A June 11 hearing paired testimony on opaque data center lending with the AI Bubble Transparency Act, a disclosure bill that would make data center and chip firms report to a federal financial regulator. The bill is designed to outlast the current chip export fight.
The Senate Banking Committee did something procedural on June 11 that has more long-term consequence than Jensen Huang's empty chair. By holding a hearing titled "AI and the American Dream" and pairing it with a new disclosure bill, the committee began treating the AI buildout as a financial-stability question, the same category of risk that produced the post-2008 bank oversight regime, rather than a Commerce Department technology question.
That reclassification is the story. The committee is not adjudicating today's chip export rules or settling the safety fight over frontier models. It is laying the institutional groundwork for a future in which AI compute companies have to report their books to a federal financial regulator. Disclosure regimes of this kind tend to grow into substantive oversight, the way mandatory derivatives reporting after 2008 became the foundation for the 2010 Dodd-Frank rules.
Ranking Democrat Elizabeth Warren invited Huang, the chief executive of chipmaker NVIDIA, to testify about the financial exposure tied to AI. He declined. Warren's office framed the absence as a choice not to answer for a company worth roughly $5 trillion whose processors sit at the center of a projected $7 trillion AI infrastructure buildout by 2030.
The same day, Warren and Senator Richard Blumenthal introduced the AI Bubble Transparency Act, a bill that would direct the Treasury's Office of Financial Research, the federal government's in-house financial-stability research arm, to collect data on AI-related exposures across the system. It also requires public companies to disclose AI-related asset values and capital-market activity. Its premise is the one Warren laid out in her opening statement at the hearing: that the AI boom is being financed through opaque data-center lending, record AI corporate debt, and a chip supply chain that is itself a national-security question.
The financial-stability framing did not arrive in a vacuum. Warren cited what she described as the Trump administration's May 2026 reversal of Biden-era export controls, clearing the way for NVIDIA to sell advanced chips into China. She also pointed to the Commerce Department's early June 2026 admission, as Warren characterized it, that it had failed to enforce the export rules it had on the books. Those two characterizations — the policy reversal and the enforcement admission — gave the Banking Committee its jurisdictional hook. The chip pipeline is not being governed by Commerce, so the committee is asking whether the financial system is the right backstop.
Witness testimony reinforced the dual track. Sarah Myers West, co-executive director of the AI Now Institute, a research group focused on AI accountability, testified on concentration risk in compute and on the lack of transparency around how AI firms are financed. A Forbes summary of the hearing details witness accounts of chip smuggling through shell entities, remote-access loopholes in export rules, a Blackwell-architecture enforcement gap (Blackwell is NVIDIA's current generation of AI processors), and distillation attacks, a technique in which a smaller model is trained to imitate a more capable one without the underlying hardware. The Foundation for Defense of Democracies, cited in Warren's statement, has reported that the Chinese military sought NVIDIA chips before and after sanctions, a fact that ties the financial-stability framing to the national-security framing rather than separating them.
The wire frame on this story is "Huang snubs the Senate," or, more politely, "Huang declines to testify." The substance is the choice of venue. Banking Committees oversee deposit insurance, systemic risk, and consumer finance. Holding an "AI" hearing there, and pairing it with a Treasury-data-collection bill, is a jurisdictional claim. It says: the AI buildout is large enough, leveraged enough, and interconnected enough with the rest of the financial system that the Senate's bank watchdog, not its tech or trade committees, is the right place to watch it.
The disclosure-first approach is also the politically durable one. The 2008 playbook was: get mandatory reporting on the books first, then let the regulatory perimeter expand. A transparency bill is easier to pass than a product ban or a nationalization debate, and the same is true here. The AI Bubble Transparency Act is a reporting mandate, not a control regime. That is the point.
What to watch next is whether the bill advances, whether the Treasury's Office of Financial Research actually begins collecting AI exposure data, and whether other Senate committees defer to the Banking Committee's framing. The June 11 hearing did not settle the chip export fight. It built a parallel track: a financial-stability track running through Treasury that can act whether or not the Commerce track does.