OpenAI’s $1.4 trillion AI plan raises ‘too big to fail’ concerns

OpenAI CFO Sarah Friar sparked controversy last week by stating that the company is seeking a “backstop” from the government for loans to help with its planned $1.4 trillion in spending on data centers and other infrastructure investments over the next eight years.

Speaking on stage at The Wall Street Journal‘s “Tech Live” event, Friar, whose comments caused the controversy, said OpenAI aims to create an ecosystem of banks, private equity, and a federal “backstop” or “guarantee” that could help finance the company’s investments in cutting-edge chips.

Following criticism, Friar later clarified her remarks, arguing that her company is not seeking government support for its AI infrastructure commitments.

Softening her stance in a post published late last Wednesday on LinkedIn, Friar said, “I used the word ‘backstop,’ and that muddied the waters. As the full context of my answer shows, the strength of American technology will come from building a true industrial capacity, which requires the private sector and government to do their part.”

CEO Sam Altman later posted a lengthy tweet to “clarify” Friar’s statement, claiming the company is not trying to become “too big to fail,” and White House AI lead David Sacks announced that taxpayers would never have to finance a bailout for AI.

On the other hand, some argue that Friar has “opened Pandora’s box.” It is not just the “size” that worries some observers. According to an analysis in The Wall Street Journal, what emerged from the collapse of Lehman Brothers in 2008, which led to a global financial crisis, was not only that banks were “too big to fail” but also that they were too interconnected.

Now, the highly multifaceted nature of OpenAI’s core investments and the tech industry’s role in supporting an economy on the brink of recession are causing some to worry that another bubble is about to burst.

OpenAI has tied up billions of dollars in “cyclical investments” with companies like Nvidia, Microsoft, AMD, Oracle, and CoreWeave, designed to support its growth, and has not yet disclosed how it plans to finance the necessary infrastructure; furthermore, an IPO is not expected in the near future.

However, not everyone is against government support. Nvidia CEO Jensen Huang, whose company briefly surpassed a $5 trillion market capitalization last week, argued on Wednesday that receiving government subsidies to support data centers is the only way to stay ahead of China in the AI race.

But as AI giants reach out for government support, investors seem to be pulling back: shares of AI companies lost more than $820 billion in value this week, marking their worst week since President Donald Trump’s tariff announcements in April.