMACK, Florida LIZZIE FLETCHER, Texas
JAY OBERNOLTE, California ALEXANDRIA OCASIO-CORTEZ, New York
JOHN JAMES, Michigan JAKE AUCHINCLOSS, Massachusetts
CLIFF BENTZ, Oregon TROY A. CARTER, Louisiana
ERIN HOUCHIN, Indiana ROBERT MENENDEZ, New Jersey
RUSSELL FRY, South Carolina KEVIN MULLIN, California
LAUREL M. LEE, Florida GREG LANDSMAN, Ohio
NICHOLAS A. LANGWORTHY, New York JENNIFER L. McCLELLAN, Virginia
THOMAS H. KEAN, Jr., New Jersey
MICHAEL A. RULLI, Ohio
GABE EVANS, Colorado
CRAIG A. GOLDMAN, Texas
JULIE FEDORCHAK, North Dakota
------
Professional Staff
MEGAN JACKSON, Staff Director
SOPHIE KHANAHMADI, Deputy Staff Director
TIFFANY GUARASCIO, Minority Staff Director
Subcommittee on Communications and Technology
RICHARD HUDSON, North Carolina
Chairman
RICK W. ALLEN, Georgia, Vice DORIS O. MATSUI, California
Chairman Ranking Member
ROBERT E. LATTA, Ohio DARREN SOTO, Florida
GUS M. BILIRAKIS, Florida YVETTE D. CLARKE, New York
EARL L. ``BUDDY'' CARTER, Georgia RAUL RUIZ, California
NEAL P. DUNN, Florida SCOTT H. PETERS, California
JOHN JOYCE, Pennsylvania DEBBIE DINGELL, Michigan
RUSS FULCHER, Idaho ROBIN L. KELLY, Illinois
AUGUST PFLUGER, Texas NANETTE DIAZ BARRAGAN, California
KAT CAMMACK, Florida TROY A. CARTER, Louisiana
JAY OBERNOLTE, California ROBERT MENENDEZ, New Jersey
ERIN HOUCHIN, Indiana GREG LANDSMAN, Ohio
RUSSELL FRY, South Carolina JENNIFER L. McCLELLAN, Virginia
THOMAS H. KEAN, Jr., New Jersey KATHY CASTOR, Florida
CRAIG A. GOLDMAN, Texas FRANK PALLONE, Jr., New Jersey (ex
JULIE FEDORCHAK, North Dakota officio)
BRETT GUTHRIE, Kentucky (ex
officio)
C O N T E N T S
----------
Page
Hon. Richard Hudson, a Representative in Congress from the State
of North Carolina, opening statement........................... 1
Prepared statement........................................... 4
Hon. Doris O. Matsui, a Representative in Congress from the State
of California, opening statement............................... 9
Prepared statement........................................... 11
Hon. Brett Guthrie, a Representative in Congress from the
Commonwealth of Kentucky, opening statement.................... 13
Prepared statement........................................... 15
Hon. Frank Pallone, Jr., a Representative in Congress from the
State of New Jersey, opening statement......................... 19
Prepared statement........................................... 21
Witnesses
Charles W. ``Chip'' Pickering, Jr., Chief Executive Officer,
INCOMPAS....................................................... 23
Prepared statement........................................... 26
Answers to submitted questions............................... 192
Ronnie Vasishta, Senior Vice President of Telecom, NVIDIA........ 38
Prepared statement........................................... 40
Additional material submitted for the record................. 44
Answers to submitted questions............................... 198
Jim Shea, Chief Executive Officer, DeepSig, Inc.................. 54
Prepared statement........................................... 56
Answers to submitted questions............................... 201
Asad Ramzanali, Director of Artificial Intelligence and
Technology Policy, Vanderbilt Policy Accelerator, Vanderbilt
University..................................................... 60
Prepared statement........................................... 62
Answers to submitted questions............................... 203
Submitted Material
Inclusion of the following was approved by unanimous consent.
List of documents submitted for the record....................... 112
Report, ``AI for connectivity: how policy makers can help
digitalisation,'' by Ian Adkins, et al., Analsys Mason, March
2025\1\
Letter of May 21, 2025, from Adams County Regional Economic
Partnership, et al., to Colorado congressional delegation...... 114
Letter of June 3, 2025, from Sharon Wilson Geno, President,
National Multifamily Housing Council, et al., to Mr. Hudson and
Ms. Matsui..................................................... 116
Letter of June 3, 2025, from Patrick Halley, President and Chief
Executive Officer, Wireless Infrastructure Association, to Mr.
Hudson and Ms. Matsui.......................................... 118
Statement by Matthew J. Platkin, Attorney General, State of New
Jersey......................................................... 120
Statement of Krystal Rawls, Director, California State University
Dominguez Hills Workforce Integration Network, June 4, 2025.... 123
Article of May 15, 2025, ``The House Is Close To Passing a
Moratorium on State Efforts To Regulate AI,'' by Adam Conner,
Center for American Progress................................... 126
Letter of May 16, 2025, from the National Association of
Attorneys General, to House Speaker Mike Johnson, et al........ 132
----------
\1\ The report has been retained in committee files and is included in
the Documents for the Record at https://docs.house.gov/meetings/IF/
IF16/20250604/118333/HHRG-119-IF16-20250604-SD13189497.pdf.
Excerpt, ``Federal Preemption of State Law,'' Bipartisan House
Task Force on Artificial Intelligence, December 2024........... 139
Letter from Community Innovation Partnership to Members of
Congress....................................................... 142
Letter of June 4, 2025, from National Digital Inclusion Alliance
to Mr. Guthrie, et al.......................................... 143
Statement, ``Consumer Reports opposes AI state preemption
language in House budget reconciliation bill''................. 151
Statement ofthe Council of State Governments, May 19, 2025....... 152
Statement, ``EPIC Opposes House Proposal to Ban States from
Regulating AI''................................................ 153
Statement, ``GOP Plan to Prevent AI Regulation Is Unhinged,
Dangerous,'' Public Citizen.................................... 154
Article of October 30, 2019, ``Not all robots take your job, some
become your co-worker,'' by Aaron Klein, Brookings Institution. 155
Article of May 28, 2025, ``Behind the Curtain: A white-collar
bloodbath,'' by Jim VandeHei and Mike Allen, Axios............. 160
Article of May 30, 2025, ``Trump Taps Palantir to Compile Data on
Americans,'' by Sheera Frenkel and Aaron Krolik, New York Times 170
Letter of May 13, 2025, from Tim Storey, Executive Director,
National Conference of State Legislatures, to Mr. Guthrie and
Mr. Pallone.................................................... 178
Statement, ``Open Markets Lambasts House Committee's Blank Check
to Silicon Valley Oligarchs''.................................. 180
Letter of May 13, 2025, from Encode, et al., to House Speaker
Mike Johnson, et al............................................ 182
Statement of Common Sense Media.................................. 185
Article of May 20, 2025, ``Critical Questions for the House
Hearing Examining a Federal Restriction on State AI Regulation,
by Liana Keesing and Isabel Sunderland,'' Tech Policy Press.... 186
AI IN THE EVERYDAY: CURRENT APPLICATIONS ANDFUTURE FRONTIERS IN
COMMUNICATIONS AND TECHNOLOGY
----------
WEDNESDAY, JUNE 4, 2025
House of Representatives,
Subcommittee on Communications and Technology,
Committee on Energy and Commerce,
Washington, DC.
The subcommittee met, pursuant to notice, at 10:31 a.m., in
the John D. Dingell Room 2123, Rayburn House Office Building,
Hon. Richard Hudson (chairman of the subcommittee) presiding.
Members present: Representatives Hudson, Allen, Latta,
Bilirakis, Carter of Georgia, Dunn, Joyce, Fulcher, Pfluger,
Cammack, Obernolte, Houchin, Fry, Kean, Goldman, Fedorchak,
Guthrie (ex officio), Matsui (subcommittee ranking member),
Soto, Clarke, Peters, Dingell, Kelly, Barragan, Menendez,
Landsman, McClellan, Castor, and Pallone (ex officio).
Staff present: Jessica Donlon, General Counsel; Sydney
Greene, Director of Finance and Logistics; Kate Harper, Chief
Counsel, Communications and Technology; Megan Jackson, Staff
Director; Noah Jackson, Clerk, Communications and Technology;
Sophie Khanahmadi, Deputy Staff Director; Brayden Lacefield,
Special Assistant; John Lin, Senior Counsel, Communications and
Technology; Joel Miller, Chief Counsel; Elaina Murphy,
Professional Staff Member, Communications and Technology; Dylan
Rogers, Professional Staff Member, Communications and
Technology; Jackson Rudden, Staff Assistant; Chris Sarley,
Member Services/Stakeholder Director; Hannah Anton, Minority
Policy Analyst; Parul Desai, Minority Chief Counsel,
Communications and Technology; Tiffany Guarascio, Minority
Staff Director; La'Zale Johnson, Minority Intern; Dan Miller,
Minority Professional Staff Member; Mary Ann Rickles, Minority
Intern; Emma Roehrig, Minority Staff Assistant; Michael
Scurato, Minority FCC Detailee; Johanna Thomas, Minority
Counsel.
Mr. Hudson. The subcommittee will come to order. The Chair
recognizes himself for an opening statement.
OPENING STATEMENT OF HON. RICHARD HUDSON, A REPRESENTATIVE IN
CONGRESS FROM THE STATE OF NORTH CAROLINA
Good morning, and welcome to today's subcommittee hearing
on examining artificial intelligence and how it is being used
in communications, technology industries. AI is top of mind
right now, not just in this country but all over the globe. It
has been used in different industries for many years, but
recent advancements in large language models, machine learning,
and generative AI have pushed this technology into the
spotlight, capturing public attention and forever transforming
how we live and how we work.
The applications for this new technology are widespread,
and we are continuing to find new ways that AI can be used to
benefit Americans' lives.
Whether you know it or not, almost everyone uses AI in our
daily lives, like when you use ChatGPT to create a shopping
list, or ask Siri for directions. Or even more specifically,
when you called the ride share to get here this morning, it
uses AI to find the fastest route based on traffic patterns.
Your cell phone provider uses AI to reduce harmful spectrum
interference on your phone, ensuring there is no lapse in
service. The entertainment industry uses AI to predict what
types of content viewers may enjoy and drives decisions on when
that content should be produced. It is being used to develop
content and enhance the editing process. Our military uses AI
to enhance efficiency with encrypted communications and perfect
precision with drones, like we have seen in the war in Ukraine.
Even the National Football League uses AI to create the
perfect schedule to limit unnecessary travel for players,
create an even playing field for teams, and maximize fan
accessibility for the biggest games.
As demand for AI grows, we must consider what physical
infrastructure will be required to continue advanced AI
development.
Storage capacity and energy consumption demands that data
centers are expected to skyrocket by 2030 due to increased AI
use. As data capacity increases, we will need robust fiber
optics and wireless connectivity to ensure powerful new AI
systems can reach their fullest potential and enable every
industry to grow.
But the United States is not the only country developing
advanced AI. China recently released its DeepSeek AI model,
which showed their advancements. Our adversaries will stop at
nothing to undermine our leadership in technological
advancement and utilize AI to threaten our very way of life. We
must continue to innovate and develop to prevent that from
happening. Competition in AI is a global issue, and it is
imperative that the United States maintains its leadership.
To do this, our country and Congress must encourage an
environment where AI companies can innovate, compete, and excel
on the global stage. Just like the light regulatory touch that
gave rise to the internet and some of the most successful and
cutting-edge companies on the planet, AI must be given the same
opportunity to ensure American companies set the standard for
the rest of the world.
This is an exciting time, and an opportunity to talk about
these issues. Navigating these new and evolving technologies
will not be without its challenges, but it is our job to meet
them head on.
Innovation has provided untold benefits to Americans and to
our economy. Today we will hear from our witnesses about how
artificial intelligence is being used across the communications
and technology industries, as well as what is required for the
United States to maintain its leadership in developing AI
models.
I look forward to hearing from the witnesses today about
these issues and how Congress can stand ready as a partner.
[The prepared statement of Mr. Hudson follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Hudson. I now recognize the ranking member, the
gentlelady from California, for her opening statement.
OPENING STATEMENT OF HON. DORIS O. MATSUI, A REPRESENTATIVE IN
CONGRESS FROM THE STATE OF CALIFORNIA
Ms. Matsui. Thank you very much, Mr. Chairman.
I am glad we are holding this hearing today, as
connectivity and artificial intelligence go hand in hand.
Already, AI is a part of everyday lives, from the improving GPS
driving directions to extending the battery life of our cell
phones and preventing network outages. As AI evolves, it will
transform how we communicate, improve network resiliency,
defending against cyber attacks, and supercharging connectivity
for consumers and businesses.
To realize AI full potential and ensure AI benefits us all
and not just a few, we must act with urgency to close the
digital divide by investing in the infrastructure and skilled
workforce that underpins AI's success.
Unfortunately, the Trump administration would rather pay
lip service to American AI leadership than act. The reality is
that they are undermining this exact goal by derailing our AI
supply chain with tariffs, gutting our AI talent pipeline by
attacking universities and slashing research dollars, and
weakening our AI infrastructure by freezing Federal
broadcasting funding.
Universal connectivity is the building block for universal
AI access. This includes the fiber networks that provide
reliable, scalable, and high-speed connections for AI
applications to process large amounts of data. To achieve this,
we must act quickly and fully carry out our Federal broadband
programs to connect the tens of millions of Americans who still
lack access to high-speed internet.
And that is why I am alarmed that the President continues
to sabotage the $42 billion Broadband Equity, Access, and
Deployment, or BEAD, Program.
BEAD is a once-in-a-generation investment from Congress to
expand affordable broadband. States are at the 1-yard line
ready to reach the end zone and get shovels in the ground. But
for almost 6 months, the Trump administration has put BEAD on
ice, blocking our States from connecting more Americans, all
while threatening to waste even more time with rule changes
that would undo the work our States have already accomplished.
To make matters worse, President Trump is weakening our AI
workforce through his cancellation of nearly $3 billion in
digital equity grants. Congress established these grants with
bipartisan support to provide communities with digital literacy
skills, training in technology, to reap the full benefits of
online access.
President Trump's attack on digital equity funds, including
trying to cancel California's $70 million grant, is leaving
behind our most vulnerable communities, including rural
Americans, seniors, Americans with disabilities, and veterans.
This is unacceptable. To be the global AI leader, America
also must lead on setting commonsense guardrails, responsible
and safe AI. Otherwise, it will harm innovation by damaging
consumer trust and weakening protections for a fair, open, and
competitive playing field for AI technologies flourish.
I have long championed policies that advance U.S.
leadership in AI and other emerging technologies. This includes
strengthening AI infrastructure from fiber and wireless
connectivity to our semiconductor supply chain. I have also
worked to preserve our States' roles in laboratories of
democracy, to provide as critical insight on AI policies where
innovation and competition thrive alongside commonsense
safeguards.
Now is the time to learn from our States and work on
bipartisan solutions to advance innovation and empower all
Americans to access the benefits of AI.
And that is why the 10-year AI State moratorium that my
Republican colleagues jammed through in their reconciliation
bill is so misguided and dangerous. We can't afford a 10-year
hold on our States' ability to identify and protect the
American people from AI-specific harms, not when AI is
developing rapidly and spreading to all parts of our lives, and
especially not without strong Federal AI safe guardrails in
place.
I look forward to hearing from our witnesses about how we
can strengthen our AI leadership. And with that, I yield the
balance of my time.
[The prepared statement of Ms. Matsui follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Hudson. Thank you.
I now recognize the chairman of the full committee, the
gentleman from Kentucky, for 5 minutes for his opening
statement.
OPENING STATEMENT OF HON. BRETT GUTHRIE, A REPRESENTATIVE IN
CONGRESS FROM THE COMMONWEALTH OF KENTUCKY
Mr. Guthrie. Thank you, Chairman Hudson, and thank you,
Ranking Member Matsui, for bringing us together for this
important hearing, and thank the witnesses for all of you for
being here and for your participation. I look forward to
hearing about all the ways the artificial intelligence is
currently being used and the promise that AI holds for the
future.
AI is a top priority for the Energy and Commerce Committee
this year. Committee's jurisdiction covers almost every layer
of the AI tech stack, from energy needed to power massive
amounts of computing power, to the fiber optic cable and
wireless connections needed to move the data through
interconnected networks, to the data itself which underpins
innumerable lines of code that together form the modeling
needed to generate outputs for the consumer commercial
application of AI technologies.
We started this Congress by holding multiple subcommittee
hearings related to AI regulation and used cases spanning all
the industries under this committee's jurisdiction, including
the committee's first full committee hearing on the existential
opportunities and risks of AI technology.
Today's hearing will focus on AI and the communications and
technology sectors where we are committed to supporting the
development of this transformational technology and to ensuring
that American innovation continues to set the global standard
for advanced networking and connectivity.
When ChatGPT came roaring into everyday life in 2022, it
provided a clearer, accessible example of the power of AI for
everyday consumer use. While it is a fantastic tool, it is only
one example of how AI can be used, and there are many ways that
AI technologies have been deployed over the last couple of
decades to support other applications.
For example, some applications use AI technologies to
detect and prevent spam robocalls. So hearing that is going
up--that is why I wasn't here at the very beginning--two floors
above us. A lot of times when we go home, we come back--most
people around here say, ``What are you hearing back home?''
Well, I can tell you robocalls is one of the number-one
priorities--of everything going on in this Congress and this
country, robocalls are one--and spam is used to prevent--I mean
AI technology is used to prevent the spam robocalls.
And there are other areas to implementing sophisticating
cyber security systems, to prevent secure--to prevent and
secure consumer data.
Law enforcement officers and other first responders in the
field can utilize real-time language translation to help assist
people facing language barriers. Even today, music artist Randy
Travis is taking full advantage of AI. After suffering a stroke
that reduced his ability to sing and speak, he used AI tools to
recreate his own voice from his own recordings to continue
producing new music.
I firmly believe that we are beginning to uncover the
possibilities that AI has to offer. America is uniquely
positioned to continue innovating as a global leader in this
sector. But this requires a commitment on our part. It is
vitally important that we hear from industry experts that is
enabled AI technology to develop rapidly and how we can allow
AI to continue to thrive in America while addressing potential
risk along the way.
At the same time, our adversaries are also developing
cutting-edge AI technologies in an effort to capture global
technological dominance. We cannot allow countries that do not
share our values to lead in technology as important as this. As
I have said before, we do not secure--if we do not secure AI
leadership, China will fill the void.
Europe is not going to fill the void. They made decisions
in their energy policy and their regulatory policy to eliminate
themselves from this transformational technology that improves
the quality of life of the people that live in our societies,
and we need to do it correctly. And if we cede development of
AI technologies to China, which as a nation, does not share the
same ethical values we do, we will also cede the ability to
shape future development of these technologies in a free and
democratic society.
As such, it is important that we take a measured approach
and strike the right balance between facilitating innovation
and providing principled guardrails where needed to address
gaps in current law.
As we have seen with Europe's approach, as I said, on
everything from energy production to data privacy, imposing
heavyhanded regulation of AI stifles innovation and stunts
economic growth.
If we can get this right--and we have to do this together--
there is no limit to American innovation and artificial
intelligence or the benefits it will unlock for all Americans.
I really appreciate our witnesses for being here. Mr.
Pickering, welcome back to Energy and Commerce Committee. I am
not sure there is any Member here on our side of the aisle that
served when you were here, but that wasn't that long ago. But
thank you for--maybe Mr. Latta might have served at the same
time you did. But anyway, we really appreciate you being here
and all of you being here, and we look forward to your
testimony. And I yield back.
[The prepared statement of Mr. Guthrie follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Hudson. Thank you, Chairman.
I now recognize the gentleman from New Jersey, the ranking
member, for 5 minutes for his opening statement.
OPENING STATEMENT OF HON. FRANK PALLONE, Jr., A REPRESENTATIVE
IN CONGRESS FROM THE STATE OF NEW JERSEY
Mr. Pallone. You notice, Chip, that he said only there may
not be a Republican, but definitely is a Democrat: me. But in
any case.
We have now had numerous hearings this Congress on
artificial intelligence, and we have heard about the benefits
and risks of AI. And while we continue to hold hearings and
debate the need to adopt commonsense guardrails to protect
consumers from bad actors using AI, some States have already
moved forward and adopted these laws providing basic consumer
protections from the negative consequences of some uses of AI.
But instead of learning from what the States are doing,
House Republicans last month passed a 10-year ban on a State's
ability to enforce their own laws protecting consumers from
AI's harms. And this provision, which was included in the
budget reconciliation, is nothing more than a giant gift to Big
Tech. And I hope--you know, we are working to see if we can get
the Senate parliamentarian to rule this out of order under the
so-called ``bird bath,'' but that is not because of the
Republicans' efforts. It is because of the Democrats' efforts.
The problem is this provision would block enforcement of
laws on the books right now that are protecting consumers from
real-world harms. Some States have laws requiring companies to
disclose when they are using AI. Others have laws protecting
against the use of deepfakes in elections and protecting
consumers when AI is used in healthcare, education, housing,
and employment.
Now, Republicans want to ban the enforcement of all these
State laws with absolutely no national bill ready to go to
address these concerns. Instead of enriching Big Tech, we
should be working toward strong Federal legislation to govern
and guide the development of these powerful AI systems which
are rapidly being incorporated into more and more aspects of
our everyday lives.
The Trump administration also continues to undermine our
progress in building the connectivity infrastructure needed to
power the AI models of today and tomorrow. For no good reason,
the administration continues to stand still in rolling out one
of the key demands of AI: that is fiber. Broadband programs
designed to bring high-capacity fiber to both data centers and
our homes are critical if America wants to continue to lead the
world in AI.
Any delays in connecting every home and business to
reliable high-speed internet only benefits our foreign
adversaries. America's strength comes from our ability to build
and deploy the most advanced technology here and then share it
with the rest of the world.
But this can't happen if everyone in America does not have
high-speed internet access. And I therefore urge the Trump
administration to get out of its own way and let the Bipartisan
Infrastructure Law's $42 billion BEAD Program move forward as
intended, letting the States deploy networks that are fast,
reliable, and can meet the technological demands for decades to
come.
Now, I also have to acknowledge that simply bringing the
internet to American homes will not allow us to lead the world
in AI. To complete the task, Americans need to understand how
to use AI. That is why House Democrats voted to include the
Digital Equity Act as part of the Bipartisan Infrastructure
Law. This Act funds programs that can help seniors, veterans,
the disabled, and others learn the skills needed to fully
participate in our digital economy.
Yet in the last couple of weeks, President Trump
unilaterally and illegally, in my opinion, decided to stop the
funding to these programs, falsely claiming that they are
racist. I can't stress, teaching grandparents and veterans, the
disabled how to use AI as well as protect themselves from scams
and scheduling doctor appointments and applying for jobs online
has nothing to do with race.
The Digital Equity Act recognizes the digital divide, and
it is not--it is not a racial divide. I mean, there is an
element of a racial divide, certainly, but it is because there
is so many people who are seniors and veterans and disabled,
regardless of their race or ethnicity, that need to know how to
use it. And it has nothing to do with race. And it is just
unfortunate that the administration is doing this.
Last thing I wanted to say, Mr. Chairman, is while we still
need guardrails to govern and guide the development of AI,
there is no question that AI has the potential to advance how
our communications networks serve the public. For instance, AI
models can be deployed in our networks to help enhance
resiliency and reliability so that when natural disasters hit
or other life-threatening events occur, Americans can rapidly
assess the damage and quickly get the help they need. And that
is why it is crucial that we fund the deployment of Next
Generation 911 across the country.
Again, it is a shame that the House Republicans want to use
spectrum auction proceeds to fund this giant tax giveaway for
billionaires and B corporations instead of helping fund Next
Generation 911, which is obviously what I have been pleading
all along. That is what we should be using that 80--$8 billion
in spectrum auction, and not to pay for the--well, you call
it--what do you call it? The Big Beautiful Bill. I call it the
Big Ugly Bill.
And with that, I yield back the balance of my time. Thank
you, Mr. Chairman.
[The prepared statement of Mr. Pallone follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Hudson. Thank you.
We have now concluded with Member opening statements. The
Chair reminds Members that, pursuant to the committee rules,
all Members' opening statements will be made part of the
record.
We would like to thank our witnesses for being here today
to testify before the subcommittee. Our witnesses will have 5
minutes to provide an opening statement, which will be followed
by a round of questions by Members.
The witnesses here before us today are Chip Pickering,
chief executive officer of INCOMPAS, and former member of this
committee, as has been noted. Welcome back.
Ronnie Vasishta, senior vice president of telecom for
NVIDIA. Thank you for being here.
Jim Shea, chief executive officer for DeepSig,
Incorporated. Thank you, sir.
And Asad Ramzanali, director of AI and technology policy,
Vanderbilt Policy Accelerator at Vanderbilt University. I
believe you have been in this room before too. Glad to have you
back. Thank you.
Mr. Pickering, you are recognized for 5 minutes for an
opening statement.
STATEMENTS OF CHARLES W. ``CHIP'' PICKERING, Jr,, CHIEF
EXECUTIVE OFFICER, INCOMPAS; RONNIE VASISHTA, SENIOR VICE
PRESIDENT OF TELECOM, NVIDIA; JIM SHEA, CHIEF EXECUTIVE
OFFICER, DEEPSIG, INC.; ASAD RAMZANALI, DIRECTOR OF ARTIFICIAL
INTELLIGENCE AND TECHNOLOGY POLICY, VANDERBILT POLICY
ACCELERATOR, VANDERBILT UNIVERSITY
STATEMENT OF CHARLES W. ``CHIP'' PICKERING, Jr,
Mr. Pickering. Mr. Chairman, Ranking Member Matsui,
Chairman Guthrie, all the Members on the committee, it is great
to be back to what I consider my home, where I served for 10
years on the Energy and Commerce Committee. I love this
committee. I love and respect deeply the role that it plays in
this institution and in setting the policies, the critical
policies, of this country.
I am the CEO of INCOMPAS. And to talk a little bit about
who we are and who I represent, we are the founding voice of
competition in network policy. We advocated over 40 years ago
for the breakup of the AT&T monopoly system, bringing the first
competition to the telecommunication networks and then, as we
went into the 1990s and into the current age, competition
across all networks and the internet.
And today, we have assembled something that is different
than and unique from all other associations in our space. We
now have members that are new energy companies, new SMR nuclear
fusion companies, traditional gas and gas pipeline transmission
and grid companies that are on one side of the data center. And
we have all the stack of data centers from small to regional to
hyper scale. And then we have all of the broadband networks
from fiber to fixed to wireless to satellite LEO systems.
And then on the other side of the market, we have all of
the leading American technology companies, but we also have the
new entrants, the new startups, the innovators, the
entrepreneurs that are creating the new AI models. So both on
the model side and on the what we now call AI infrastructure on
both sides of the data center.
And we think that makes us uniquely credible. To be honest,
we try to create a membership that matches your jurisdiction.
You are the only committee in all of Congress that is on both
sides of the data center and has the jurisdiction on all of the
major elements and questions of AI policy.
And you have a serious obligation and responsibility to get
this right. We are now in a race against China to win the
future of an AI economy, an AI-driven national security and
cybersecurity, and we need a national framework of AI policy on
each of the major questions to win that race.
So today I would--I want to start a conversation of how do
we do that? And how do we win that race? And there are two
major objectives: one is to maximize competition among all
models because, as you maximize competition, you get the
greatest investment and the greatest innovation. In this race,
unlike what we faced in the World War II and the nuclear race
and in the Cold War, the space race--those were Federal
Government-funded initiatives.
The AI race is primarily funded by the private sector and
private actors. So what we need to do is give the
predictability and the certainty of a national framework so
that the investments made to win that race will be made without
a patchwork of unpredictability and uncertainty.
We need to be able to build--this is the time to build the
infrastructure of a new age that will bring back American
manufacturing and bring enormous benefits in every sector of
our economy and for the healthcare, education, and workforce of
our country. And to do that, we need to build as fast as
possible. We need bipartisan permitting reform to clear the
obstacles that delay and cost our companies who are building--
fiber, wireless, satellite companies who are building the data
centers and are building the new energy and the new energy
supply and the transmission and grid that we need to power AI.
So I fully support the speed to BEAD, and the need to get
the shovels in the ground to build the broadband networks.
We need to infuse, through spectrum auction authority, new
spectrum into the marketplace. We need to do everything that we
can to close the digital divide. And so, Congressman Joyce, and
working on a bipartisan basis--how do we build fiber networks
across the railroad so that we can close the digital divide
without the railroad industry holding us up, delaying us, and
charging exorbitant prices?
It is critical that--in most communities, that you have to
still cross the tracks to reach the full community and to close
the digital divide.
In the Senate, there is a piece of legislation that makes
sure that BEAD is not taxed, so we don't need to tax the
broadband grants and defeat the purpose of the deployments to
every corner of the country.
Finally, INCOMPAS fully supports the effort of this
committee in creating a national framework to have a pause or
temporary moratorium that would give you the time that you need
to form a national framework around all AI issues. I want to
commend Congressman Obernolte and Congressman Lieu and the
bipartisan workforce that has created a set of recommendations
and one of the most extensive reports in Congress and the
leadership on a bipartisan basis. The moratorium is simply a
means that we don't have 1,000 State actions that could slow,
delay the investments we need on both sides of the data centers
and the full models and the full stack of the AI ecosystem.
This committee has had a rich history, from 1934, of
bringing telephone service to every American, electricity to
every American, and internet infrastructure, internet age.
Every time, it was this committee in this room that made those
policies. And we need this committee, under present bipartisan
leadership, to find a way to create the national framework so
that we win the race and bring all the benefits and secure our
national security and America's future. And I look forward to
working with everyone to that purpose and to that end.
[The prepared statement of Mr. Pickering follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Hudson. Thank you.
Mr. Pickering. Thank you.
Mr. Hudson. Mr. Vasishta, you are recognized for 5 minutes
for an opening statement.
STATEMENT OF RONNIE VASISHTA
Mr. Vasishta. Thank you very much.
Chairman Hudson, Ranking Member Matsui, thank you for--and
members of the subcommittee, thank you very much for the
invitation to speak to you today.
As said, my name is Ronnie Vasishta, and I am senior vice
president of telecom at NVIDIA.
NVIDIA, as you may know, is an American full stack
accelerated computing company, proud to be helping drive
American technology leadership globally. We have spent over
three decades inventing the technology that powers modern AI.
As you have heard from the Members' opening statements today,
AI is not just another app or algorithm. It is the engine
behind a new industrial revolution. And just like the roads and
electric grids of the past, the countries that build AI
infrastructure will reap the rewards of this next era.
At this critical point in time, there is also the need to
redefine the telecommunications infrastructure around the world
as well by leveraging AI. The convergence between AI and
telecoms presents an unprecedented opportunity for renewed U.S.
leadership globally.
But we must act quickly. Over the last few decades, the
telecom industry has evolved through generations of standards
known as 2G, 3G, 4G, 5G. And the industry is now working, as
you may suspect, on 6G, with a target completion date of 2030.
Now, while 2030 might sound like a long time to go, we are
actually--already the train has left the station, and we are
losing time.
The early deployments of 6G may start as early as 2028.
What is already clear is that whoever seizes the advantage in
the development and the deployment of AI-native 6G will win the
6G race.
The United States invented the foundational cellular
wireless technology, but today there are no--there are no
American wireless equipment providers. Now is the time to act
to regain the U.S. leadership in 6G.
And AI offers a number of advantages for the wireless
industry. First, AI can be applied to network operations to
increase energy efficiency, to enhance security, to improve
network resiliency, and, very importantly, increase spectral
efficiency.
Second, future networks will additionally support an
entirely new kind of traffic, not just voice, video, and data,
but AI traffic, the control and connectivity autonomous
vehicles, smart glasses, robotics, and many more applications
that we have yet to think of.
Third, putting wireless processing and AI on the same
infrastructure--that has never been done before--will enable
new economic opportunities for telecos.
AI networks need to be software-defined. This will enable
the same infrastructure to underpin both the telecommunications
infrastructure and the AI infrastructure. New features and
generations and new standards of wireless--think of this--will
be software updates rather than the requirement to completely
overhaul telecom infrastructure.
And fourth, AI will enable enhanced cybersecurity for
telecom networks. AI is essential for real-time threat
detection and automated remediation and incident response, as
we have heard.
AI can process massive data streams, can quickly identify
and neutralize attacks, whether they are occurring on a device
or at the network edge or in the cloud. This convergence of AI
and wireless infrastructure will fundamentally reshape the
global telecommunications landscape.
NVIDIA is working closely and actively with partners across
industry and academia to provide the tools and platforms and to
drive American innovation for this global ecosystem. Just for
an example, in the last couple of days we have been here with
other members of the AI WIN Project, which was announced in
just March of this year. Booz Allen Hamilton, Cisco, Mitre, the
ORAN Development Company, and T-Mobile were all working
together to enable this U.S. leadership. This project will
deliver American AI-native full stack software defined and
secure wireless platform that will enable that U.S. leadership
in 6G.
But ensuring U.S. leadership in next-generation wireless
networks requires industry and public sector to work together.
Congress and the U.S. Government can help ensure this
leadership by supporting R&D and continued innovation, by
working with U.S. companies together as we set and place our
requirements into universal global standards, and enabling and
promoting U.S. companies to win at every layer of the AI
infrastructure stack.
I am confident that together we can maintain U.S.
leadership in AI and regain leadership in wireless
communications through the development and global deployment of
AI-native wireless telecommunications networks.
Thank you very much for the opportunity to speak to you
today, and I look forward to all your questions.
[The prepared statement of Mr. Vasishta follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Hudson. Thank you very much.
Mr. Shea, you are recognized for 5 minutes.
STATEMENT OF JIM SHEA
Mr. Shea. Chairman Hudson, Ranking Member Matsui, and
members of the subcommittee, thank you for the opportunity to
be with you and testify today.
I am Jim Shea, CEO of DeepSig, a small business bringing
artificial intelligence software for wireless communications
and sensing to market. Today, I will outline how AI is
transforming 5G, enabling superior spectrum management, and
providing essential capabilities for national defense and
everyday applications.
DeepSig, Inc., headquartered in Arlington, Virginia, was
founded in 2016 by me and Dr. Tim O'Shea to evolve and bring to
market AI wireless technology that Tim developed at Virginia
Tech. The company's 48 employees are focused on delivering AI
sensing and communications software to meet commercial and
defense needs.
Working with partners such as Intel, NVIDIA, and Qualcomm,
DeepSig's AI-based OmniPHY software replaces traditional
wireless processing algorithms with AI and 5G systems.
When you make a call, the signal from your mobile travels
to a base station, often reflecting off of multiple buildings
and obstacles while competing with interfering signals.
Traditional wireless algorithms employ a simplified one-size-
fits-all approach to pull signals out of noise. AI, on the
other hand, can learn the local wireless environment by
monitoring received signals to rapidly develop AI models that
better sort signals from noise and impairments, dramatically
improving spectrum efficiency and reducing dropped calls.
This breakthrough extends to 5G massive MIMO technology,
flight panel antennas that focus wireless signals at individual
users. Traditional massive MIMO algorithms use a library of
fixed antenna beams, but AI can learn the best beam shapes and
directions for each user to optimize performance and the user
experience.
Taken together, these and other AI technologies are
referred to as AI, wireless, native wireless. AI-native is seen
as a key enabling capability in 6G as it evolves through the
standardization process.
U.S. leadership in AI-native, coupled with Open RAN Access,
ORAN, that replaces custom hardware with commodity servers,
will dramatically reduce costs and offer a path to compete with
Huawei and other concerning network vendors.
AI is enabling rapid sensing of the wireless environment.
Traditional sensing approaches are painfully slow and require
significant analysis by skilled engineers and weeks or months
of effort to develop new code when new signals are encountered.
DeepSig's OmniSIG AI sensing software has been trained to
detect, classify, and locate nearly all types of signals
ranging from narrow-band handheld radios to wide-band radars,
often up to 1,000 times faster than traditional approaches.
Like other AI systems, OmniSIG can learn new signals in a
matter of hours after being presented with new signal data.
Exponentially increasing mobile data needs for consumers,
industrial, and defense applications require more wireless
spectrum. Sharing with DoD and other incumbent users is under
study, but only on a limited basis. The CBRS band shares
spectrum with U.S. Navy ship-based radars. However, new
proposed bands have incumbent ground and airborne users that
move far faster than ocean-based ships. With the integration of
AI, rapid detection can inform commercial networks to vacate or
steer wireless beams away from the incumbent user.
Another important application involves addressing wireless
intrusion and base station spoofing. Persistent and growing
threats concern both network owners and the government due to
their increasing scale and sophistication. By continuous
monitoring of the spectrum, AI can detect fake base stations
such as those discovered last month in Turkey that were
covertly transmitting information about local leader population
back to China.
Finally, as we have seen in Ukraine, spectrum sensing and
other electronic warfare are capabilities where the U.S. must
regain leadership to protect our forces. Adversaries
continually change their signals and can only be countered by
timely detection. The ability to quickly learn new signals is
paramount to ensure our defense systems can quickly respond to
threats. Small, innovative businesses such as DeepSig deliver
off-the-shelf AI software, accelerating the likes of Anduril,
CACI, Lockheed Martin, and other partners, giving them the
ability to respond to rapidly changing threats.
Leadership in AI-native and wireless sensing technologies
offers the U.S. a path to reclaim global leadership in mobile
wireless technology. A special thank you to this committee for
your leadership and helping the NTIA Public Wireless Supply
Chain Innovation Fund become a reality. DeepSig has been
honored to receive three grants in partnership with Airspan,
Qualcomm, and Fujitsu, and is rapidly advancing our AI software
into 5G and 6G to take part in the resurgence of U.S. wireless
leadership.
The convergence of AI wireless represents a transformative
moment. The United States has been the innovative ecosystem,
talent, and industrial partnerships necessary to lead this
transformation. Working together, American industry and
government can ensure that the next generation of wireless
infrastructure and sensing technology is made in America.
Thank you for providing me the opportunity to testify, and
I would be happy to answer any questions that the committee
members may have. Thank you.
[The prepared statement of Mr. Shea follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Hudson. Thank you. Very well said and on time.
Mr. Ramzanali, you are recognized for 5 minutes for your
opening statement.
STATEMENT OF ASAD RAMZANALI
Mr. Ramzanali. Chairmen Guthrie and Hudson, Ranking Members
Pallone and Matsui, members of the subcommittee, thank you for
holding this important hearing and having me back to this room.
My name is Asad Ramzanali, and I am the director of AI and
tech policy at the Vanderbilt Policy Accelerator. I previously
worked in the tech industry and in government, including for a
member of this subcommittee. My testimony reflects my own views
and does not represent Vanderbilt or anybody else.
Today I will speak to three things: First, I will share a
framework for analyzing the whole picture of AI. Then I will
describe how to achieve long-term American leadership in AI.
And finally, I will describe this committee's history in
enacting guardrails to mitigate real harms from powerful
technologies.
To understand AI, many use the framework of a technological
stack. My written testimony illustrates one version of that
with four layers on how AI is used and developed. It is
applications, models, cloud computing, and chips.
Let me go through that. Applications, like chat bots, are
what you use. They sit atop models which are large pieces of
software that operate in cloud computing data centers that are
full of chips, often specialized for AI.
Of interest to this subcommittee, those data centers are
connected to high-capacity fiber. This framework helps
illustrate that each layer has distinct features and can be
analyzed for its distinct policy questions and has known policy
solutions.
Second. American leadership in AI is critical to our
geopolitical competition with the People's Republic of China.
Many interpret this to mean building larger versions of today's
AI models. However, just building larger data sets for today's
AI models will not yield a lasting national advantage. For
long-term American leadership in AI, we should pursue a
strategy based on our historic technological advantages, like
public investment in R&D, supporting startups, enabling all
Americans to benefit from technology, and mitigating its harms.
I detail these in my written testimony, but I will focus on
the latter two for now.
This subcommittee has led the charge to increased broadband
access and adoption. Put simply, America can't truly lead in AI
if not all Americans can benefit from AI. The bipartisan BEAD
and Digital Equity Programs are critical to closing the digital
divide, and they should get back on track.
Next, this committee should continue its tradition of
encouraging powerful technologies while mitigating real-world
harms through bipartisan legislation. During the second
industrial revolution, at the end of the 19th century, society
faced a different powerful technology. It was that of
railroads. Chairman Hudson, you know that Franklin, Virginia,
had its beginnings as a rail stop. Ranking Member Matsui, you
know that Sacramento was the terminus for the Transcontinental
Railroad. Farmers, however, got the short end of the stick for
a long time. Railroads charged farmers more for their short-
haul shipments than they charged large companies to ship across
the country. So States stepped in and they passed laws to ban
this kind of price discrimination. These laws didn't stop
industrialization or slow the technology. They mitigated a
real-world harm.
Based on State laws, Congress later stepped in to enact
these kinds of protections nationally. Specifically, your
predecessors on this committee passed bipartisan laws, like the
Hepburn Act, named for the former chair of this committee, to
require fair terms for farmers, price transparency for small
businesses, and an end to vertical integration that harms
competition. I tell this story to illustrate that State and
Federal laws can encourage positive aspects of a powerful
technology while mitigating its harms.
Just as they did with railroads, States today are leading
the charge to mitigate the harms of AI. For railroads, Congress
passed Federal laws commensurate with State protections. On AI,
the House passed a 10-year moratorium on State guardrails. This
is a different path representing a major policy shift in AI and
from how powerful technologies have been regulated in the past.
This moratorium would wipe away real guardrails protecting real
people and strip millions of Americans of rights promised to
them by their State lawmakers without commensurate Federal
protections.
This doesn't mean Congress shouldn't act on AI. As I said
earlier, investing in R&D, supporting startups, those are
necessary. The bipartisan House framework that Mr. Obernolte
and others have put together has a lot of really good ideas for
policy.
This committee should also reinvigorate bipartisan efforts
on privacy and to protect kids and work with the administration
on advancing BEAD and Digital Equity Programs.
In closing, thank you for inviting me to be here with you
today, and I look forward to your questions.
[The prepared statement of Mr. Ramzanali follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
Mr. Hudson. Thank you very much.
We will now begin the questioning, and I will recognize
myself for 5 minutes.
Mr. Pickering, while advancement of AI has so much
potential to improve efficiencies across many industries, it
can also be used maliciously and cause serious damage. While we
have seen a few instances of placing measured safeguards on
certain uses of AI, we must balance any regulation with
enabling innovation.
Can you also give us an example of how Europe has inhibited
industry from advancing AI due to their overbearing
regulations?
Mr. Pickering. Now, if you look at the European model,
which has shifted a lot of the AI investment to America and
given us a comparative competitive advantage, it doesn't
allow--it is more of a permission-based system versus a risk-
based system. And I think America should continue its tradition
of light-touch, risk-based rules and regulations at a national
level on something of a national priority that needs a national
framework.
And so if you look at how we have done in the internet age,
setting a national framework that allowed full competition of
open internet that--whether you are a small business or a large
business, you can have equal access to the internet--the same
is going to be true on the AI models. From open-sourced to
closed and proprietary, if you have a gatekeeper that is going
to check whether a new entrant that will be punished the most
from a regulatory framework of a European-type model, the large
tech companies will be uniquely positioned and the resources to
manage the complexity of whether it is 50 States or one
regulatory framework that would be modeled after that--our
approach, I think, is the best, which is maximizing
competition, not regulating in a way that slows innovation, the
investment in the private sector, and letting all models
compete against each other as quickly as possible.
And if we do that, we will be successful. And so thank you
for the question.
Mr. Hudson. That makes a lot of sense.
Now, so one of our concerns is having a patchwork of State
regulations. You mentioned this in your testimony. Are you
familiar with Colorado and kind of what is going on with their
State regulations?
Mr. Pickering. Yes. And the former member of this
committee, Congressman Polis, has raised concerns and
objections about what his State law and legislature has done
that really could stifle investment in the Colorado economy and
investment in what the AI economy means.
So whether it is Colorado--there have been--and Congressman
Obernolte brought this up in the reconciliation markup--1,000
different pieces of legislation across the 50 States
introduced. If you go back to this committee's history, it did
The Cable Act in 1992, the wireless spectrum auctions in 1993,
The Telecommunications Act of 1996, the commercialization of
the internet, privacy, COPA, copyright privacy, internet tax
moratorium. So every category of the internet regulatory
framework, it was done in this committee. It was done by this
leadership on a bipartisan basis.
And that is what we need today if we are going to win the
race. We cannot afford the delay and the unpredictability of a
patchwork approach when we need--just like in the Cold War and
the nuclear race in World War II, we need a national urgency of
getting a framework in place. And again, I go back to
Congressmen Obernolte and Lieu and the leadership. There is
enough bipartisan commitment and consensus on this issue.
There's a lot of issues to fight over in the country and in
Congress, but on--AI policy is a place where we can find common
ground and a sustainable policy that is of national urgency and
imperative.
Mr. Hudson. I agree with you. I think this committee is
prepared to work in a bipartisan way to address this.
What advice would you give us, though, to make sure that
any regulations we do are reasonable and they don't stifle
innovation?
Mr. Pickering. You know, the market of maximum competition
has worked in every decade. So if you think through Ronald
Reagan's breakup of the AT&T system in 1982, and then you had
long-distance competition that replaced copper analog networks
with fiber digital networks. Then this committee passed The
Cable Act, which brought cable and satellite into competition
with one another. One was digital, one was analog. The
competition made both of them build out the last mile of
infrastructure that allowed the internet to be commercialized
and reach every American. The 1996 act, full competition of all
networks of everyone competing against each other with
interconnected, interoperable networks and devices, works.
So the only recommendation that I would give to this
committee is, whatever you do, use maximum competition to
achieve your objective. And that is the best way to regulate
the market and to give consumer protection, and also to give
you the innovation at this table for national security,
cybersecurity, health, education, workforce solutions.
Mr. Hudson. Right. Thank you for that. My time has expired.
I will now recognize the ranking member, Representative
Matsui, for 5 minutes for your questions.
Ms. Matsui. Thank you very much, Mr. Chairman.
Our State and local governments are on the front lines of
leveraging the benefits of AI while protecting consumers from
AI-specific harms and ensuring workforce protections keep pace
with rapid technological change. As home to 32 of the world's
50 leading AI companies, California is a national leader in
ensuring that innovation and competition thrive alongside
commonsense safeguards.
Mr. Ramzanali, how would the 10-year AI State moratorium,
as passed by the House Republicans, impact U.S. AI innovation
and protections from millions of Americans?
Mr. Ramzanali. Thank you for that question, Madam Ranking
Member. And I think your State of California has done great
work, as you said, both on encouraging innovation and also
mitigating its risks.
The way I like to think about it at the most simple level
is responsible innovation shouldn't be afraid of laws that go
after responsible practices. The kinds of State laws that have
been passed go after deepfakes. They go after scams. There is a
lot that require transparency as well. But that is how we know
when issues pop up.
So, to me, there is a great way to do responsible
innovation while also mitigating harms.
Ms. Matsui. Thank you very much. I always hear about--I
hear about moratorium and I hear about competition and
innovation. They don't go together at all, as far as I am
concerned. And so those are two sides of it that we have to
really deal with, and I am sure that this conversation will
take place on this subcommittee.
To lead the world in AI, America needs fast, reliable, and
futureproof networks to power AI-driven data centers, networks,
and homes.
Mr. Pickering, you mentioned the need to expand fiber
networks to reach rural and underserved communities at risk
being left behind in the AI economy. Why is this so critical
for AI innovation?
Mr. Pickering. So the connectivity to every American gives
the greatest opportunity for every American and to every small
business, every community. It doesn't matter if it is my home
State of Mississippi, which is primarily rural, or your
district, which is urban. Connectivity, in today's world, is
employment. It is opportunity. It is the greatest way to get
both the opportunity that America has to offer--the education,
the workforce, the healthcare--and so being able to have
universal access to the fastest, highest-capacity networks, it
is a national commitment that members of this committee and
Congress made coming out of the pandemic.
Ms. Matsui. Absolutely.
Mr. Pickering. We kept all Americans connected during the
pandemic, and from that a commitment to bring broadband to
every community--
Ms. Matsui. To every household in America. Right.
Mr. Pickering. And I agree with you. We need to speed BEAD.
I agree with Chairman Hudson.
Ms. Matsui. Right.
Mr. Pickering. We need to give clear guidelines, cut the
red tape, get shovels in the ground, and----
Ms. Matsui. There are States waiting right now, ready to
go.
Mr. Pickering. Yes.
Ms. Matsui. So I think we need to move forward.
I work hard to advance policies that support the
development and deployment of open radio access networks--or as
we call it, ORAN--including NTIs helping Wireless Supply Chain
Innovation Fund under the CHIPS and Science Act. Open RAN
increases supply chain diversity, which has significant
economic and national security benefits.
Mr. Vasishta--right? And Mr. Shea--how do technologies like
Open RAN help us leverage AI technologies for next-generation
connectivity and maintain U.S. technological leadership?
Mr. Vasishta. Thanks for the question. Maybe I will take it
up first.
So, as you quite rightly said, Open RAN was enabling
interoperability in a system that was traditionally propriety
in a closed system. What Open RAN was able to do, by enabling
that interoperability, was enable new players to come into the
industry as well as enable some of the interfaces to be more
openly developed.
What we need to do now is really kind of accelerate Open
RAN and make Open RAN competitive. One of the challenges also
around Open RAN was the competitiveness of that compared to the
proprietary systems. And so that first step that was made with
Open RAN now, with what is called AI RAN----
Ms. Matsui. Right.
Mr. Vasishta [continuing]. Which is including AI into those
open standards as well, and enabling the network to become
software defined--so you are completely writing new features
and capabilities in software. You can now open up that
ecosystem still further as well as create more competition in
that ecosystem, while at the same time taking advantage of AI
to improve the spectral efficiency and operational efficiency.
Ms. Matsui. OK. Thank you.
And I want to ask Mr. Shea to----
Mr. Shea. ORAN enabled our business in 5G and 6G. You know,
when we first started the company, we traveled around to many
of the vendors and we had great conversations, but they didn't
want to give us any access to their code. With ORAN, working
with initially Intel and now NVIDIA, we are able to actually
build operating 5G base stations. In fact, our headquarters in
Arlington, Virginia--we have two operating 5G base stations
that are built based upon ORAN and are enabled with our AI
technology, something that we couldn't have done previously
with closed systems.
Ms. Matsui. Thank you very much. My time has run out. Thank
you.
Mr. Hudson. Thank you.
The Chair now recognizes chairman of the full committee,
Mr. Guthrie, for 5 minutes----
Mr. Guthrie. Thank you. Thank you. And the ranking member
and I have been partners in spectrum, and--I am over here on
this far end--spectrum and other things, and she just asked the
first question I was going to ask. So we think alike. And there
is a lot of cooperation on both sides of the aisle in trying to
figure out how do we defeat China. All of us want to defeat
China to make sure that we are the platform that the world uses
for AI.
So, Mr. Vasishta, we have been focused on--almost every
hearing that we have had has been focused on how do we beat
China to AI. It is all subcommittee--anything that brings all
the jurisdictions of the committee together, it is AI. And so
the big part is energy. A lot of guys will tell me that--guys
and ladies, I say ``guys'' generically--but men and women will
tell me that in this industry that we have the brain power, we
have the capital. We need the energy and we need the regulatory
structure where we can succeed.
And so in terms of energy is producing the energy, but it
is also being more efficient with energy that we have. If you
look at the delta between China's production of energy and
ours, it is scary, and we have to catch up. But we also have to
get better with the energy that we have.
And so would you talk about, as a leading chip manufacturer
in AI software, how NVIDIA is using advancements in chip
technology and AI to improve data center efficiency in energy
consumption?
Mr. Vasishta. Yes. Absolutely.
Obviously, that is a very important point when it comes to
the deployment of AI, but it is also a very important point in
the design of AI.
AI, as we see, is generated from a hardware infrastructure
to do processing, networking, and many other functions. And as
we move down that technology curve, what we are able to do is
improve the efficiency and capability of those chips.
At the same time, what we have been able to see in the last
few years, by using accelerated compute as opposed to the
traditional compute, say, with CPUs only, we are able to
achieve much higher energy efficiency. Think of it as
performance per watt. Accelerated compute really has given the
enablement--led to the enablement of AI. And energy efficiency
is really--is created by that accelerated compute such that we
can achieve the functions we want to achieve in the silicon
technologies that we have.
Mr. Guthrie. So my question--so that--thank you. That was
my question., but also added to that, if you look at the gap
between what China's producing in energy and we are producing
in energy, can we make that up in efficiency alone? Or is it
going to require us producing more energy?
Mr. Vasishta. I think there is going to be a requirement
for more energy. Energy is, you know--just the massive scale of
deployment to be competitive in AI will require more energy
even though we are improving energy efficiency of the
infrastructure itself every generation also.
Mr. Guthrie. Well, thank you. Thank you.
I had a question for Mr. Shea, but that was a question that
Ms. Matsui asked. So I may get back if you want to follow up
because I think you ran out of time.
But let me go to Mr. Pickering first. Can you elaborate on
what kind of growth our networks will require to support
American innovation in AI, and how soon this infrastructure
needed--how soon is this infrastructure needed so the United
States can compete?
Mr. Pickering. Mr. Chairman, so if you think about fiber
capacity to the data center, one of my member companies, their
CEO recently said in the next 5 years, fiber capacity to the
data center is projected to multiply 6 times. Now, if any of
you all were to go to the Northern Virginia data center hub,
and if you were to realize the abundance of fiber capacity into
that data center hub is enormous--to comprehend six multiples
of capacity demand growth is really hard to emphasize how
important that is.
And I see Chairman Latta here as well. The same thing is
going to happen in our grid, our transmission, our energy
supply and generation. And we no longer look at a separation
between the energy market, the data center market, and the
fiber market. It is converged into one AI infrastructure
market.
And those inputs, going in and meeting at the data center--
if you look at my home State of Mississippi, which has always
been last in economic growth, this year they are second in GDP
growth because of huge investments that have been made possible
by a fiber route built by Zayo, one of our member companies,
that went from Atlanta, Georgia, to Dallas. It passes through
Birmingham; Meridian, Mississippi; Jackson, Mississippi;
Vicksburg; Monroe. That has become AI alley, with tens of
billions of dollars of data center investments, Birmingham,
Meridian, Jackson, Vicksburg. It is causing all of our energy
production--grids, transmission--to see massive growth and
upgrades.
Mr. Guthrie. Thanks. My time--I know you know the rules
here. My time expired----
Mr. Pickering. You got me excited because the growth in the
economic development comes with all this new capacity.
Mr. Guthrie. It is exciting. We are looking forward to
working together on that. Thank you.
Mr. Hudson. Thank you, Chairman.
The Chair now recognizes Representative Soto for 5 minutes
for your questions.
Mr. Soto. Thank you, Mr. Chairman.
We know AI is going to be an increasing part of our daily
lives, in our homes, in our offices, small businesses, helping
out with medicine and education, entertainment, and we are
going to need advanced microchips, a strong internet
connectivity, growing workforce, and huge amounts of storage in
energy to make this happen.
It was recently announced a $70 million artificial
intelligence partnership between UF and NVIDIA. We appreciate
that.
At NeoCity in our district in Kissimmee, we are making
advanced aerospace microchips and AI capacitor microchips. It
has been named an NSF Engine, and Chris Malachowsky is a UF
grad and cofounder, so we appreciate that partnership.
Mr. Vasishta, we see huge tariffs being levied, especially
today, 50 percent on steel and aluminum. Ten percent across-
the-board tariffs still remain in place, as well as higher
elevated tariffs for places like Canada and Mexico. How does
this affect manufacturing of advanced microchips?
Mr. Vasishta. Firstly, yes, thank you very much for the
acknowledgement of Chris Malachowsky, who is, as you said, a
founder of NVIDIA, and very connected to University of Florida.
You know, my real specialty is telecommunications and AI,
so I am going to have to defer on the question of tariffs, if
you don't mind.
Mr. Soto. So you don't use steel or aluminum to help make
microchips?
Mr. Vasishta. I am sure they are used, but I don't get to
see that on a daily basis.
Mr. Soto. I also want to talk about immigrant labor. You
know that your CEO, Jensen Huang, is an immigrant. We see a lot
of CEOs are immigrants who come into the country.
How important is it for some of these visas and to allow
some of this highly skilled talent to stay, and what effect
could deportations and a chilling effect on immigrants who come
and want to be U.S. citizens and contribute to the economy for
the future of microchip technology?
Mr. Vasishta. Yes. So NVIDIA, we pride ourselves in being a
global company. We have employees around the world, and also we
have some amazing talent that has been able to come to the U.S.
to really exhibit their talent and grow their talent. And I am
actually a recipient of that, being able to come to the U.S.
and grow my career, and now at NVIDIA.
So having the right availability of talent where you need
it and when you need it, and the enablement of that talent to
be trained on the most current technologies and bringing the
brightest and best, and leveraging that across the globe I
think is absolutely essential for us in the U.S. to grow.
Mr. Soto. So if we allow them pathways to stay, that could
help the United States. If we let them go back to other
countries, that actually increases the competition, especially
if we are talking about nations that don't share our values.
We also see the CHIPS Act under attack, $52 billion for
chips manufacturing. You had mentioned also the importance of
telecom manufacturing, $1.5 billion to incentivize that.
How key is maintaining the CHIPS Act to helping have enough
resources, both public and private, to develop advanced AI
microchips?
Mr. Vasishta. Yes. As I said in my opening remarks, I think
the NTIA or public wireless fund that was created is absolutely
correct. There is a requirement for public and private
partnership when it comes to research. We have some of the best
researchers on AI within NVIDIA, and we work constantly with
research organizations. And we need--they also need funding as
well to enable them to do their best work.
Mr. Soto. Thank you so much. It has been 135 days since
President Trump has taken office; 135 days of delays to the
BEAD program, rural broadband, high-speed internet for folks in
underserved areas; 135 days of nothing happening, even though
50 States have already approved their plans.
Mr. Ramzanali, what is the cost to places in rural America,
like my district in south Osceola and east Orange, and to local
agriculture and other small businesses in rural America by this
delay to the rural broadband program?
Mr. Ramzanali. I appreciate the question. The cost is the
delay, not just of people having access to world-leading tools
that we believe should be developed in the U.S., but it is also
the cost of the economic value that all of those people could
be producing through jobs, through remote learning, through so
many other things that the internet enables.
Mr. Soto. Thanks so much. It is time to get this done. And
I yield back.
Mr. Hudson. I thank the gentleman.
I now recognize Representative Allen for 5 minutes to ask
your questions.
Mr. Allen. Thank you, Chairman Hudson, for holding this
important hearing, and I thank the expert witnesses for joining
us today.
Mr. Vasishta, could you provide a working definition of AI
and its impact in context of this hearing today?
Mr. Vasishta. Yes. Obviously, everybody talks about AI, and
we heard some examples of some of the uses of AI in your
opening--in the committee Members' opening remarks. AI is
really the ability for computers to predict, to think, to
perceive as a human would do, and hence that is the definition
of AI.
Mr. Allen. Mr. Pickering and Mr. Vasishta and Mr. Shea,
let's talk about natural disasters. My district experienced a
significant telecommunications disruption from Hurricane
Helene. Took weeks in some areas for phone service to return to
normal.
Could artificial intelligence help mitigate these
disruptions in the future, and, if so, how? Mr. Pickering?
Mr. Pickering. The answer is yes. The AI applications,
whether it is in our fiber, wireless, or any of our networks,
is able to both manage and optimize redundancy, resiliency, and
to be able to get systems back up and running and identify
where issues are much more quickly than in the past.
As I mentioned, being from a State that is both tornado-,
flood-, and hurricane-prone, this is a critical issue.
Mr. Allen. Right.
Mr. Pickering. And AI, just like in every sector, I think
will give benefits of getting our communications back up,
managing our networks more efficiently and effectively. And so
I think it is a great application that we can all cite as one
of the reasons AI is a good thing.
Mr. Allen. Mr. Vasishta?
Mr. Vasishta. Yes, so AI has the capability--and we are
starting to develop some of that capability even more at
NVIDIA--to actually have some level of prediction of weather
outcomes as well.
So the first stage is, obviously, when you start to enable
predictions to happen and be able to proactively react and make
decisions based upon those weather predictions and likely
outcomes. A lot of work is happening with NVIDIA around that
about--with something that we call the Earth-2 model, and we
are working with researchers around the world to make that
happen.
And then, of course, there is the observe, orientate,
decide, and act aspect of AI, which AI is able to then make
those decisions real time autonomously, and then be able to act
on those decisions autonomously and in an agentic way. This is
really the year of AI agents that can make those kind of
actions and decisions autonomously and rapidly, and then, of
course, make decisions to be able to react afterwards to put
the right logistics in place for a complex supply chain.
Mr. Allen. Mr. Pickering?
Mr. Shea. I have to, you know, go with Mr. Vasishta's
comments that with digital twin technology we can predict what
coverage we have left over with the resources that are existing
after a natural disaster.
And the AI in particular is good, as I mentioned in my
talk, about pulling signals out of noise and interference. So
although you may not get the full capacity, you can at least
get some capacity to everyone, so first responders and people
in need will have some coverage no matter where they are in the
cells' capabilities with what you have available.
So, yes, it is a great way to recover from a disaster, know
where you need to put your resources to go, which cell sites to
fix first, how to get the maximum capacity up to serve the
people.
Mr. Allen. Right. Yeah, it was critical to our first
responders and law enforcement and others in dealing with that
disaster.
Mr. Pickering, how can AI be used to enhance efficiency
within our communication networks?
Mr. Pickering. We already know and see in our wireless
networks the optimization of spectrum, how you can more
efficiently dynamically share, use spectrum beam and target
spectrum, mitigate interference.
So the maximization and the efficiency that AI brings to
our wireless networks can also be used in our fiber networks,
it can be used in data centers, and it can be used in low Earth
orbit satellite. So every communication system now is going to
be embedded with AI efficiency.
The same thing is true in our energy networks in grid----
Mr. Allen. All right.
Mr. Pickering [continuing]. And transmission.
Mr. Allen. Good, good.
Mr. Shea, I have got 20 seconds. How can we use AI to
secure our communications infrastructure from malicious actors?
Mr. Shea. Well, with the infrastructure, you can actually
locate fake base stations, denial-of-service attacks, and other
type of capabilities that people, adversaries, might bring
against you. So at the physical layer, it helps you protect it.
Then you have your cybersecurity at the back end that can work
hand-in-hand to protect our networks.
Mr. Allen. Good. Thank you, thank you all. I appreciate it.
And I yield back, sir.
Mr. Hudson. I thank the gentleman.
The Chair now recognizes Representative Clarke for 5
minutes to ask her questions.
Ms. Clarke. Thank you very much, Mr. Chairman. Let me thank
our panel of witnesses for appearing before us today.
Our committee has a long history of working in bipartisan
manner on issues of connectivity and protecting consumers. More
recently, we have had robust bipartisan conversations about AI
regulation, including the opportunities and challenges
associated with this emerging technology, which is why I am
disappointed to see my Republican colleagues turn their back on
our bipartisan work and sell out to Big Tech millionaires who
have bought their way into our Government.
Last month, Democrats sat in this hearing room for over 24
hours relentlessly combatting the bad provisions included in
the Big Bad Ugly Reconciliation Bill. And in the middle of the
night, Republicans voted to approve a 10-year moratorium on
State and local enforcement of their own AI laws and provided
no Federal safeguards in their place.
So let me be clear. I say this is a giveaway to Big Tech at
the expense of Americans' personal freedom, privacy, and safety
online. And until my Republican colleagues finally get their
act together after 3 years in the majority, there will be no
recourse or guardrails in more than half of the States that
have responded quickly to their residents' concerns about the
risks posed by AI.
New York City is one of the early movers in this space.
Since 2023, we have had an effect on AI bias law designed to
regulate the use of AI in employment decisions. This is just
one of the hundreds of State and local laws my Republican
colleagues would sweep away.
Mr. Ramzanali, can you please speak to some of the real-
world harms and unintended consequences of the moratorium,
especially those related to bias and discrimination?
Mr. Ramzanali. I appreciate the question. And the New York
City law is a good example because the employment
discrimination it is going after. Let me tell you the kind of
harm that we have seen with AI systems, in resume-screening
software in particular. There was a firm that was using resume-
screening software for computer science jobs, and it was
screening out women. Now, the company dealt with that, but that
is the kind of information that leads to huge problems.
The kinds of laws that are out there are not just the New
York City employment discrimination law, but you also have laws
that create transparency so that we can know when there is a
problem.
Ms. Clarke. Thank you.
Ensuring that we have proper safeguards in place for
sophisticated AI systems is only one piece of the puzzle. It is
also critically important that consumers understand the
abilities and shortcomings of AI systems that are poised to
become an increasing part of our everyday lives.
AI is already disrupting the way we live and work,
supercharging scammers and refocusing our coal industries. Now
perhaps more than ever before, digital literacy and AI literacy
will decide who can participate in our modern economy.
That is why I am concerned with President Trump's misguided
efforts to roll back the Digital Equity Act. This statute was a
vital investment in making not only internet access available,
but it also educated users on how to use it. We have seen far
too often during this administration, we were once again on the
finish line of getting $2.75 billion of Digital Equity Act
grants out the door, and then the President determined, by way
of Truth Social, that this program was woke, racist, and
unconstitutional, and directed the Department of Commerce to
stop the congressionally authorized and appropriated funding.
Apparently, he was triggered by the word ``equity.''
The truth is, gutting the Digital Equity program will only
hurt vulnerable populations like seniors, veterans, low-income
communities, and communities of color who already suffer from
the digital divide. It will sacrifice critical AI trainings
that would have helped seniors understand online scams, upskill
workers, and help more Americans incorporate AI into their
everyday lives.
Without these programs, we risk building AI bridges to
nowhere, creating a new digital divide in which certain
communities can benefit from these new technologies and others
slip further behind. This not only harms the economic health
and well-being of our local communities but jeopardizes
America's AI competitiveness that my colleagues on the other
side of the aisle say they care about.
So I urge my Republican colleagues to join us in calling on
President Trump to cease his efforts to destroy the Digital
Equity Act programs.
Mr. Chairman, I would like to submit for the record a
letter from the National Digital Inclusion Alliance describing
the importance of the Digital Equity Act in making sure all
consumers can take advantage of AI.
[The information appears at the conclusion of the hearing.]
Ms. Clarke. And with that, I thank you and I yield back.
Mrs. Fedorchak [presiding]. The Chair recognizes--excuse
me.
The Chair recognizes Mr. Latta for his 5 minutes of
questions.
Mr. Latta. Well, thank you very much, Madam Chair. And to
our witnesses, thank you so much for being here.
AI is on all of our minds, as many Members have already
said. So many questions, so little time.
Mr. Pickering, in your statement, you made very, you know,
eye-opening remarks when you said that in 2024 the United
States invested 12 times more into AI than the Communist
Chinese, but that lead is not guaranteed.
And one of the things I know that we have talked about in
the past is about permitting, and we got to get it done. And it
is almost like the top, when we were talking about on the
communications side and we were talking about on the energy
side, and it is all coming together.
But could you talk briefly about if we don't win this race,
what is going to happen, especially if we don't get our
permitting done in this committee and in the House?
Mr. Pickering. So as China builds out their infrastructure,
both energy and fiber networks and data centers, they are not
going to experience the type of permitting delays that the
energy, data center, and fiber industry are experiencing.
I want to commend you for your leadership in the
reconciliation on having a national framework on permitting for
pipelines, that if it is a multi-State pipeline, that there is
a means by which you can have a time-certain approach of a
year, with an extension of 6 months.
Yesterday I met with Congresswoman Fedorchak about what
they do in North Dakota and the accountability, the
transparency, shot clocks, and--you can have an accountable,
transparent process that protects our natural resources and
communities, but to act in a timely way.
And so our industry, on both sides of the data center,
wants to work on a permitting route reform that gets us, as
quickly as possible, to build as fast as possible.
Mr. Latta. Let me ask you real quickly if I can just follow
up because, again, if we are looking at if we don't get this
done--and as you said, that lead is not guaranteed--how much
time do we have left?
Mr. Pickering. It is time to build. It is time to go and--
you know, we have horror stories on fiber networks and railroad
crossings and long processes that take, on the energy side,
sometimes 10, 15 years, on the fiber side 18 to 2 months, when
we need to be able to have shot clocks of 30, 60 days of
getting the permits that we need to build.
Mr. Latta. Well, thank you very much.
Mr. Vasishta, you know, with so many data centers coming
online that we are seeing across the country, can current
telecom networks handle the amount of traffic that current and
also future AI and data centers are going to bring?
Mr. Vasishta. Yes, so you are right, there's a lot of data
centers coming online. A lot of those data centers are actually
connected through fiber to each other or to the
telecommunications network. The traffic that comes over the
telecommunications network, particularly the wireless part of
the network, is continuously growing, and that is where we
really see the need for AI, because that traffic is as the AI
models are trained, and then as the AI models are inferenced
and get consumed by the consumers and by enterprises, that
traffic is going to grow considerably.
And AI is essential, and that is why the fusion, as I said
earlier, of radio access network infrastructure on top of also
the AI infrastructure will really help enable that AI traffic
to be distributed and consumed.
Mr. Latta. Let me follow up with another question to you
because, again, you know, when we were talking about the race
to 5G, and now we are in the race to 6G, and I know I had
asked, you know, different witnesses that were here, where are
we at--are we winning, are we losing--and, you know, I always
hear that we are doing well, and all of a sudden we are not
hearing people saying we are doing that well.
I got a question. What if we don't win that race to 6G--
because also in your testimony you talk about the real threat,
the detection, to automated remediation, and incident response,
especially looking at some of these cyber attacks coming in.
What happens if we don't win that race to 6G?
Mr. Vasishta. Yes, so I think it is an imperative, but let
me answer the question that you have stated. The challenge with
this--the challenge and benefit, the pro and con of the
convergence of AI and 6G standards, is that there is really
only one other country that is thinking about this, and that is
China.
So traditional implementations of radio access network have
not required AI. There is some AI infusing, but the ability to
really take advantage of AI, I think, leaves us at a
significant disadvantage from all different facets, both
productivity but also security and overall growth.
So I think--I hate to answer the question because I hope we
never get to that point.
Mr. Latta. Well, we have a lot of work to do in this
committee and in this Congress. And, again, I appreciate all
the witnesses for being with us today.
And, Madam Chair, I yield back the balance of my time.
Mrs. Fedorchak. The Chair recognizes Representative Peters
for his 5 minutes of questions.
Mr. Peters. Thank you, Madam Chair.
First of all, I want to say where I think there is
agreement here. I really do believe that permit reform is very
important and have been working on that. Would love to see some
bipartisan action around that.
Also, we had a privacy bill here passed with a single
national standard, I think it was 55 to 2. I don't know where
that bill is. We should bring that back and we should pass it,
because I believe that in some things you really have to have a
Federal standard. I would say this is one where we also have to
have one Federal standard.
We should take the best ideas from New York City, take the
best ideas from California, Mississippi, whoever is passing
these bills, and we should put them in one Federal standard
because it is impractical for us to have not just 50 different
States regulating it, but now localities, counties. I mean,
this is nuts. That won't work. I agree.
Let me tell you my problem, Mr. Pickering, since you were
in Congress. I heard all this talk about urgency, but the
Republicans came up with this notion we should have a 10-year
moratorium. What timing does that signal to this Congress is OK
for setting a standard? Doesn't it really basically say you got
10 years?
Mr. Pickering. Two precedents to consider. The internet tax
moratorium was a 3-year moratorium that was extended twice, and
eventually it was made permanent by President Obama, who signed
a permanent internet tax moratorium. Now, the result of that,
as e-commerce emerged, has been at least $5 trillion----
Mr. Peters. Yeah.
Mr. Pickering [continuing]. In economic development because
we didn't have, you know, hundreds of tax jurisdictions on e-
commerce.
Mr. Peters. Sure. Yes, yes, yes.
Mr. Pickering. The second precedent for the committee to
consider was the beginning of spectrum auctions. It was in the
Budget Reconciliation Act of 1993, which was President Clinton
and Gore, and they included it in because the spectrum auctions
would create revenue.
Mr. Peters. Right.
Mr. Pickering. And at the same time, they preempted any
State regulation on the rates and on the entry because that
would have devalued this new emerging technology.
The moratorium and modernization provision in
reconciliation has the same principles and same concepts. We
believe an all-of-government modernization of every Government
service--from Department of Defense, to Energy, to Medicare, to
Medicaid--will have tremendous savings if they adopt AI uses
and applications and technology. But if we have 50 different
States regulating----
Mr. Peters. No. Actually----
Mr. Pickering [continuing]. It undermines the basis of
this----
Mr. Peters. I am not sure it can be done in reconciliation,
and I don't disagree with the theory of it. I just think that
10 years--in the face of this talk about urgency, we had Eric
Schmidt come in here. You could have heard a pin drop when he
talked about how important this was.
Ten years is completely out of line. You know, I think if
you are talking about a mor--that is a ban. That is not a
moratorium. A moratorium is 2 years.
Mr. Pickering. You know, to me, whether it is 10 years or
something less than 10 years, as long as this Congress has a
window of pause to set the Federal framework--you just
mentioned you are close on privacy. We are very close on
permitting on the infrastructure.
Mr. Peters. We actually have a bill passed on privacy which
we can't get back here to actually--
Mr. Pickering. So the question is, what is the right time
to give you the opportunity on this committee to create a
national framework on the major questions?
Mr. Peters. OK. My answer is, this term, which has about a
year and a half left. And that is the appropriate time for a
moratorium. So I am all for doing all--accomplishing all the
goals, but if there is really a sense of urgency, let's get it
done this term, is my answer.
Mr. Pickering. I agree.
Mr. Peters. I had another question for Mr. Ramzanali. Just
this concept of normal technology. In your testimony, you
stated that AI is a normal technology that needs normal
regulation. I wish I understood what you meant by that.
But how can this committee treat AI like normal technology
when it is so complex, and what maybe is the priority, you
think, for us to attack first?
Mr. Ramzanali. I appreciate the question. So this framework
of a normal technology is not meant to say it is not powerful.
It is powerful. We should apply it across so many different
domains. It is going to have really important impacts in a lot
of scientific domains.
The idea is to say this is not the kind of problem where we
don't have policy tools from our history that we can apply. So
that is the idea, is we can treat it in a way where we can look
to historical precedent, we can look to the policy toolkit we
have, and apply that.
Mr. Peters. OK. Well, I am looking forward to this. Mr.
Pickering, I am just playing with you because you are a former
Member of Congress. It is fun to see witnesses out there.
Mr. Pickering. And I want you to know, I agree, the sooner
the better. We have a national----
Mr. Peters. Yes. I think the thing is that I also want to
indicate that this Democratic reason has to be a national
standard. This Democratic reason has to be permit reform, but I
just laugh at the notion that 10 years is the right timing. And
if we are really--if we really have a sense of urgency about
it, this term.
I yield back.
Mrs. Fedorchak. The Chair recognizes Mr. Bilirakis for his
5 minutes of questions.
Mr. Bilirakis. Thank you. I appreciate it very much, Madam
Chair.
Mr. Shea, you briefly mentioned how technology and AI have
been used to combat threats to the Uyghurs by both China and
Turkey. As cochair of the International Religious Freedom
Caucus, I have been a long-time advocate for the Uyghurs, so
the importance of this really stuck with me.
Can you talk more about how AI is being used or can be used
to prevent human rights abuses, war crimes, and other acts of
persecution?
Mr. Shea. I think the thing you can do with proper
monitoring, you can make sure that adversaries aren't getting
into your network, putting up fake base stations, doing other
types of things that they are then using for surveillance.
Because, you know, the problem with the fake base station is
they grab a call, and then suddenly the information going
across there often can be deciphered.
So I think it is important for us to provide the tools,
both to our partners and our country, to be able to protect
these networks to make sure that it is not being used, you
know, for hostile-type intent.
Mr. Bilirakis. Very good. Thank you.
At the same time as this hearing, of course, you know, the
E&C is also having the oversight hearing on robocalls. So it is
fitting to bring this topic up at this particular time, which
is very important to my constituents.
So, again, with regard to robocalls, ever since the TRACE
Act was passed, I think the average American became familiar
with how AI capabilities have addressed spam calls. So we made
some progress. People can now see a likely spam message pop up
when an unknown number calls, helping prevent fraud, and that
is great progress--it really is--toward protecting vulnerable
people from identity theft. But we have to do more.
But actually catching these criminals continues to remain
elusive, and prosecutions are rare, unfortunately.
So, Mr. Pickering, how can AI be used, utilized, to
actually track down online and by-phone criminals that are
preying on our seniors in particular and finally shut them down
once and for all? If you could answer that question, I would
appreciate it.
Mr. Pickering. Yes. The great thing about AI is, if you
give AI a problem, it will create a greater, productive way and
solution to attack any problem, whether it is fraud, criminal
conduct, or find a good cure for cancer. It will be faster, it
will be more productive, and it will be able to bring, you
know, all the knowledge of any particular case to be able to
solve it and to meet the objectives, whether it is in criminal
justice or in national security or in energy or any other
sector.
And so the applications and the growth of our large
language models and what comes next with the quantifiable
language models is really going to be an ever-increasing way,
whether it is robocalls or fraud, to be able to identify,
authenticate, and recognize--identify bad actors and then give
tools to pursue them.
Mr. Bilirakis. Very good. Thank you.
One more question for Mr. Pickering. One of my senior
staffers just got back from a trip, a delegation trip to
Israel, including the Gaza Strip. One of the items discussed
with Israeli leaders was the potential of AI to help identify
and address the significant rise in anti-Semitism online. And
with recent events in Colorado, it is clearly not only
something of importance to Israel but a potential aid to a
worldwide problem.
Mr. Pickering, again, can you explain how AI is currently
being used by your member companies to identify threats of
violence and potentially prevent violence against religious
groups and what potential AI has in the near future to continue
to address this particular issue?
I know you touched on it, but if you could elaborate, I
would appreciate it.
Mr. Pickering. You know, this is a subject I am not as
familiar or have not been privy to those types of applications.
But I am confident if----
Mr. Bilirakis. In general, yes.
Mr. Pickering. In general, if there is an ability to,
whether it is a prediction of a natural disaster or a
cybersecurity or a potential crime, I think AI can be a
resource and a tool for law enforcement, for national security,
or for Homeland, to be able to better predict or identify, and
then hopefully prevent disasters or attacks on the U.S.
So we have to have--again, this is why--I think a Federal
framework in each category of the major questions so that we
could use the full resources of the modernization of Government
services that include national security, cybersecurity, and
homeland security.
Mr. Bilirakis. Thank you. I yield back. Thank you.
Mrs. Fedorchak. The Chair recognizes Mrs. Dingell for her 5
minutes of questions.
Mrs. Dingell. Thank you, Madam Chair.
Artificial intelligence is transforming nearly every aspect
of our lives, which we all know, with great potential benefits
and serious risks. As AI becomes more powerful and more deeply
embedded in our economy, we have got to take comprehensive
action to assure this technology strengthens ourselves, our
health, safety, economy, and national security so we can reap
its benefits. But we also know that it has got serious
potential harms. So, I mean, just some examples is AI-driven
robocalls, deepfakes, and deceptive advertising. And we have to
be mindful of how much AI would widen existing digital divides.
But I want to be positive too. There are real
opportunities. When paired with next-generation connectivity
like 5G and 6G, AI can revolutionize healthcare, improve
customer service, and help power the future of the automotive
industry, boosting innovation, creating jobs, and improving
lives across the country.
But in the few minutes I have, I want to stay focused on
robocalls. I think there is a growing threat of robocalls and
robotexts, many of which are increasingly powered by AI and
disproportionately target vulnerable populations. In 2024
alone, Americans received over 52 billion robocalls, nearly 200
calls per adult. Nearly half were scams or unwanted
telemarketing calls. They are not just annoying, they are
dangerous for a lot of people.
Consumers reported losing a record $12.5 billion to fraud,
with $2.95 billion lost to impostor scams where bad actors used
AI and deepfake tools to convincingly mimic trusted voices and
identities.
Additionally, robotexts and phishing emails are tricking
people into clicking malicious links or sharing sensitive
financial information, scams pretending to be from Medicare,
law enforcement, bank accounts, or even family members in
distress. We used to get the old one, ``I am,'' you know,
``stranded.`' Now they are mimicking these voices of family
members. They are using cloned voices to build trust. They are
getting more and more sophisticated.
And as AI voice cloning, spoofing, and deepfake tools
become more accessible, threats continue to evolve, and
Congress, the Federal Government, and regulators have to keep
pace.
Mr. Ramzanali, can you speak to how bad actors are using
artificial intelligence to exploit seniors, people with
disabilities, other at-risk groups?
Mr. Ramzanali. I appreciate that question. And you are
right, this is a pernicious problem. It is not just annoying.
People are getting scammed in dangerous ways.
The robocall problem is accelerated by AI in a number of
ways. Generative AI is being used to generate scripts that can
be compelling for particular audiences. It is being used to
build impersonation of specific people that a person might
know, and it is being used to read that script on a call to try
to scam somebody out of their money. That is a problem with the
technology.
In my view, regulating a problem like that doesn't hurt
national security. And so this is the kind of regulation where
we can reduce real-world harms and not hurt our leadership in
the world.
Mrs. Dingell. I want to build on that. Earlier this year,
the FCC Chairman emphasized that cracking down on illegal
robocalls remains a top priority. The Commission has committed
to expanding the use of Do Not Originate lists and
strengthening call-blocking tools, both of which are essential
to protecting consumers. Though I would say to you, I am on the
Do Not Originate list, and I get 20 calls a day. So I am trying
to figure out and get that figured out.
But we must ensure all providers are fully implementing
these tools. Last month, the Commission adopted new rules that
required caller ID authentication on non-IP networks, helping
ensure uniform robocall mitigation across platforms.
Mr. Ramzanali, how can artificial intelligence and machine
learning be leveraged not only to detect but proactively block
or trace--trace--illegal robocalls and robotexts? How can AI be
part of the solution?
Mr. Ramzanali. I appreciate that question. The FCC had a
proceeding last year to ask that exact question and has some
answers. But I would say, where you went, where you are on the
Do Not Call list, I am too. We both get calls all the time.
That is not acceptable. We can't be at a place where we accept
that. We wouldn't accept that with bank fraud. We wouldn't
accept that with our spam emails. So I do think it is worth
continuing to push the FCC.
The other hearing that is happening on robocalls, the
Consumer Federation of America has a lot of good ideas on what
else can be done. I urge the committee to consider those.
Mrs. Dingell. Thank you.
I had questions for you, Mr. Pickering, but I am out of
time. So I will yield back and submit them for the record.
Mrs. Fedorchak. The Chair recognizes Mr. Obernolte for his
5 minutes of questions.
Mr. Obernolte. Well, thank you very much, Madam Chair.
Mr. Pickering, we have heard in your opening remarks about
your support for the moratorium on the enforcement of State
regulation of AI. And you have heard in the opening statements
of several Members here and then in some of the subsequent
questions, answering, some assertions about the moratorium. So
I wanted to ask you specifically about a few of them.
One of the assertions that we have heard is that the
moratorium should be stripped out under the Byrd Rule because
it is policy and isn't related to fiscal matters.
Now, as you know, the reason the moratorium was included is
because we are making a $500 million investment in procuring AI
to make Federal Government more efficient and effective, and it
is nonsensical to make that enormous investment if all these
Federal agencies are going to have to navigate this morass of
50 different State regulations.
Would you agree with the assertion that has been made that
this should be stripped out under the Byrd Rule, or would you
disagree?
Mr. Pickering. I would disagree. And, again, going back to
other precedent, 1993, Bill Clinton and Al Gore started
spectrum auctions with a Democrat majority, both the House and
the Senate, on a partisan vote. And it was in budget
reconciliation, so that you would have all the value created by
spectrum auctions. You would also create a competitive
industry, and you would preempt States from regulating wireless
entry and rates, so that you would get the maximum value.
I think that is one of the best parallel precedents to the
AI moratorium so that we can maximize all of the savings and
all of the efficiencies across government, if we were to adopt
AI technologies. We think--you know, just in the Department of
Defense alone, we have an example of one of our companies,
Granite, that does telecommunications services, replacing the
old network and communications with new AI-generated options
and services. And they believe just in one branch that they can
save $100 million a year on changing that type of technology
from obsolete, antiquated to new.
And if you do that across the board in every department,
every agency--you know, last night you were at the AI award
dinner. The vice admiral that heads or is the Director of the
U.S. Geospatial-Intelligence Agency has probably created more
wealth in our country in the private sector with taking the
geospatial and putting it in all of our devices that we now
know as Google Maps, or Apple Maps, or precision farming, or
transportation and distribution.
We think that the AI adoption governmentwide will have
tremendous savings and also help us grow our economy once the
government products and solutions also go into the commercial
markets.
Mr. Obernolte. Well, thanks. It is helpful to point out
that this has precedent. It has been done this way before.
Mr. Pickering. Yes.
Mr. Obernolte. Another assertion that was made by several
of my colleagues in their opening statements is that the
moratorium would prevent States from enforcing laws that
protect consumer safety and prevent deceptive business
practices. Would you agree with that assertion?
Mr. Pickering. I would disagree. The general application of
law, whether it is civil rights, consumer protection, consumer
fraud, criminal conduct, the language that is in the
modernization and moratorium in the reconciliation does not
disrupt any of the enforcement of those laws.
Mr. Obernolte. Right. Yes. There is specific language in
that bill that says that, as long as something isn't narrowly
targeted on AI, it can be enforced.
Mr. Pickering. That is correct.
Mr. Obernolte. All right. Last assertion that has been made
is that this is a giveaway to Big Tech. Is that something you
would agree with or disagree with?
Mr. Pickering. As an advocate for competitors for 40
years--for the new entrants, the new technology, the upstarts,
the innovators, the entrepreneurs--it is the exact opposite.
If you are a small startup, you cannot afford the patchwork
of 50 States, the complexity of it, to create a model that you
hope to deploy in 50 States and nationally and globally if you
have to go through the hoops of every different State's
regulation. It is the small startup that suffers the most under
that system versus having one predictable national framework
that then helps the new start and the competitor enter without
a regulatory burden and cost that the big companies can afford
and manage and have the resources in every State and here in
Washington to manage. It is the small and the new entrant that
does not.
Mr. Obernolte. Right. So this favors entrepreneurs and
actually is anti-Big Tech because it encourages competition.
Mr. Pickering. And, you know, one of our----
Mr. Obernolte. I am sorry. I don't want to let you go over
here. I am----
Mr. Pickering. Sure, sure. Well, OK, thank you.
Mr. Obernolte. I see we are out of time, but I thank you
very much for your time.
Mr. Pickering. Thank you very much.
Mr. Obernolte. I yield back.
Mrs. Fedorchak. The Chair recognizes Ms. Barragan for her 5
minutes of questions.
Ms. Barragan. Thank you, Madam Chair.
I would like to focus on AI's role in the real-world, life-
or-death situations. Nearly 68 million Americans speak a
language other than English at home. That is 1 in 5 Americans.
In addition to immigrant communities, millions of tourists
contribute to our economy, and their safety matters too. For
all these people, receiving emergency alerts or calling 911 in
their own language can mean the difference between life and
death.
AI can help bridge those gaps, translating weather alerts
or 911 calls in real time. But if we are not careful, errors
and delays in translation could cost lives. States need the
power to enforce rules, and Congress must ensure these systems
are safe, effective, and fair.
Mr. Ramzanali, in your testimony, you highlighted AI's
potential for real-time language translation during 911 calls.
How can AI and Next Generation 911 work together to better
serve our diverse communities and save lives?
Mr. Ramzanali. Thank you for highlighting the community
that has that need, and let me describe the problem. When
someone calls into 911--someone who doesn't speak English--
there are translation services available. It just takes the
operator time to even know what language is being spoken.
That time is critical life-and-death time. That is where
technology can be helpful. Now, this is not at the, like, let's
deploy it out to a hundred percent of the 911 operators today,
but it has promise. I will also point out that the transparency
requirements that a lot of States have, that is the kind of
transparency requirements you want going hand-in-hand with
deploying a technology in this kind of a situation.
Ms. Barragan. Well, thank you. That is one of the reasons I
am disappointed that my Republican colleagues have abandoned a
plan to fund NextGen 911, money that could have ensured that
everyone, regardless of language they speak, could get help in
an emergency. Instead, they gave billionaires yet another tax
break, putting lives at risk.
Mr. Ramzanali, could you describe how AI translation
technology has already been used to deliver life-saving
information to millions of Americans who speak a language other
than English?
Mr. Ramzanali. As I described, it is starting to be used in
911 systems today. Some of the vendors are rolling it out. It
is not yet at mass scale, but it is something where we are
seeing that happen and--in fact, I heard from one of the
government officials who works on NG911 this week that it can
save up to 6 minutes in the delay between having access to 911
response in the language that somebody needs and not having
that for 6 minutes.
Ms. Barragan. Wow, that is a lot of time when it comes to
an emergency and response.
I would like to share another example. In 2023, the
National Weather Service partnered with an AI translation firm
to train a language model in weather terminology. Thanks to
that collaboration, forecasters reduced the time to translate
hurricane forecasts from an hour to less than 10 minutes,
potentially saving countless lives.
When the Trump administration let that contract collapse in
April of this year, they created a dangerous gap in information
for millions of Americans. Even when helpful AI systems are in
place, Republicans find ways to undermine them.
Despite impressive advances, I am concerned about letting
the fast-growing AI industry go unchecked, especially in
critical areas like emergency communications. If House
Republicans' reconciliation bill becomes law, States will lose
the ability to enforce new AI regulations for 10 years.
Mr. Ramzanali, what are the risks of underregulating AI
technology in emergency communications, and how might that
endanger lives?
Mr. Ramzanali. I think you made the point well of when
these systems aren't tested well, when they are deployed too
quickly, that can endanger lives.
I also want to go back to your point on the weather data.
The National Weather Service is part of NOAA, which is in the
Commerce Department. Part of what these AI systems that can do
weather--part of what they are trained on is data that NOAA
produces. Think about buoys in the ocean that know when a
tsunami is coming, because we too are a Pacific country, that
data is managed by employees of NOAA. So when the science
agencies hit a cut, that is the kind of work that I get worried
about.
Ms. Barragan. Great. Thank you.
With that, I yield back.
Mrs. Fedorchak. The Chair recognizes Mr. Carter for 5
minutes of questions.
Mr. Carter of Georgia. Thank you, Madam Chair, and thank
each of you for being here.
You know, AI is fascinating, but it is also daunting,
particularly for those of us who don't consider ourselves
experts in this area. But just about every profession--I am a
pharmacist, and it is going to impact my profession. It is
going to impact almost everything that we can think of. But it
is daunting, and it is--we know that the advantages are
immeasurable, but then again, we have got to be careful.
We know that it is going to help us in detecting cyber
threats, increasing network reliability and spectrum
efficiency, or combating robocalls--we all want that. So the
benefits speak for themselves, but the innovation has got to be
balanced. It is got to be balanced with responsibility. And I
am very, very concerned--I think we all are--about that.
I am particularly curious about how AI can help us modify
and improve and deploy broadband, especially in rural areas. I
have the honor and privilege of representing the entire coast
of Georgia, but I have a lot of rural south Georgia as well. In
Georgia we say there's two Georgias: There is Atlanta and
everywhere else. Well, I represent everywhere else, and there
is a lot of rural area in south Georgia.
Mr. Pickering, how can Federal policy promote equitable
access to AI-powered tools and services, particularly in rural
and underserved areas?
Mr. Pickering. Well, one thing that this committee has
spoken earlier this year to address is how the BEAD funds need
to go out as quickly as possible, the speed to BEAD, and to be
able to cut the red tape so that the deployment of broadband
networks to rural parts of the country, like your district, can
proceed as quickly as possible with as little regulatory red
tape as possible.
AI can help us on the permitting side, both for the local
mayor, city council, to be able to have the resources through
AI to solve permitting issues, to make it faster so that you
can build faster.
As many of the panelists have talked about, wireless
coverage in rural areas can be improved by AI and the spectrum
management and how our networks operate and interoperate. So
for rural America, AI on our networks, on our broadband
deployments--I will give you one last example.
We have a company that is building in the Midwest, and they
are able, through an AI software application, to know when a
contractor is completing a fiber deployment by the square foot.
And they can pay them in real time. And so the incentive is to
build faster, because as soon as they build, they can get paid.
And so that is just one example.
Mr. Carter of Georgia. That is a nice incentive. For
someone who was in business for 32 years, I can assure you that
is a nice incentive.
Another component of AI that I think is critical for our
race is from 5G to 6G and the technology there. I believe that
we are in another arms race with our adversaries, such as
China, and we need to be first in the world to reach 6G.
The Salt Typhoon--all of you remember that--last year, it
was a Chinese-sponsored infiltration of nine major American
telecommunications companies' networks and systems.
Mr. Shea, what lessons did the industry learn from the Salt
Typhoon cyber attack? Did we learn anything? I mean----
Mr. Shea. I am not really qualified on cybersecurity to
comment. I just know that, you know, it was a substantial
penetration, and there certainly is a lot of concern in the
industry.
Mr. Carter of Georgia. Anyone on the panel want to take a
stab at that?
Good. I win. I baffled the--I did it. I have been wanting
to do that for 11 years, so thank you all.
What do you think is the role, Mr. Shea, of AI in improving
spectral efficiency and resilience in national critical
infrastructure?
Mr. Shea. Well, we think--you know, as I mentioned in my
testimony, we are trying to share bands with incumbent users
that are very mobile. And prior attempts were very slow. They
took maybe minutes to make a spectrum change.
I think with AI, you can, in a matter of milliseconds,
understand what is happening in the spectrum and have the
people that are using the band that are not the incumbent user
make accommodation for the user. So I think it is opening a
whole new world for spectrum sharing.
Mr. Carter of Georgia. Great. Again, I want to thank all of
you for being here, and I want to remind you, for many of us--
for most of us, I would go as far as to say--this is
fascinating, but it is also daunting, and it is also, quite
honestly, scary. So bear with us.
Thank you, and I yield back.
Mrs. Fedorchak. The Chair recognizes Mr. Menendez for 5
minutes of questions.
Mr. Menendez. Thank you.
Madam Chair, I am deeply concerned that we are sitting here
while our colleagues across the aisle push for deregulation of
AI at the same time that the Trump administration is
weaponizing AI to make Americans less safe.
Let me explain. People across the country are and should be
concerned by recent reports that the Trump administration is
using Palantir's AI technology to consolidate Americans' most
sensitive data from across Federal agencies into one
centralized database. This data could include Americans'
medical history, disability status, bank account numbers,
immigration status, and even real-time geolocation information.
Make no mistake, the Trump administration is taking this
unprecedented step to create a surveillance apparatus and is
doing so with zero oversight from congressional Republicans.
While I am in favor of making government more efficient, we
have known for years that synthesizing this much sensitive
information into one centralized database is ripe for abuse and
makes us all vulnerable to cyber attacks.
Mr. Chairman, I ask unanimous consent to enter this article
from The New York Times about the Trump administration using AI
to merge government data into the record.
Mrs. Fedorchak. Without objection.
[The information appears at the conclusion of the hearing.]
Mr. Menendez. Mr. Ramzanali, you highlight in your
testimony that AI has been used in troubling ways, specifically
as it relates to surveillance. In a few words, how could the
Trump administration's use--how could the Trump administration
use Palantir's AI technology in a centralized database of
Americans' most sensitive information to monitor and track our
constituents?
Mr. Ramzanali. Citizens give a lot of government to their
data, and they have to. That is part of the social contract we
have. For the services you need to be able to do that.
In the 1970s, when the Privacy Act was passed, the fear of
a combined database that is going on right now, that was the
exact fear for the reason that the 1974 Privacy Act was
created, because they didn't think--at that time, Congress
thought that we should be afraid of a government that knows
everything about its citizens.
Mr. Menendez. And think about how much more, right, data we
are providing to the Federal Government to make government work
for our constituents, right? And they should have confidence
that in doing so, that information that they provide, some of
the most sensitive information, will not be weaponized or used
against them. Is that correct?
Mr. Ramzanali. That is right. When--we want people to have
trust that when they are providing information it should be
truthful. We want people to give the IRS truthful information
without fear that it will be abused somewhere else.
Mr. Menendez. Correct. And as AI tools become more
advanced, is there a need for high-end consumer protections for
the Government's collection of sensitive data?
Mr. Ramzanali. Absolutely, yes.
Mr. Menendez. And can you expand on that?
Mr. Ramzanali. There is a lot of--since the 1970s, not only
has the technology changed, our thinking on privacy has also
changed. There are new principles, like data minimization
principles, that the Privacy Act has some of those, but we can
go much further and not just collection, but we should minimize
what we retain and what we process and how we link those
databases.
Mr. Menendez. Great. The last piece of major Federal
privacy legislation was passed in 1998. That is almost 30 years
ago, before many of the major social media platforms were even
launched. Even so, my GOP colleagues continue to sit here and
repeatedly call for Congress to avoid any guardrails on the use
of AI at both the Federal and State level. In fact, their
reconciliation package that passed the House just 2 weeks ago
include a decadelong moratorium on any State law that addresses
AI.
Just yes or no, are commonsense AI guardrails critical for
protecting Americans' data from being weaponized?
Mr. Ramzanali. Yes.
Mr. Menendez. And would the Republicans' moratorium of
State AI legislation wipe away the current guardrails that
protect Americans from their data being misused and weaponized?
Mr. Ramzanali. Yes.
Mr. Menendez. So while the Trump administration is using AI
to collect and exploit Americans' sensitive data, House
Republicans are rolling back AI regulations and preventing
States from filling in the existing regulation gap with their
own policies to--with their own policies to protect our
constituents. This will leave the American people without any
protections as the Trump administration uses AI to act
recklessly with their data, and we will miss a short window to
pass meaningful legislation at both the State and Federal
level.
The bottom line is that the Trump administration cannot
continue to misuse sensitive data with impunity. Congress must
act to implement commonsense guardrails on the Government's use
of AI technology.
Quickly, with respect to the consolidation of Americans'
information across Federal agencies, it is not something that
we have seen before, it is unprecedented, and does it make us
more or less cyber secure as a country?
Mr. Ramzanali. Less. It makes us way more vulnerable.
Mr. Menendez. And if you were China or Russia or North
Korea, right, and you knew that the Federal Government was
consolidating our sensitive data into one database, would that
not be your prime target to attack and hack?
Mr. Ramzanali. I would be shocked if it is not already
their prime target.
Mr. Menendez. I agree. Thank you so much.
Mrs. Fedorchak. The Chair recognizes Mr. Dunn for 5 minutes
of questions.
Mr. Dunn. Thank you, Madam Chair.
This committee has vast jurisdiction over technologies, AI
being no exception. And along with this committee, I also serve
on Speaker Johnson's AI Task Force led by my friend Mr.
Obernolte from California.
We focused on AI regulations, regulations on AI labeling,
watermarking standards, harmful risks such as deepfakes, fraud
prevention, cybersecurity protocols, et cetera. We need America
to be at the forefront of the technology, not China.
President Trump also made important decisions around this,
including his Executive orders to remove barriers to American
leadership in AI and advance AI education for American use.
Congressional action is necessary to support these efforts and
protect Americans from serious risks but also to support
innovation.
Startups and small businesses are already benefiting from
AI. And specifically with telecom, prioritizing advancement of
the American global competitiveness in 5G and 6G growth is
really on the top of everybody's mind.
So I would like to take everything into account on that. A
current example is finding and acquiring the resources to build
new data centers.
Representative Pickering, I want to pivot for a moment to
the realities my district faces, which, due to natural
disasters that affect the Florida panhandle, we rely heavily on
emergency preparedness and response. And as technology
advances, emergency alerts and updates to telecommunications
networks are a welcome advancement, and I support the FirstNet
emergency telecom network in my district. FirstNet is up for
reauthorization in 2027, and these emergency networks are
crucial for us.
So given INCOMPAS' proven success in a wide range of
communication networks and backup infrastructure, how do you
see AI being used for public safety, and how do we protect
emergency networks and systems from, you know, people in China?
Mr. Pickering. It is a great question. And this is where
ORAN and AI can really help on the public safety front and in
predicting a natural disaster and the response to it. One of
the great problems in any disaster is the communication of
local public safety with Federal public safety, whether it is
FEMA or Homeland Security. And by using AI to create better
interoperability of the networks and the communications, the
response and the recovery and the prediction of any natural
disaster, I think, through AI will be improved. And I would
welcome any other comment from the panel, that this is really
where we can strengthen FirstNet with AI applications and the
networks that are coming through the AI movement.
Mr. Dunn. Thank you for that.
Mr. Shea, you spoke about U.S. leadership in AI, and
DeepSig's tech, like your AI-native wireless capabilities,
coupled with Open RAN networks will reduce costs to replace
custom hardware and allow us to better compete with China and
whatnot, specifically Huawei. Can you elaborate on your
company's vision, how you see that entrepreneurship leading to
the conversations, but also the actions of the companies in
creating new technologies?
Mr. Shea. As I mentioned before, ORAN is just enabling this
by disaggregating the network into, you know, components that
can be built with individual commodity-to-base servers. So what
it is letting companies like DeepSig do is participate with
other companies to come up with very cost-competitive
technology. You know, when you buy a server, you have many,
many options. And so by getting away from custom hardware, you
can go commodity on the actual hardware components of it.
And now, the actual radio units are what is left. And
through the NTIA program, there has been a great effort to
reduce the cost and simplify these radio units so they can also
be part of the cost reduction.
So we think this is going to really help drive competition,
which will ultimately drive down costs.
Mr. Dunn. Thank you very much.
In the few seconds left to us, Mr. Vasishta, can we utilize
spectrum bands for wireless advancement? I mean, I think that
is important. I know we are not in a classified setting, but to
the extent that you can, will you share your thoughts on how
our military can diversify spectrum usage to remain, you know,
innovative and stay hidden?
Mr. Vasishta. Yes, this is where, actually, the ability for
AI to do dynamic spectrum allocation, dynamic spectrum sharing
with 4G and 5G, and sensing of the spectrum can be really
beneficial, because many of those applications can be very
beneficial for military applications as well as, you mentioned
earlier, public safety.
So bringing AI and the spectrum together as well as, of
course, additional spectrum that can be made available for the
6G domain, I think is one of the real benefits of AI.
Mr. Dunn. Well, thank you for mentioning dynamic sharing.
That is what I wanted to get from you. I appreciate that.
I yield back.
Mrs. Fedorchak. The Chair recognizes Ms. McClellan for 5
minutes.
Ms. McClellan. Thank you, Madam Chair.
This is probably one of the most important hearings that we
are going to have all year. And I want to put this in a little
bit of context about why I am so excited about it and to hear
from you.
So I graduated law school in 1997, and rather than
practicing toxic tort litigation, like I thought I was going
to, I began working for an incumbent local exchange company
implementing the Telecom Act and spent 25 years in that
industry. And I saw the transition from the Princess phone
plugged into the wall, to this, to telephones in people's
sunglasses, and how rapidly that transition changed as we got
farther and farther away from 1997.
And at my last legal conference in that job, I heard some
statistics--this was in 2018. I heard some statistics about AI
and the ability of AI to create fake news being on pace to
outpace its ability to detect it.
And we talked about--then, we were in the fourth Industrial
Revolution, and we talked about how with each Industrial
Revolution, just as it brought extraordinary advancements, they
brought extraordinary challenges. I don't know if we are on the
way to the fifth Industrial Revolution with the transition from
5G to 6G and how fast AI is evolving, but I think we are
woefully behind getting ahead of the challenges.
Yesterday I was speaking to a group, and they asked me,
specifically about AI, What is Congress not talking about that
it should? And, What are industry leaders not talking about
that they should? And lo and behold, there is an article in
Axios on May 28th called ``AI jobs danger: Sleepwalking into a
white-collar bloodbath,'' where Dario Amodei said that AI could
wipe out half of all entry-level white-collar jobs and spike
unemployment to 10 to 20 percent in the next 1 to 5 years.
Madam Chair, I would ask unanimous consent to introduce
this article into the record so I can get straight to my
questions about it.
Mrs. Fedorchak. Without objection.
[The information appears at the conclusion of the hearing.]
Ms. McClellan. I am actually shocked we haven't heard any
questions yet about workforce. But for all of the witnesses,
how can industry and Congress work together to manage this
workforce transition that is coming?
It came in every Industrial Revolution. But the difference
is here it is going to happen so fast that I don't know if we
are going to have time to retrain workers, I don't know if
there will be jobs to retrain them to, and I don't know if we
can adjust school curriculum fast enough to teach today's
students what they need when they graduate to succeed in this
new world.
So what should we be thinking about to address that issue?
Mr. Pickering. One recommendation that I would have is in
the BEAD Program, there is both deployment and nondeployment.
Louisiana, which--Republican State, on the forefront, they are
using about 60 percent of their funding to deploy and connect
every Louisianian and about 40 percent in workforce training.
So if you think about AI--basically, what our AI networks
that will take the AI applications and content over the
broadband networks that we are building and the infrastructure
that we are building, but you have to complement it with
workforce.
So Mignon Clyburn and I work together with INCOMPAS and the
AI competition center, and one of our key pillars is AI
workforce. And everybody at this table--NVIDIA is extremely
engaged in AI workforce training. Microsoft, Google, Amazon.
The companies that are building the AI models realize that if
we don't start from K through 12, community college and
university, and adult retraining and workforce, this huge
transformation and the benefits could be jeopardized for the
good that it could do. But we need to train now and use some of
the BEAD money in flexibility for both deployment and training.
Mr. Vasishta. Maybe I will just quickly follow on from
that.
AI gives tremendous opportunity for distribution training,
the ability to provide individual students one-on-one tutoring
rather than sitting in a classroom and listening to one
teacher, trained at the same pace, for instance.
But that also requires communications network as well. And
the ability to provide that connectivity to every person where
they need it and every student when they need it, I think,
enables us to train quickly. As mentioned before by Mr.
Pickering, NVIDIA is very much engaged with many IS leads,
meaning independent software vendors, developing those types of
applications such that they can be delivered over the network.
And then, of course, as I said earlier, the ability to
provide a very competitive U.S. telecommunications
infrastructure provider will also bring many jobs into the U.S.
Ms. McClellan. Thank you, Madam Chair. I hope the other two
will send me their responses for the record. And I yield back.
Mrs. Fedorchak. The Chair recognizes Mr. Joyce for 5
minutes.
Mr. Joyce. Thank you, Chair and Ranking Member, for holding
today's hearing. Thank you to the witnesses who have agreed to
come here and testify.
My district in Pennsylvania, and throughout the
Commonwealth of Pennsylvania, is on its way to becoming one of
the epicenters of data centers throughout America. The energy
resources that sit beneath the feet of my constituents is what
makes these centers viable. While AI has the potential to
continue to be developed, without the significant energy
capabilities and the technology will ultimately become
ineffective. This demand for massive energy capabilities
fundamentally goes hand-in-hand with the need for fiber
deployment.
Fiber provides the backhaul that is needed for data
centers. But without the right permitting laws in place, we
will be incapable of speedy deployment. If we continue to
inhibit the growth and the deployment of fiber with obstacle
after obstacle, then we are at risk of losing the AI race to
China. You have all stated that and acknowledged that here this
morning.
This is particularly true when it comes to deploying fiber
under, across, and around railroad tracks. The continual
delays, unreasonable permitting, the back-and-forth between
providers and rail are causing massive problems for my
constituents whose homes reside near these tracks and rely on
providers to deploy to their homes.
Mr. Vasishta, how are you developing and innovating your
technology to address the growing demands of AI? Do you concur
that fiber is the necessary backbone of this system? And what
type of technology do you see becoming critical in the future
to network security as the demand for AI continues to develop?
Mr. Vasishta. So, yes. We do agree that the backhauling of
data centers is very important, and fiber plays a critical role
in enabling that to happen.
If you think about the compute density of--the density of
compute required for AI, it can be at the chip level, it can be
at the data center level, it can be at the campus level, or
multiple data centers. And the connectivity of all those points
within a data center or--data center relies often upon fiber
networks.
One of the things that we are continuously doing is
innovating in that ability to provide low power but also
optical connections and fiber connections to enable that to
happen.
Mr. Joyce. Mr. Pickering, INCOMPAS has been a great
supporter of my legislation that is focused on streamlining the
railroad permitting process. The intersection of fiber
deployment and railroad tracks is just one of the many
obstacles that providers, such as your members, face.
I am the grandson of Pennsylvania railroad workers. I know
how important the rail was for connecting East to West
throughout America. I know that Polish and Irish and German
immigrants built that connectivities. And I feel the rail
industry is no longer responsive to understanding the
connections to occur via rail, but they need to connect with
the ability for fiber to go between, under, and around those
rail crossings.
How are these permitting relays delaying and giving that
advantage to China when rail is not cooperative in United
States?
Mr. Pickering. Thank you. And thank you for your leadership
on this issue.
Whether it is our companies who are building long-route
fiber routes or fiber to the home and fiber to the community,
when they get to the railroad track, there is a regulatory gap.
It is one of the few places that, from a permitting right-of-
way perspective, there is no regulatory oversight.
So the FCC does not have oversight of rail crossings, and
the Federal Rail Administration has no regulatory oversight.
And as a result, there is no incentive for the rail industry to
work with another network industry, the broadband fiber
industry, to give them fair access, timely access, and an
actual cost to cross their railroad.
And as a result, we have story after story of up to 18
months of delays of being able to build. And if you are a fiber
company with capital that is just waiting and losing your money
and your time to build to the other side of the tracks to close
the digital divide, then you are not going to build there. Or
you are going to avoid or--just cost all the community the lost
time and opportunity for a broadband connection.
The other thing: Sometimes we hear exorbitant fees of up to
$40,000 of crossing, making deployments uneconomic to be able
to build and deploy. So this is a major problem. And your
legislation is a very important progress that I hope that we
can make in this Congress and in permitting reform ahead.
Mr. Joyce. The connectivity that rail allowed America to
achieve a century ago needs to continue today with the
deployment of fiber. And the ability of rail to cooperate with
this development, unfortunately, now needs to be legislated. I
look forward to doing that.
I thank all of the witnesses for being present today. And I
yield back the balance of my time. Thank you.
Mrs. Fedorchak. The Chair recognizes Ms. Castor for 5
minutes.
Ms. Castor. Well, thank you, Madam Chair, and thank you to
the witnesses for being here. I am excited about the potential
for AI, all of the advancements and efficiencies in
communications networks and in our lives. And thank you for
pressing this committee on our responsibility to adopt a
national framework.
The problem is that my GOP colleagues have abdicated that
responsibility. And what they did in sneaking in this 10-year
AI immunity moratorium takes us off track, takes us away from
the debate we should be having on that actual framework, and
instead, we are going to have to fight over this big gift to
Big Tech contained in the billionaire tax giveaway. It is
really unfortunate.
And let me say to all of the State and local leaders out
there, I really do appreciate everything that you have done
while Congress has been absent, while Congress has abdicated
its responsibility in this area. And I want to make sure that
we know this is a bipartisan issue across the country. The
National Association of Attorneys General, 40 of them have
written to us that the impact of such a broad moratorium would
be sweeping and wholly destructive of reasonable State efforts
to prevent known harms associated with AI. They had previously
recommended that the Congress act on a framework, especially
addressing high-risk areas. But rather than follow the
recommendation, instead this immunity for Big Tech and AI,
again, takes us off track.
So let me try to get us on track on some of these
provisions. Yes or no for all of you. If Congress were to act
in a framework, would you agree that we should address AI
having highly sexualized conversation with minors, even
encouraging minors to harm themselves? Is that an area that we
should address? Yes or no.
Mr. Pickering. Yes. Child safety, yes.
Ms. Castor. Yes.
Mr. Pickering. At the Federal level.
Mr. Vasishta. Sorry. What was the question again?
Ms. Castor. Sure. When Congress is going to adopt a
framework, is this a topic that we should address?
Mr. Vasishta. I think that is important topic. Yes.
Mr. Shea. I would agree, for minors particularly. Yes.
Mr. Ramzanali. Definitely.
Ms. Castor. Yes.
How about--you know, a number of cities have banned AI-
driven rent-setting software used by large landlords after
evidence that they were using algorithms colluding to push
rents up and reduce housing availability. Is this a topic for a
national framework? Yes or no?
Mr. Pickering. And let me just----
Ms. Castor. Can you do yes or no? I have limited time. Or
you can pass.
Mr. Pickering. Well, existing civil rights laws, I think,
cover that.
Mr. Vasishta. I think I have to pass if that is an option.
Mr. Shea. Likewise. I am not much into regulation. Thank
you.
Mr. Ramzanali. Yes.
Ms. Castor. OK. How about some of the--let's see. There was
another good example here. How about just plain transparency so
that a consumer understands when they are--that AI is on their
phone or guiding their decisions? Yes or no.
Mr. Pickering. As long as it is in a Federal framework,
yes. Fifty different State transparency requirements, no.
Mr. Vasishta. I think transparency is always important.
Mr. Shea. I agree with that. People should understand what
they are being--working with.
Mr. Ramzanali. Yes.
Ms. Castor. So I wonder also--at the end of the last
Congress in December, they issued a Bipartisan House Task Force
Report on Artificial Intelligence. Have you all read this?
Yes, Mr. Pickering?
Mr. Vasishta. No.
Ms. Castor. No.
Mr. Shea. I haven't either. Sorry.
Ms. Castor. You haven't.
It is just very interesting. I am going to ask Madam Chair
that we put in the record the key findings relating to
preemption. That bipartisan work group that some of the members
of the committee here sat on, they said Federal preemption of
State law on AI issues is complex. It has--Federal preemption
has benefits and drawbacks. It can allow State action subject
to floor or ceiling.
But the ultimate recommendation is that the Congress
continue to study this. Nowhere in here does it say that we
should sneak in a 10-year immunity moratorium for all AI
regulation.
Mr. Ramzanali, is the Congress being consistent here?
Mr. Ramzanali. I think you had it right. Senator Blackburn
recently had a very great, reasonable view on this, which was
Congress shouldn't have a moratorium but it should consider
preemption when reasonable protections are being put in place
in a similar area. And that applies because her State of
Tennessee has the Eldest Act.
Ms. Castor. Right.
Mr. Ramzanali. She has the Federal bill, but that makes
sense to preempt when--that debate should happen when it is
time.
Ms. Castor. Thank you very much. I yield back.
Mrs. Fedorchak. The Chair recognizes Mrs. Houchin for 5
minutes.
Mrs. Houchin. Thank you, Madam Chair. Thanks to the
witnesses for your testimony today.
Artificial intelligence is no longer a far-off concept. It
is already changing how we live, work, and communicate, but its
success still depends on the basics: fast, reliable broadband
infrastructure, access to spectrum, cybersecurity, and reliable
energy.
For rural communities, like in southern Indiana, like mine,
and across the country, those fundamentals are not always
guaranteed. That is why I am fighting to close the digital
divide and make sure that those small communities aren't left
behind. Today I want to highlight just a few key priorities.
First, we need commonsense permitting reform to cut red
tape and get broadband and wireless projects off the ground
faster.
Second, we should be harnessing the power of AI to speed up
deployment and strengthen our cybersecurity. It is especially
important for small providers who don't have the massive teams
or unlimited resources to keep up with compliance and ongoing
threats.
We should also be prioritizing the deployment of clean,
reliable energy like SMRs, small modular nuclear reactors, to
power AI and data centers. I hope we can focus on how
Washington can be a partner, not a barrier, in driving
innovation and expanding the reach of responsible AI across
sectors in communities.
Mr. Pickering, you have emphasized that universal broadband
access is a prerequisite to universal AI access. What specific
barriers are your members running into most often, and where
can Congress help?
Mr. Pickering. Well, you had mentioned earlier in your
comments, commonsense permitting reform would be at the top of
our list.
Creating AI connectors in corridors--for example, you may
be building a long-fiber route from an urban area to a rural
area like in Indiana, and it may not be economic to have the
long-route fiber that connects data center hubs. So making that
eligible for BEAD as well as permitting reform--because once
you get the data center hubs, what will happen next is a second
wave, which will be advanced manufacturing, that I believe will
be coming to rural and remote parts of the country that would
never be considered in the old world with old technology as a
manufacturing hub.
But in AI and with advanced manufacturing, as long as you
have energy and if you have data centers and if you have fiber,
then you will see the economic growth come to the middle parts
of the country, not just the coastal.
And so I would encourage permitting reform and BEAD
flexibility.
Mrs. Houchin. Yes. I have been a chief proponent for BEAD
flexibility, removing some of the more costly aspects of BEAD
that are not resulting in deployment of those funds to
broadband, as well as technology neutrality.
Is there a role for AI itself in helping to accelerate
permitting? Could Federal support for AI-based project review
tools actually make the deployment process faster and more
predictable?
Mr. Pickering. The great thing about AI, it is going to be
the greatest technological advancement in human productivity in
the history of the world. And when you apply it to--whether it
is rushing new cancer treatments through FDA approval, the
iterative ability to speed cancer treatment can also be applied
to everyday practical things like, how do we permit faster, how
do we plan better? And the AI applications in every part of
American life can make us more productive and faster in every
area.
Mrs. Houchin. And what is your take on proposals that
classify AI-supporting infrastructure like data centers or
high-capacity fiber? What do you--if we work to classify those
as strategic infrastructure eligible for a fast track in
permitting, is that necessary?
Mr. Pickering. Yes. We are in a critical race against
China. And whatever we can do, especially on the multi-State
energy deployments and fiber deployments, we need some type of
national framework that consolidates reviews, accelerates
reviews, and approves in a time-certain manner.
Mrs. Houchin. And to all the witnesses, if Congress could
just pass one reform this year, whether it is related to
permitting, interagency coordination, targeted incentives, what
would have the greatest impact on unlocking AI's potential
across the U.S. economy?
Mr. Pickering, I will start with you.
Mr. Pickering. Just remember, infrastructure, whether it
was the internet infrastructure of the previous age, once we
built fiber long-haul satellites that were digital, cable that
was digital, and then we had all of the infrastructure in
place, on the wireless and wired side, you could then do
something like this.
The same thing is going to be true in AI applications. If
you build the infrastructure that is now both energy and fiber
and other broadband networks of all technologies, add the data
center, it will unleash unlimited new research, new
manufacturing, precision agriculture, and all of the different
uses that we think will grow the economy.
Mrs. Houchin. Since my time has expired, agree? Disagree?
Mr. Vasishta. Agree.
Mr. Shea. Agree also.
Mr. Ramzanali. I would prioritize other things.
Mrs. Houchin. OK. Thank you. I yield back.
Mrs. Fedorchak. The Chair recognizes Mr. Landsman for 5
minutes.
Mr. Landsman. Thank you, Madam Chair.
I want to talk about the satellite versus fiber broadband,
you know, which seems to be at the heart of this conversation
in terms of where we go long-term and whether or not we have
the capacity. I want to pick up where you just left off, which
is your different priorities. I would love to hear those as it
relates to what we should be doing.
But the question I have, fair or not, is whether or not
broadband and fiber, which we know matters more--it is more
reliable, it is what is going to power all of this--versus
satellite--that satellite has the advantage because of folks
like Elon Musk, right? Because he has got Starlink, and this is
what he is pushing. He has obviously got enormous influence.
Maybe that has changed over the last 48 hours. I don't know.
Do you worry about this--that satellite will somehow, you
know, get the best of us? You all? In terms of resources? Do
you see that? Is this sort of a--just a Starlink thing? Where
is the power coming from on the broadband fiber side?
Mr. Pickering, I would--and that is not a setup. I am just
curious.
Mr. Pickering. So INCOMPAS has members that are fiber,
wireless, fixed wireless, and LEO satellite, Amazon's Kuiper
that just launched and is competing for broadband grants across
the country and competing against Starlink. We think
competition in LEOs and fiber and wireless and everything will
make it better.
And we think that States having the flexibility to choose
whatever technology is best for them--Colorado may need
satellite, and other places could densify their fiber and their
fixed wireless, and that might be a better combination. But we
think the States should decide--look. The great thing is we
have new deployments of new networks that I think are so much
better on all fronts. Fiber is always going to be the
foundational network that everything comes back to, wireless,
fixed, and satellite.
Mr. Landsman. Fiber doesn't have at the moment--and I am
not trying to be funny or--fiber doesn't have a, you know, the
world's richest man saying, you know, ``Invest in fiber or
broadband.'' And I agree. I mean, obviously every State is
different. But we have to provide the regulatory and investment
framework. And I am worried, as we all should be, that, you
know, one guy and the thing that he owns, Starlink, is going to
have more influence than what we know to be true, which is
that, yes, every State is different, but for the most part,
fiber and the broadband is the most reliable.
Can you just share what you were going to say? It was a
good question in terms of where should we go and what we should
be investing in. I just want to----
Mr. Ramzanali. It is a good question. I think you are right
that--I would actually say the work has already been done. This
subcommittee--this committee--this Congress passed the
Bipartisan Infrastructure Law that had specific instructions on
how to implement the law. The work was done to think about what
that means for different technologies and the NTIA.
This administration has paused that. They want to revisit--
they get the authority to put their policy priorities on top of
it. But that work was done. That is how we got to a fiber
preference.
The way I think about this historically is rural
electrification. We got electricity to all of America. We would
have never put up with a second-class technology for
electricity going to rural Americans.
Mr. Landsman. That is a really good way to frame it. I do
think speeding this up--I mean, we have to get this stuff out
much more quickly. And so hopefully that bipartisan commitment
is there and that will get to a good place because it goes
beyond BEAD and everything else, but we have got to get things
out much more quickly.
That is all. Thank you so much. I yield back.
Mrs. Fedorchak. The Chair recognizes Mr. Goldman for 5
minutes.
Mr. Goldman. Thank you, Madam Chair. Thank you to the panel
very much for being here today.
Mr. Vasishta, how is NVIDIA ensuring that AI use and
telecom infrastructure is resistant to foreign interference,
especially from adversarial state actors like China?
Mr. Vasishta. So NVIDIA, we build the infrastructure that
enables the AI to be leveraged by our ecosystem of partners,
whether it be original equipment manufacturers or telecom
operators or people developing software.
So as the vulnerabilities of the layers of the applications
and the software on top of that, a lot of our partners are
building that in. What we have done, though, is we have built
in the infrastructure underneath that to enable much of that--
those software vulnerabilities to enable to be exposed.
And I will give you an example of what that means. Often,
cybersecurity can be enabled and created by anomaly detection.
There is a normal pattern of use and data, and then there's
anomalies. And to be able to detect those anomalies fast and at
line rate, meaning as they happen, can happen within that
infrastructure that we build. We have some silicon capability
that allows that to happen and the connectivity to allow that
to happen, right at the edge of the network. So it doesn't have
the opportunity to infiltrate into the cloud.
So that is just part of and just an example of some of the
things that we are doing in working with our ecosystem.
Mr. Goldman. Thank you very much.
Mr. Pickering, Fort Worth in my district has a growing film
industry, with over 550 million in economic impact since 2015.
How can INCOMPAS's work on AI-driven network improvements, like
better 5G and fiber connectivity, help support Fort Worth film
studios with real-time virtual production and data-heavy post-
production?
Mr. Pickering. Just like every sector of the economy, it is
enabled by the network. It is enabled by the infrastructure.
And if you have a creative movie production, the ability to use
AI and then the networks will allow better quality, better
performance, better distribution, and better and higher value.
There is going to be great challenges as we look at the
property rights and the intellectual property in this sector.
But as far as the infrastructure that then enables every sector
to grow, I am excited about the future for each sector.
Mr. Goldman. Great. Thank you very much.
And Mr. Shea, is DeepSig currently seeing any risk of IP
theft or cyber infiltration targeting U.S. defense AI
contractors? And what countermeasures are you developing?
Mr. Shea. No, we are very concerned with that. We are not
aware of any exfiltrations have happened of our data, but we
spend a lot of effort protecting our information, you know,
having cyber threat detection in our network. But it is a major
concern because we know AI technology is right at the forefront
of where the world is going, both on the commercial and defense
side. So it is a high priority for us.
Mr. Goldman. All right. Thank you very much.
Madam Chair, that is all I have. I yield back.
Mrs. Fedorchak. The Chair recognizes Mr. Fry for 5 minutes.
Mr. Fry. Thank you, Madam Chair. You are doing a great job
today, so appreciate that.
Mr. Pickering, I was perplexed a little bit. You were
talking with Mrs. Houchin earlier about barriers that slow down
AI-related infrastructure buildout. And two things that you
mentioned were permitting reform and BEAD flexibility. And I
think y'all talked about the BEAD flexibility a little bit. But
specifically, when it comes to permitting reform, what do you
envision that looking like from a specific policy standpoint?
Like, what specifics----
Mr. Pickering. So I will give you an example of two
successful permitting reforms that have happened in recent
history. One was for the small sale deployments that went with
5G. We created a Federal framework with shot clocks and time
certainty, and that if the jurisdiction did not approve within
the time certain, it would be deemed granted.
And so the times of certainty and predictability is
critical. ``One touch, make ready,'' which means that if one
fiber deploys on a pole, that they do all the deployments at
one time, one touch, instead of doing six different, you know,
service trucks for each different company. And that accelerates
the deployment of a long pole.
So there's some commonsense things, but everything is about
one touch, did once, one review, time certainty, those types of
principles on permitting reform. And if you can have national
projects for grids, pipelines, transmission, fiber routes that
are multi-State so that you can coordinate a whole-of-
government approach to deploy the infrastructure that we need
for the AI race against China.
Mr. Fry. At least when it comes to the permitting
perspective of six different points, one different trigger--or
one trigger, would that also--in your mind, if a permit was
previously issued, say, a few years ago, would that also be
almost grandfathered in, that you could use that existing
permit for----
Mr. Pickering. I will give you an example. If you have
multiple agencies--a lot of times, they will all require their
own environmental review when you should just have one
environmental review that would then be adequate for all the
different agencies. And if you have a review that is a year ago
or 2 years ago and then you have a new part of a construction,
that previous review should be considered as adequate for the
current project if it is not substantially different.
And so those types of commonsense reforms that I think can
protect and preserve our resources and our communities and at
the same time speed the deployment that everybody needs today.
Mr. Fry. Thank you. Thank you for that.
Mr. Vasishta, NVIDIA is on the frontlines of both AI and
the telecom infrastructure. What role do you see for AI-native
wireless networks in the transition to 6G, and how can Congress
support companies leading that effort in that transition?
Mr. Vasishta. Thanks for the question.
As I said in my opening remarks, AI-native wireless is
really what 6G is going to be about. And what I said earlier is
that there is a requirement, really, for help between public/
private partnership to make that a reality. We need a lot of
research to come straight to production. So with this AI WIN
Project that NVIDIA is part of with other companies, we are
already embarking upon taking research from, say, Mitre to
production with, say, T-Mobile as fast as possible and getting
guidance from that.
A lot of that relies upon research that is coming out of
the universities, and so the accelerant and the enablement of
AI-native wireless research. There is only one other company--I
said earlier China, but I really meant one other company that
is kind of working on this AI-native approach, and that is
Huawei, on the world stage. And so, once we are able to
accelerate this within the U.S., we will then be able to take
these same developed platforms out globally and once again be
able to create a global platform for telecommunications within
those standards committees.
Mr. Fry. Thank you for that.
You have emphasized software-defined networks as a game
changer. How do these networks enhance both performance and
cybersecurity compared to our traditional infrastructure?
Mr. Vasishta. Yes, so traditional infrastructure has more
of a closed propriety system. Now, you might think that is a
positive, but the benefit of having software defined is
continuous integration, continuous deployment so you are able
to add features very quickly, you are able to implement new
capabilities much faster than if you had to do that in
hardware.
AI-native approaches, like integrated sensing and
communication, which can be--can be threat detection, you can
really sense the airwaves for threats--that can be a software-
defined feature that can be integrated by defense but also in
commercial reason--commercial aspects.
Mr. Fry. Thank you for that. I see my time has expired.
Madam Chair, I yield back.
Mrs. Fedorchak. The Chair recognizes Ms. Kelly for 5
minutes.
Ms. Kelly. Thank you so much.
I am very excited to see the emergence of artificial
intelligence. However, like many emerging technologies, it
needs to be implemented with care, weighing the risks posed to
American consumers, like my constituents back home in Illinois.
As I discussed on a very long markup not too long ago, I
joined many of my Democratic colleagues in concern about the
reconciliation package including a 10-year moratorium on State
and local enforcement of AI laws. Such a moratorium without
Federal safeguards in place leaves Americans unprotected from
data-driven discrimination in critical areas, such as housing,
employment, credit, education, healthcare, and insurance.
It is crucial that we stop bowing to Big Tech and start
doing our jobs to protect American consumers by addressing the
potential for discriminatory outcomes, especially as AI
technologies advance at an unprecedented pace, both
domestically and internationally, which brings me to my next
point. It is critical that the U.S. is positioned to win the
race for global AI leadership, which I believe all of us want
that.
Mr. Ramzanali, how will the BEAD Program's investment in
futureproof internet infrastructure, like fiber, allow America
to remain a world leader in AI innovation?
Mr. Ramzanali. I appreciate that question and your hard
work in this area.
As I said in my opening, America can't lead in AI if all
Americans don't have access to AI. The way that happens,
especially as we think about where the technology might go that
requires higher-capacity throughput, is through futureproof
networks like fiber.
Ms. Kelly. Does anyone have anything else to add? No?
In your testimony, you state that AI's power and usefulness
will be fundamentally limited if all Americans are not able to
access that use AI. Recognizing this, how does the President's
unilateral decision to stop funding programs providing digital
skills, including AI training, to such groups as seniors,
veterans, and the disabled jeopardize America's ability to
innovate in AI ahead of places like China or the Middle East?
Mr. Ramzanali. The way to think about the Digital Equity
Act programs--maybe I will give a couple of examples that were
in the letter to the record that was submitted earlier. There
is a program in rural Kentucky that was hoping for a digital
equity grant that helps seniors with digital skills, including
how to use the internet for job applications. They were
helping--they were going to help the seniors also use AI in a
way. And so that is one example.
Another example is in Hurricane Helene, the areas that
experienced that disaster, they were doing device access. Those
are the kinds of people that need our help in a time like that.
To me, those kinds of programs deserve our support.
Ms. Kelly. Thank you. And I yield back.
Mrs. Fedorchak. Thank you. The Chair recognizes Mr. Kean
for 5 minutes.
Mr. Kean. Thank you, Chairwoman. And thank you to our
witnesses for being here today.
Mr. Shea, I recently heard from a constituent who lives in
Warren County, a rural part of my district in New Jersey. Her
small town is tucked between two steep ridges that flank either
side of the Delaware River. Because of the unique geographic
features of this area, she and her neighbors frequently have
unreliable cell service and even difficulty reaching emergency
services.
In your testimony, you discussed the cases of integrating
AI technology into consumer wireless communications. What about
for an application like--what about for an application like
this? What integration of AI technology can help overcome the
challenges posed by difficult geography, like this one, to keep
people connected and able to reach first responders when
needed?
Mr. Shea. Well, actually, the NTIA--first NOFO award we
received was about how to come up with better ways of measuring
the local environment to assure high-quality service. So I
think with AI, what is called ``digital twin technology,'' we
are getting a much better understanding of how signals
propagate and where we can put in small cells to fill in these
gaps.
So AI, I think, is going to be revolutionary for these type
of applications where people have poor service. And I think we
are going to be seeing the fruits of that technology within the
next couple of years.
Mr. Kean. Thank you.
Mr. Pickering, in your testimony, you discuss how
deployment of technologies, like fiber networks, is a necessary
input to the management of AI applications. Given that
responsibility for these regulations is shared among several
different stakeholders, including the State and local level,
how can Congress best facilitate smart, forward-looking
policies that will enable us to compete in AI?
Mr. Pickering. I think there is a way in the permitting
reforms that this committee will put forth that will respect
the local jurisdictions and the counties and the States and how
they have responsibilities on permitting. What we have tried to
do as an industry is to create a blueprint of best practices
and then find voluntary incentives and community engagement.
But on national deployments, multi-State, I do think that there
is a rural and across Federal lands and Federal properties that
this committee can really speed and accelerate the time to
deploy.
I have been talking earlier, we are in a race with China.
We need to build as fast as possible. They are not having
permitting delays like we have, if they have permitting at all.
And so we need to find a way to speed at every level and
give the resources and the technology and the tools and the
incentives for the best practices with time certainty,
transparency, and the best technology tools to permit as fast
as possible.
Mr. Kean. Thank you.
Mr. Vasishta, I agree with you as to the critical
importance to making sure the United States is a leader in the
future of wireless communications on the global stage. What
steps do we need to take to make sure that American innovators
are leading on AI within international standard-setting bodies?
Mr. Vasishta. So let me take that as a telecommunications
question.
Firstly, as we start to look at the definition of the 6G
standards, which has already started to happen, America needs
to have a very strong voice in the 6G standards, both
corporately as well as from the NTIA. And I think that is
starting to happen. But we need to really double down on that
to make sure that as we define those standards, those standards
are defined in a way that is meaningful to our advancement as
well as AI RAN, as we call it, which is the infusion of AI and
the radio access network. And as I said before, that is
probably something that many others are not really thinking
about. But it gives us, within the U.S., the opportunity to
take that leadership position because we have that AI
leadership position.
The other thing, of course, is to make sure that we have
most of, if not all, the developers in the world developing AI
on America infrastructure. That is very important. Many
developers around the world don't reside within the U.S., but
we need to make sure they have access to that American
infrastructure because that is what improves the capability and
the performance of our AI.
Mr. Kean. Thank you. I yield back.
Mrs. Fedorchak. The Chair recognizes myself for 5 minutes.
So AI has been described as not only a powerful tool but an
incredible weapon. And I don't feel like the public necessarily
understands that or sees it exactly that way. But given that
China today produces 10,000 terawatt hours of power a year and
the U.S. produces 4,000 terawatt hours of power a year, we are
already quite behind.
So given this, I am wondering, you know, do you share this
concern, Mr. Pickering, about maybe China being in a better
position? If so, what do we need to do about that? And then
also, to comfort me and my colleagues, what barriers is China
facing in their deployment of AI?
Mr. Pickering. So far, you know, our country--as you look
at what is powering the data centers that are running the large
language models' intensive energy demand, I think new solutions
that would bring behind-the-meter or dedicated energy to the
data center is a critical reform that is needed to meet the
demand needs today that would not overtax the existing grid and
the residential customer. So that is one solution.
Too, as we have talked about the permitting reforms--and
this committee in reconciliation has something that addresses
multi-State pipeline construction and being able to have a
consolidated review and a time-certain review on that.
As we look at traditional energy meeting current demand,
how can we also look at future options and solutions? And SMR
technology, the small modular nuclear reactors, fusion
technology, those types of things that can give us clean
energy, reliable, abundant energy, and a very small footprint
that is much safer and much more sustainable--getting those
technologies as quickly as possible into the market. So X-
energy is one of our member companies that has a big agreement
with AWS and Dominion utility in Northern Virginia. It came out
of one of our DOE labs. It is an amazing advancement in nuclear
technologies, very similar to what we do in our nuclear fleets.
And so those types of new solutions long term. But in the
short term, building as fast as we can, and all-of-the-above
strategy, and giving data centers the ability to have behind-
the-meter solutions or dedicated solutions for the demand
today.
And no, China does not have barriers or impediments that we
do.
Mrs. Fedorchak. Shoot. I was hoping there were a few, at
least.
I have questions for a few other of you, but I do want to
invite you all--my office--I am very concerned about having the
power to meet the demand of the moment for AI, and so my office
is leading an AI energy working group, and I would invite all
of you to participate in it. I know, Mr. Pickering, you are,
and hopefully others are as well. But we are very much working
on this framework of solutions to meet the energy needs.
Mr. Shea, you haven't gotten a question for a bit. I wanted
to ask you, from your vantage point as a smaller innovator,
what are the specific barriers to deploying AI-driven wireless
infrastructure, especially in rural underserved areas, and what
kind of changes could Congress bring about to help that to
address those barriers?
Mr. Shea. I think for Congress to continue the push at ORAN
is probably the most help that can be provided because we are
finding the ORAN vendors we work with more forward looking on
AI. They are looking for ways to leapfrog technology, which is
what we need to compete with China. So I think the openness of
that type of standard is what is key to make this all happen
quickly.
Mrs. Fedorchak. Excellent.
And, Mr. Ram-za-me--how do you say your name?
Mr. Ramzanali. ``Ram-za-nall-ee.''
Mrs. Fedorchak. Ramzanali. You stated with Mr. Menendez
that the moratorium--or Mr. Menendez stated that the moratorium
on new State regulations that we included in the One Big
Beautiful Bill erodes the current frameworks passed by States,
and you agreed with that statement. So help me understand how
that is so when the moratorium is on new regulations and
doesn't do anything to the existing regulations that States
have already enacted.
Mr. Ramzanali. So there's questions about how it applies to
privacy laws that are being--privacy bills that are being
considered and privacy laws that exist, because some of them
use definitions for automated systems that could be caught up
in the way that the bill is written. That is one way that it
could directly go at the privacy concerns.
Mrs. Fedorchak. But you agree that the moratorium is on
States developing new regulations, not the existing ones.
Mr. Ramzanali. I would have to read it more closely, but I
believe there has been some debate about how that would be
applied. And that is the kind of thing that would take a while
for courts to work out too.
Mrs. Fedorchak. OK.
Let's see. Seeing there are no further Members wishing to
be recognized, I would like to thank our witnesses for being
here today. I ask unanimous consent to insert in the record the
documents included on the staff hearing documents list. Without
objection, that will be the order.
[The information appears at the conclusion of the hearing.]
Mrs. Fedorchak. I remind Members they have 10 business days
to submit questions for the record, and I ask the witnesses to
respond to the questions promptly. Members should submit their
questions by the close of business on Wednesday, July 18th.
Without objection, the subcommittee is adjourned. Thank you
all.
[Whereupon, at 1:23 p.m., the subcommittee was adjourned.]
[Material submitted for inclusion in the record follows:]
[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]
[all]