06/11/2026 at 10:00AM
The Committee on Banking, Housing, and Urban Affairs will meet in open session to conduct a hearing entitled, “AI and the American Dream: Promoting Innovation, Affordability, and American Dominance.”
Witnesses:
- Mike Flynn, Senior Vice President and Counsel, The Information Technology Industry Council
- David Feith, Senior Fellow, Hudson Institute
- Will Rinehart, Senior Fellow, The American Enterprise Institute
- Dr. Sarah Myers West, Co-Executive Director, AI Now Institute
Myers West:
In the first quarter of 2026, Amazon, Google, Meta, and Microsoft reported spending $130.65 billion in capital expenditures for data centers—71 percent higher than what they spent during the same quarter of 2025. These four companies anticipate spending approximately $700 billion this year. Morgan Stanley analysts project that combined, Big Tech companies will spend nearly $3 trillion through 2028, but only generate half of that amount in cash. To finance this buildout, tech companies are taking on an increased amount of debt: in 2025, big tech firms issued $121 billion in new debt, but this number is anticipated to balloon in the years to come: Morgan Stanley is estimating debt issuance will top $500 billion this year. … Private equity firms are key conduits for the vast amounts of capital going into data centers. In 2025, private equity investment into data center transactions/deals reached $45 billion.
OpenAI is on the hook for over a trillion in deals with other AI firms and chipmakers, including a $300 billion deal with Oracle for its compute infrastructure, $250 billion with Microsoft for its Azure infrastructure (which includes Microsoft’s 26% return ownership share of OpenAI); $38 billion to rent access to Amazon Web Services servers, (a deal which could expand under a revised structure), $22 billion with CoreWeave for use of its data centers, a deal with Google Cloud, a “strategic partnership” to deploy 10 gigawatts of AI data centers with Nvidia (in return for Nvidia’s $100 billion investment into OpenAI), a multi-billion dollar chip deal with AMD (a deal that enables OpenAI to take a 10% ownership stake in AMD), and a partnership with Broadcom to develop and deploy chips that OpenAI would design.
The revenue needed for the industry to break even is becoming astronomical: a widely cited study from Bain estimates that $2 trillion in new revenue is needed by 2030 to fund the current AI scaling trend.
Last month, after burning through its entire 2026 AI budget in only four months, Uber’s COO asserted the costs of AI were getting harder to justify because they were not translating into useful customer features. And Uber is not alone: a June 2026 Bain study found that nearly 40% of companies said their cost reductions from AI were significantly less than expected. More troubling, 44% of companies based their next wave of AI investments on previous rounds of savings—savings that have consistently come in below expectations.35 As the study put it, “self-funding the next wave from past returns sounds like discipline. In reality, it is a circular bet with a structural leak.” Scaling revenue may be further challenged by delays in data center construction: recent reporting in the Financial Times found that 40% of data centers planned for 2026 are delayed, further increasing financial risk at the level of individual firm bets.
An AI bubble burst could wipe out over $20 trillion in American household wealth, three trillion more than the financial crisis.
Private credit companies deploy capital that comes from 401(k) accounts, life insurance plans, and pensions; they’ve made a casino of American workers’ financial security. For example, both New York and Pennsylvania’s state pension plans are invested in Blue Owl’s $7 billion digital infrastructure fund, which in turn has loaned out money to finance data centers for Meta in Richland Parish, Louisiana. If Meta fails to post revenues that justify its planned $125 billion in spending for 2026—a move that led to a 6% decline in its stock attributed to investor anxiety—the effects will be felt across the country.
The subsidies that AI firms have received are extraordinary, ranging from federal backing of a $1 billion loan to bring the Three Mile Island nuclear plant back online to power Microsoft’s A.I. data centers, to offering up $1 billion in AI funding through the Big Beautiful Bill, to allocating federal lands for data center construction, outlining new ‘private public partnerships’ at national laboratories that house treasure troves of genomic data that can be leveraged for commercial use, to its Export AI initiative, which leverages the apparatus of the federal government in support of deal-brokering on behalf of AI firms. Looking at the evidence, a bailout of the AI sector has arguably already begun, before Congress has meaningfully acted to protect the public and the economy from the risks introduced by the industry.