Theres an old saying in Washington: never believe anything until its been officially denied. And now that David Sacksdubbed the AI czar in the Trump administrationhas declared that there will be no federal bailout for AI, we can begin speculating about what that bailout might look like when it happens.

It turns out that the chief financial officer of the AI giant OpenAI has already floated an idea of what such a rescue could involve. In a recorded interview with the Wall Street Journal, Sarah Friar said the sector would need federal guarantees to enable the massive investments required for the United States to lead in AI development and deployment. Friar later clarified her remarks in a LinkedIn post after Sacks reacted, saying she had muddied the point by using the word backstop, and that she meant that AI leadership would require government involvement. That aligns more closely with what she said in the WSJ interview.

You might wonder: why would the hottest industry in the worldflush with hundreds of billions in investor fundingneed a federal rescue at all? Notably, AI expert and commentator Gary Marcus predicted ten months ago that the AI sector would seek a government bailout to compensate for overspending, poor commercial decisions, and massive future commitments it is unlikely to meet. For example, in a recent podcast hosted by one of OpenAIs outside investors, CEO Sam Altman appeared uncomfortable when asked how a company with only 13 billion dollars in annual revenue and ongoing losses could meet 1.4 trillion dollars of spending obligations over the coming years. Altman did not actually answer the question.

So what justification might the AI industry craft to secure government support, loan guarantees, grants, or other forms of assistance? For years, one of the most reliable ways to get Washingtons attention has been to say some variation of China is bad China must be beaten. And thats exactly what Altman has been telling reporters. But that does not explain why OpenAI specifically should be the recipient of federal aid over any other company.

In what appears to be an attempt to contain the fallout, Altman wrote on X that OpenAI is not seeking direct federal support, then later explained how the government could provide indirect help by building numerous public data centers that could be leased to AI firmsallowing them to avoid bearing the full capital costs themselves.

Perhaps Im mistaken, and what were seeing is not an early negotiation between the AI sector and Washington over what a bailout might look like. And lest anyone think this industry has operated so far without government help, the Associated Press notes that more than 30 U.S. states already offer incentives to attract data centers. Not everyone is pleased to have these facilities in their communities. These centers have driven up electricity costs as utilities and consumers compete with data centers for supply, while utility companies request funding to build additional capacity to power them. In effect, current electricity customers are underwriting the expansion of AI data centers by paying for new power plants and transmission lines.

The deeper problem is that AI, in its current form, appears constrained by structural limitations that will prevent it from taking over many human tasks and from being embedded in critical systems (because it makes too many mistakes). The extraordinary claims made by AI boosters are thoroughly dismantled in a lengthy critique by Ed Zitron.

I increasingly see AI as a boondogglea term that Dictionary.com defines as a wasteful or pointless project carried out for political, commercial, or personal gain. So far, the AI sector fits that definition quite well. But Ill borrow a broader interpretation from Dmitry Orlov, author of Reinventing Collapse: a modern boondoggle should not only be useless, but ideally should also create additional problems that can only be solved by new, equally pointless projectssuch as the need for massive new electricity capacity that may later prove unnecessary if AI turns out to be far less useful than advertised. The boosters say AI will have a massive impact on society. I agree completelybut not in the way they imagine.