AI Senate Hearing Notes

Senate Committee on Commerce, Science, and Transportation Hearing

Opening Remarks

AI as a transformative technology, comparable to the internet, driving a new global industrial revolution.
Potential to improve quality of life, create jobs, and stimulate economic growth.
The United States must lead in AI to shape the 21st-century global order, especially against China's AI ambitions.

The AI Race: US vs. China

China aims to lead in AI by 2030, investing heavily across various sectors, including manufacturing and defense.
The US faces a pivotal choice: embrace entrepreneurial freedom or adopt European-style command-and-control policies.

Lessons from the Internet's Dawn

In the 1990s, Washington embraced the internet with a light-touch regulatory approach.
The Telecommunications Act of 1996 deregulated the sector, while tariff agreements protected intellectual property.
A 10-year internet tax moratorium was enacted in 1998 to prevent state laws from hindering e-commerce.

Outcomes of Pro-Innovation Policies

By 2000, the US saw five years of historic productivity and investment growth.
The US became a top tech exporter with substantial private investment in the digital economy.

EU's Regulatory Approach

EU countries pursued heavy-handed regulations that proved costly.
In 1993, US and European economies were similar in size; today, the American economy is over 50% larger, driven by tech and the shale revolution.
Only 6% of global AI startup funding goes to EU firms, one-tenth of the amount going to American companies; blamed on the EU's regulatory approach.

Concerns Over US AI Policy Alignment with EU

The Biden administration's AI executive order casts AI as dangerous and lays the groundwork for audits, risk assessments, and regulatory certifications.
This approach threatens startups, developers, and AI users with heavy compliance costs.

Critiques of Overregulation

A testing regime to guard against AI "discrimination" and government "guidance documents" is seen as paternalistic and unnecessary.
Harmful regulations include Biden's AI diffusion rule on chips and model weights, which would cripple American tech companies and benefit China.

The Importance of Innovation and Adoption

US dominance in AI depends on innovation and adoption.
Innovation drives breakthroughs, while adoption empowers American workers and businesses.

Recent Developments

Over $1 trillion of new AI projects, including investments in Texas, such as the CoreWeave data center in Plano and Project Stargate in Abilene by OpenAI, Oracle, and others (500 billion).
Adopting a light-touch regulatory style will require Congress to work alongside the president, as with President Clinton.

Proposed Regulatory Sandbox

A new bill will create a regulatory sandbox for AI, modeled on the approach taken at the dawn of the internet.
This aims to remove barriers to AI adoption, prevent state overregulation, and allow the AI supply chain to grow in the US.

Ranking Member Cantwell's Remarks

Focusing on winning against China with an open AI approach.
Key areas: computing power, algorithms, and robust data sources.
Continued investment in NSF as a public-private partnership.
Importance of the Chips and Science Act for domestic investment and supply chain security.
Need to lead on future chip designs and implementation, addressing data centers and electricity sources (potentially 12% of electricity demand).
Microsoft's agreement with Fusion Energy in Everett, Washington, for a power source supply.

The Need for Electricians

The US needs hundreds of thousands of new electricians.
Having electricity and data source centers in the US is crucial.

Export Controls

Export controls should not be used as a trade strategy.
Standards should encourage broad distribution of US-manufactured AI chips and technology.
Partners must comply with US rules to prevent circumvention to China, ensure access, and allow US data and cloud-based companies in the market.
The US must move fast to avoid another Huawei situation.

Collaboration

Collaboration is key to innovation, as per Paul Romer's quote.
Avoiding political infighting is essential to get policy right and enable innovation.

Witness Introductions

Sam Altman, CEO of OpenAI (ChatGPT).
Lisa Su, Chair and CEO of Advanced Micro Devices (AMD).
Michael Integrator, CEO and co-founder of CoreWeave.
Brad Smith, Vice Chair and President of Microsoft.

Sam Altman's Testimony

ChatGPT is used by over 500 million people a week and is the fifth-largest website globally.
Significant productivity increases reported by users.
AI development must focus on innovation and infrastructure investment.
The next decade will be about abundant intelligence and abundant energy.
He recounts his childhood computer experiences and their role in founding OpenAI.
Importance of the spirit of American innovation and entrepreneurship.

Lisa Su's Testimony

AMD builds high-performance computing chips for the modern economy, supporting critical missions.
Proudest moments are public-private partnerships, such as supercomputing with the Department of Energy (Oak Ridge and Lawrence Livermore National Labs).
AI is transformative, but US leadership is not guaranteed.
Maintaining leadership requires excellence at every layer of the stack.

Five Key Points

Continue to run faster, ensuring computing availability.
Embrace open ecosystems to foster innovation and reduce barriers to entry.
Focus on a robust domestic supply chain for semiconductors.
Invest in talent to make the US the best place to study and work in AI.
Understand the importance of national security related to export controls but facilitate widespread adoption of US technologies.

Michael Integrator's Testimony

CoreWeave is at the forefront of America's AI infrastructure revolution, managing over 250,000 GPUs across 30 data centers.
Revenue has surged by 12,000%, reaching 1.9 billion in 2024.
Modern AI requires specialized infrastructure beyond traditional cloud computing.
The computing power necessary for advanced AI models has multiplied approximately 100,000-fold since 2018.
Focus on policy elements: strategic investment stability, energy infrastructure development, global market access, and public-private partnerships.

Brad Smith's Testimony

The AI tech stack requires infrastructure, platforms, and applications, emphasizing interconnectedness.
Congress should focus on innovation (more infrastructure, faster permitting, more electricians and investment in universities).
AI diffusion requires skills investment and education.
Export strategies must consider that only 4.5% of the world's population lives in the US.
The goal should be to build machines that help people become better and pursue more interesting careers.

Question and Answer Highlights

Essential elements for US AI leadership: more electricians, broader AI education, streamlined ability to build large things,policies ,sensible supply chain policy.
Incentivizing companies to do business in America: supporting compute, infrastructure development (data centers, electricity), universities, regulatory framework.
Importance of NIST setting standards: establishing a common vocabulary for acceleration.
The future of work, electricity, and public-private cooperation being the future of innovation.
Dr. Sue's point that usage spurs innovation.
Focusing on streamlining the ability to build quickly with energy, which is paramount with the discussion on national resource infrastructure.
Protecting children, which is another discussion that is being prioritized heavily.
The development of Stablecoin ecosystems and electricity is extremely important for the U.S. to succeed.
The issue of bias, also prioritizing inclusion within this technology.
The impact for scientific research, which may become one of the most important contributions to the world from AI.
Concerns about high priced energy, potentially having a negative impact to households and ratepayers as more data centers are put into place.
Emphasizing public-private partnerships as the future of progress and innovation for AI.