Securing AI Supremacy: Key Takeaways from the Senate Commerce Committee // Philip Brazeale

Securing AI Supremacy

On May 8, 2025, the U.S. Senate Commerce Committee convened a pivotal panel featuring some of the most influential voices in the artificial intelligence (AI) and semiconductor space. This wasn’t just a discussion about data ethics—it was a strategic session aimed at securing America’s long-term dominance in the global AI race.

The panel featured:

Sam Altman, CEO of OpenAI
Dr. Lisa Su, CEO of AMD
Michael Intrator, CEO of CoreWeave
Brad Smith, Vice Chair and President of Microsoft

Together, these leaders laid out a vision not only for how the U.S. should lead in AI development but also for how federal policy can accelerate innovation, secure national interests, and prevent the nation from falling behind geopolitical rivals like China.

1. The Strategic Imperative: America Must Lead the AI Race

One of the panel’s most pressing themes was the urgent need for the U.S. to lead the development of artificial general intelligence (AGI) and, eventually, artificial superintelligence (ASI)—much like the country pioneered the internet in the 1990s. Several strategies were proposed to ensure this leadership.

Protocol Leadership: U.S. companies should define the foundational standards and protocols of the AI era, just as they once did with internet technologies like TCP/IP and HTTP. These protocols will shape how AI systems communicate, evolve, and govern data.

Onshoring AI Chip Production: A recurring concern was the dependence on foreign semiconductor manufacturing—especially in a geopolitical climate where adversarial powers like China are actively working to close the technological gap. They emphasized the importance of not only manufacturing high-performance AI chips on U.S. soil but also controlling the full production pipeline, from design to fabrication.

This includes establishing trade regulations that explicitly prevent countries like China from acquiring cutting-edge U.S.-made chips. Ensuring a technology gap would secure U.S. leadership by default, creating a strategic moat around innovation.

Capital Investment and Government Support: Continued—and increased—federal support is critical to accelerate investment in American AI companies. This includes direct funding, tax incentives, and public-private partnerships. All of this is necessary to prevent rival nations from surpassing U.S. efforts through sheer volume of funding.

The Rise of DeepSeek: The panel specifically highlighted DeepSeek, a Chinese AI project developed by a team of recent college graduates. The project’s deliberate avoidance of “traditional thinking” allowed it to iterate rapidly—an alarming sign that China is fostering agile, unconventional R&D environments to bypass the U.S.’s established technological edge.

2. Regulatory Acceleration and Infrastructure Expansion

A clear throughline in the panel discussion was the urgent need to modernize and streamline regulatory frameworks across sectors that underpin AI development—particularly energy, construction, and immigration.

A. Unlocking Energy Production from All Sources

Panelists agreed that energy availability is becoming the ultimate throttle on AI development. As demand for compute power surges, AI companies are running into energy constraints not just in data centers but in entire regions. The consensus: unlock all energy sources.

This includes:

Natural gas and nuclear energy, both of which are dependable and scalable.
Fusion research, which remains long-term but promising.
Solar, particularly in sun-rich states, as a supplementary option.

Sam Altman emphasized that lowering the cost of energy directly raises the quality of life. The implication: a future where cheap energy powers not just AI, but public prosperity.

B. Streamlining Physical Buildouts

The current pace of permitting and construction approval is too slow for the speed at which the AI industry needs to grow. AI companies want to build:

Data centers
Power generation facilities
Cooling and networking infrastructure

But the bureaucratic lag—from environmental reviews to zoning laws—creates costly delays.

The panel urged Congress to create fast-track processes specifically for AI infrastructure. Michael Intrator of CoreWeave went further, describing the risk of retroactive regulation—where companies make billion-dollar investments only to face new restrictions—as “sub-optimal” and potentially fatal to U.S. competitiveness.

C. Consistent, Long-Term Regulatory Frameworks

Both Altman and Brad Smith warned against the dangers of a shifting regulatory landscape. For long-term infrastructure investments to make sense, the rules governing those investments must stay consistent across administrations. If not, companies may:

Withhold investments
Redirect R&D to friendlier jurisdictions
Accelerate offshoring of talent and facilities

This is similar to the early 2000s, when much of Europe’s tech sector migrated to the U.S. due to rigid EU regulations. He fears history could repeat itself in reverse.

D. Immigration Policy: Expanding the H-1B Visa Cap

To stay ahead in AI, the U.S. must attract and retain the world’s top talent. One proposed solution: expand the H-1B visa program, allowing highly skilled researchers, engineers, and scientists to work and stay in the U.S. longer.

The idea isn’t just to accelerate domestic development, but also to stall international competitors. When the best minds relocate to the U.S., they’re no longer building for foreign labs. However, this also raises the risk of corporate espionage.

3. The US vs. the EU: Diverging Regulatory Models

A comparison between the regulatory philosophies of the U.S. and European Union framed much of the discussion. The panel was unanimous: the EU’s stringent rules are slowing its own innovation and leaving it less competitive in the global AI race.

A. Europe’s Delay in Model Deployment

Sam Altman noted that OpenAI cannot release new models or features on Day One in Europe. Due to the EU’s labyrinthine compliance frameworks, model rollouts are delayed, often by months. This not only stifles product innovation but makes Europe a second-tier market for AI technologies.

B. US-UK Free Trade Agreement Opens Doors

Full Announcement The US and UK will open up free trade between each other, and part of that will include services. Meaning that it will likely be easier to allow for software developers to work with each other’s companies.

C. Investor Flight from the EU

As Altman put it, investors are becoming wary of the European AI scene—not because of a lack of talent, but due to overregulation. The panel warned that capital will naturally flow toward jurisdictions that:

Offer legal clarity
Minimize red tape
Don’t try to pre-regulate future innovations

In their view, the EU is becoming a case study in how not to foster a high-tech industry.

D. Free Speech and AI: Philosophical Clashes

The First Amendment was a recurring subtext. While the EU often imposes speech restrictions tied to hate speech and misinformation laws, U.S. companies operate under a different regulations aimed at: free speech first, with opt-in safety mechanisms.

Altman argued for giving adults the tools to explore AI on their own terms, while also providing guardrails for misuse. He was careful to distinguish AI from social media—suggesting that while the two often intersect, they shouldn’t be regulated in the same way.

E. Industry-Led Standards and Avoiding Regulatory Patchwork

Sam Altman issued a stark warning against a “patchwork” of differing AI regulations across U.S. states, deeming it a potential detriment to the industry’s growth and a national security threat that could undermine America’s lead over China. He advocated for a light-touch, uniform federal regulatory framework, cautioning that allowing states to individually regulate would create significant issues. None of the other panelists disagreed with this assessment.

Echoing this, Michael Intrator highlighted that the risk of making large-scale infrastructural investments only to have them stymied by subsequent, unforeseen regulations is “sub-optimal” and unacceptable from an investment standpoint. He argued that such regulatory uncertainty would deter the massive capital commitments needed for meaningful progress, potentially causing technology industries to relocate to more stable jurisdictions, much like the EU tech migration to the US in the 2000s.

4. China, Export Controls, and National Strategy

The urgency was clear: the U.S. must secure its leadership in AI not just as a matter of economic competitiveness, but as a matter of national security and global influence.

A. Deep Seek: China’s Rising AI Threat

Sam Altman and others repeatedly referenced Deep Seek, a Chinese language model that has made rapid progress—reportedly developed by a cohort of recent college graduates, deliberately selected to avoid traditional thinking from older colleges. The implication was clear: China is not just copying U.S. innovations; it’s iterating aggressively and may surpass in key domains without coordinated countermeasures. The panel stressed that the emergence of Deep Seek underscores the necessity for the U.S. to vigorously back its domestic AI industry.

B. Export Controls and the Chip Race

The senate emphasized the importance of tight export controls on cutting-edge semiconductor technologies, particularly for AI-focused GPUs. Current U.S. regulations restrict China’s access to advanced chips (such as those from Nvidia and AMD), but regulatory gaps remain and regulations with trade partners must ensure China cannot access these critical components.

C. Domestic Resilience and Onshoring

All four panelists agreed that onshoring AI-related manufacturing and infrastructure is vital. This means not only chip fabs (such as those incentivized by the CHIPS and Science Act), but also the broader stack:

Power infrastructure (plants, substations, cooling)
Fabrication, testing, and packaging facilities
Talent pipelines and immigration frameworks
Data center construction—fast-tracked through regulatory streamlining

The underlying message: if the U.S. wants to win the AI race, it must own the means of production, not just the IP.

D. The AI Stack as a Strategic Export

Perhaps one of the boldest proposals came from Sam Altman: Make the U.S. AI stack the global default.

In other words, the U.S. should build AI standards, protocols, and platforms so foundational and so widely adopted that the rest of the world aligns around them—just as it did with the TCP/IP protocols and the Internet. This involves leveraging U.S. government power and authority to ensure global adoption and protect the industry from external competition.

This vision includes:

Onshored model training and inference infrastructure
Open standards, but driven by U.S. research institutions and private companies
Exportable cloud services that comply with U.S. norms
Diplomatic advocacy to get allied nations on board

In his view, this is how the U.S. wins: not just by building better AI, but by embedding its values and standards into the global technological fabric.

5. Project StarGate and State-Level Investment Strategy

How can individual states position themselves as key players in this AI transformation?

This section of the panel explored what state governments, utility providers, and local ecosystems must do to attract investment from AI labs, chip manufacturers, and data center operators.

A. Power Availability is the First Gatekeeper

The single most important asset? Power. Not just existing capacity, but future-proof availability, with a need to build much more power supply to protect all ratepayers.

The compute demands of training frontier models and running inference at scale require massive and stable power supplies from diverse sources including natural gas, nuclear, fusion, and potentially solar. And that means states that want to become AI hubs must:

Expand energy production aggressively
Offer redundant supply pathways
Invest in grid resiliency

Altman also emphasized that the cost of electricity directly correlates with quality of life and long-term prosperity. As energy becomes cheaper and more abundant, states become more attractive not just for AI investment, but for high-tech growth more broadly

B. Skilled Workforce and Immigration Pathways

Infrastructure alone isn’t enough—people build the future.

Panelists stressed the importance of a skilled local workforce.

C. Streamlined State-Level Regulations

Michael Intrator drove this point home: any region that wants to host AI infrastructure must remove red tape.

From environmental permitting to zoning and construction regulations, fast-tracking is key. The investment required to build an AI-grade data center or power plant is measured in hundreds of millions or even billions of dollars. If companies face post-investment regulatory hurdles—or worse, if rules change midstream—they may abandon the project altogether. Texas was mentioned as a key partner in demonstrating how streamlined regulations can attract such development.

Brad Smith added that these regulations must be predictable and long-term, not subject to changing political winds or administrations. AI infrastructure takes years to build and decades to amortize; any uncertainty threatens the entire business model.

D. Public-Private Utility Partnerships

One of the more unexpected moments came from Brad Smith, who detailed Microsoft’s approach to utility investment. He revealed that Microsoft petitioned for a rate increase on itself—voluntarily. Why? To offset the capital costs required by power companies to expand generation capacity and meet Microsoft’s upcoming demand. He argued that companies have a responsibility to make investments with utilities to increase baseline capacity.

This signals a future in which major companies don’t just consume power—they partner with utilities, co-fund infrastructure, and contribute to broader regional development as co-investors in the infrastructure of innovation.

6. Science, National Labs, and the Role of AI in Fundamental Research

A particularly compelling segment of the panel focused on the intersection of AI and fundamental scientific discovery, underscoring that science research has been the core of U.S. innovation and economic growth for over 50 years. While much of the public discourse centers around AI as a commercial or industrial force, the panelists reminded the audience that AI is rapidly becoming an essential tool in science itself.

A. AI as an Accelerator for Scientific Discovery

Sam Altman and Brad Smith both emphasized that AI is poised to redefine the pace and methodology of scientific research.

Examples include:

Material Science: Using AI to discover new materials for superconductors, batteries, or semiconductors.
Drug Discovery: Leveraging language models and protein-folding AI to model complex biological systems and accelerate new therapeutics.
Climate Modeling: Integrating AI to build more accurate, granular, and fast-updating models for climate change forecasting.

B. National Labs as AI Powerhouses, Enhanced by AI

Panelists noted that these national labs have already seen direct improvements by integrating AI into their own processes.

Vital partnerships between national labs and AI labs will continue to be crucial for future advancements.

The vision is clear: AI as a national accelerator for scientific discovery, rooted in the public interest and driven by collaboration across academia, government, and industry.

7. The Geopolitical Stakes of AI: National Security, Chip Supply, and Global Competition

As the panel progressed, it became increasingly clear that AI is not merely a technological challenge—it is a geopolitical one. Multiple senators, including Senator Ted Cruz, homed in on national security risks and U.S. strategic interests surrounding AI development.

This portion of the discussion touched on everything from semiconductor supply chains to AI arms races, underscoring that leadership in AI confers more than just economic power—it could redefine global influence and security in the 21st century.

A. Semiconductors: The Critical Bottleneck

AI’s dependency on massive computational resources led to repeated references to semiconductors, particularly GPUs (Graphics Processing Units) and AI-optimized chips like those from NVIDIA or AMD.

Brad Smith and Sam Altman both made it clear: No compute, no frontier AI.

Senators and panelists alike emphasized the importance of:

CHIPS Act implementation: Ensuring that U.S. manufacturing capacity for advanced chips is revived and scaled rapidly.
Onshoring global supply chains: Reducing dependency on foundries located in a geopolitically sensitive area.
Encouraging domestic R&D: Building incentives not just for fabrication but for chip design innovation within the U.S.

8. Societal Transformation: Employment, Adaptation, and the Challenge of Deepfakes

A. Employment Impact and Human Adaptation

Sam Altman projected that the impact of AI on employment could be both rapid and substantial, potentially outpacing transformative shifts seen in past technological revolutions. He views AI as arguably the largest technological revolution in human history. Despite the potential for disruption, Altman expressed an optimistic view of human adaptability, believing that people will adjust quickly. His proposed solution is to empower individuals by putting AI tools directly into their hands, allowing them to discover new employment solutions and navigate the changing landscape. He urged an approach marked by humility when confronting these monumental changes.

B. Addressing the Deepfake Dilemma

The proliferation of deepfakes and manipulated media was a key concern. Brad Smith advocated for a proactive approach, suggesting that AI itself should be employed to identify and detect deepfakes. Once identified, content that crosses legal or ethical lines should be taken down.

Sam Altman offered a complementary view, acknowledging that stopping the creation of deepfakes entirely is likely impossible. Instead, he emphasized the critical need for societal education. Society must be equipped with the knowledge and tools to identify manipulated content and respond appropriately, fostering a more resilient and discerning public.

Original Notes

Ensuring that that US leads the AI development race to AGI/ASI. Much in the same way that the US led the development of the internet in the 90s.
1. US AI companies lead in the creation of the standardized protocols.
2. Onshoring chip production for AI hardware.
  1. Regulations with trade partners to ensure that China isn’t able to access the cutting edge chips to create a guaranteed production lead for all US based companies.
3. Continued government support to increase capital investment in US AI companies ahead of foreign companies.
4. Deep Seek is a formidable competitor, and requires the US to back the US AI industry.
  1. It was developed by mostly fresh college graduates to prevent traditional thoughts from being brought into the process.
Reducing regulations to allow for:
1. Increased power production from all sources.
2. Streamlined processes for the chosen companies to build data centers, power plants, and any other needed physical facilities.
3. Long term regulation framework to ensure future administrations can’t change the rules that the AI companies are following. See the above point about increasing investment.
4. Increase the allowance for H-1B Visas to bring in high talent researchers for every level of AI development. (This would not only accelerate development, but would likewise stall other countries as their own talent leaves. However, the inverse issue would be corporate espionage risks go up.)
US regulation environment vs EU’s
1. OpenAI is not able to release it’s new models/features day one in Europe due to the regulations that they’ve put in place which stalls the roll out process.
2. US Trade Deal With UK 2025-05-08 “The US and UK will open up free trade between each other, and part of that will include services. Meaning that it will likely be easier to allow for software developers to work with each other’s companies.”
3. The EU will continue to fall further behind as investors choose not to deal with the EU’s regulations around AI and become more aligned with the US stack.
4. Speech regulations around social media will also have an outsized impact around the development of AI. Obviously stands in contrast to the US’s first amendment views on freedom of speech.
5. Sam doesn’t view AI and social media as the same, but does believe that adults should have the freedom to use AI as they see fit, and trust them to be responsible with it. While also creating some safety guardrails.
6. The AI industry must be left alone to discover and design the standards and then the government needs to codify those into laws and regulations to support the industry.
7. Sam Altman believes that a “patchwork” regulatory framework across the states would be a determent to the industry, and national security threat toward the US remaining its lead ahead of China’s development.
  1. Supports a light touch framework from the federal, but allows states to regulate would be an issue.
  2. None of the other panelist disagreed with his assessment.
8. Michael Intrator points out that the risk of making a large scale infrastructural investment that could then be regulated after the fact is “sub-optimal”. (Regulations that interfere with cutting-edge research will result in technology industries relocating to other jurisdictions that don’t regulate. Much as how EU tech moved to the US in the early 2000s)
  1. The investment requirement to make meaningful progress, are so large that any risk of regulation hurdles after the fact are prohibitive and not acceptable from the investment side.
Get the US government to use it’s power/authority to ensure that the rest of the world has to us the US “AI Stack”, and product the industry from outside competition.
Microsoft’s testing process for co-pilot and all AI models.
1. Multiple rounds of red teaming.
2. External testing is important even after internal test and release.
AMD chip development
1. Chip and Science Act helped.
2. Open models of AI chips allows the innovation to happen faster.
How states can attract investment from Project Star Gate
1. Power availability
2. Skilled workforce
3. Streamline regulations to allow for fast development
  1. Texas mentioned as a key partner toward this
4. Build much more power supply to protect rate payers
  1. Sam asserts that as cost of energy goes down quality of life goes up.
  2. Natural gas, Nuclear, Fusion, and possible solar
  3. Brad Smith thinks that the companies have a responsibility to make investments with the utilities to increase the baseline capacity.
    1. Microsoft argued for a rate increase on themselves to help offset the investment requirements for the infrastructure needed to deliver the power to their data centers.
5. Brad Smith echoes that any regulation around the industry must be done in such a way that it’s consistent across the administrations and subject to the changing winds of national politics.
Science Research is the core to the US innovation and economic growth for the past 50+ years.
1. National labs have seen direct improvements by using AI in their own processes.
2. Partnership between the labs and AI labs will remain vital into the future.
Privacy is important, but the new AI models will require more personal information from the users than any technology beforehand.
On the question of if the US is currently leading the AI development race:
1. Sam Altman believes that the US leads and continue will do so.
  1. The US infrastructure and integrated supply chain must be developed and maintained within the US.
  2. He wants to increase foreign talent being brought to the US.
2. Brand Smith believes that China will continue to be a major investment opportunity, but that if the US continues to bring the brightest minds to the US and supports the industry as it develops, that US can win the race.
Employment Impact:
1. Sam Altman thinks that the impact could happen rapidly. Faster than seen in the past.
  1. Thinks the best solution is to put the AI tools in the hands of the people and allow them to find the employment solution.
  2. Believes AI to be the largest technological revolution in the history of humanity.
    1. Humans will adapt and rapidly.
    2. Approach it with humility.
Deep Fakes:
1. Brad Smith thinks that AI should be used to find the fakes and then take them down where they’re found to have crossed the line.
2. Sam Altman thinks that the creation can’t be stopped but that society needs to be educated on how to identify it and respond.
Search Engine:
1. AI will replace some aspects, but Google will remain strong competitor.