Tech titans hit the Senate: Altman and Su pitch massive AI infrastructure

Silicon Valley power players just took the stand in Washington, and they brought a unified message: if America wants to stay ahead in AI, it needs massive power plants, fewer regulations, and a smarter approach to global trade.

OpenAI CEO Sam Altman, AMD CEO Lisa Su, Microsoft Vice Chair Brad Smith, and CoreWeave co-founder Michael Intrator testified before the Senate Commerce Committee on the accelerating AI race against China.

Demanding a dual revolution

Altman didn't mince words about the stakes, telling lawmakers that AI will be at least as big as the internet, maybe bigger. But to get there, tech companies need a massive buildout of physical infrastructure.

Altman urged senators to back what he called dual revolutions in both AI development and energy production. In short, AI models are burning through electricity at unprecedented rates, and computing power is directly limited by how much juice the grid can supply.

Senate Commerce Committee Chair Ted Cruz framed the hearing around geopolitical stakes, pointing out that China intends to lead the world in AI by 2030. Cruz argued the U.S. faces a choice between leaning into technological innovation or copying European command and control regulations.

While both parties agreed on keeping the U.S. dominant, political tensions broke through. Democrat Sen. Tammy Duckworth blasted recent federal funding cuts pushed by Donald Trump and Elon Musk, arguing that if their DOGE initiative had existed decades ago, the research project that created the internet would have been defunded as wasteful.

Meanwhile, Republican Sen. Bernie Moreno pressed executives on whether previous sustainable energy policies had actively choked power supplies for new AI data centers.

The unintended blowback of export bans

Beyond asking for energy and faster project approvals, chipmakers and AI labs issued a direct warning against overzealous trade restrictions.

Following recent restrictions barring Nvidia H20 and AMD MI308 accelerator chips from being exported to China—a move Nvidia estimated would cost it 5.5 billion dollars—executives argued that extreme bans can backfire.

Su warned that while national security matters, cutting off foreign markets pushes customers toward alternatives. Those foreign alternatives may be less capable today, but guaranteed market share will let them mature fast.

Altman backed that up with an ecosystem analogy. He noted that America wields massive global leverage because the entire world wants to carry an iPhone and search on Google. The U.S. should aim to have its full AI stack adopted globally rather than walling it off and creating an opening for rivals.

Why it matters

This hearing signals a permanent shift in how Big Tech engages with Washington. AI leaders are no longer pitching consumer tools to regulators; they are positioning themselves as critical national defense infrastructure to win policy and energy concessions.

We think Lisa Su's critique of strict export controls hits the nail on the head. AI chips and foundational models require massive global revenue to fund their next-generation R&D cycles. Locking U.S. hardware out of foreign markets doesn't simply deny adversaries chips today; it guarantees that foreign competitors get domestic demand and runway to catch up.

However, there is an inherent contradiction in Silicon Valley's posture. Altman and Su want fast-tracked domestic energy permits and minimal regulatory red tape, yet they also want to keep selling compute abroad while Washington treats AI primarily as a geopolitical weapon. As access to computing power becomes the defining front of international rivalry, keeping AI tech open and globally distributed looks increasingly difficult to pull off.