The $1 Trillion Question: Who Would Bail Out OpenAI?
The company says it doesn’t want federal guarantees. Last week’s letter to the White House appears to say otherwise.
Is OpenAI, feeling a sense of vertigo after agreeing to take on more than $1 trillion in financial obligations for its massive compute build out, angling for the plush cushion of a federal bailout should the bubble ever burst?
That was the topic of a lot of discussion last week, after the company’s CFO Sarah Friar told The Wall Street Journal, in a live appearance that was recorded on video, that the company was looking for a financial “backstop” from the federal government, among other institutions, that would allow its financing to flow more freely.
The backstop would make sense because, Friar implied, because AI is “almost a national strategic asset.” Cue the requisite mention of competition with China.
Altman himself has made similar remarks, telling economist Tyler Cowen last month that, “When something gets sufficiently huge ... the federal government is kind of the insurer of last resort, as we’ve seen in various financial crises … Given the magnitude of what I expect AI’s economic impact to look like, I do think the government ends up as the insurer of last resort.”
OpenAI’s finances are a bit opaque, but it’s clear that they’re the leaders of the big bet everyone has collectively laid down on A.I. During a chat with Altman last week, OpenAI investor Brad Gerstner cited figures that the company had some $1.4 trillion in financial obligations it had taken or was planning to take on, in the face of just $13 billion in revenue. Altman pushed back, and appeared a bit peeved, even.
“Brad, if you want to sell your shares, I’ll find you a buyer,” Altman said, before acknowledging that “we might screw it up.”
With the Great Recession and notorious federal bank bailout in 2008, which has influenced — and helped distort — politics since, still a sore wound, the company’s insinuations about federal financial backing drew instant anger from around the country.
Even Republicans, who have in many ways been savvier than Democrats in channeling populist anger over inequality, jumped on board, with people like Florida Gov. Ron DeSantis and Trump confidant Steve Bannon lambasting the idea.
“No,” DeSantis wrote on X. “Privatized profits and socialized losses. The enduring quest of being Too Big to Fail.”
Gary Marcus, a neuroscientist and A.I. founder who is critical of the industry, underscored this sense of righteous indignation on his Substack.
“If you thought the 2008 bank bailout was bad, wait til you see the 2026 AI bailout,” he wrote. “Tell your congress person — today — that you don’t want your taxes used to bail out overhyping and economically shaky AI companies that spend far more than they earn. Workers, already feeling the knife from layoffs, should not be footing the bill.”
One writer, Sinéad O’Sullivan noted that OpenAI’s valuation alone would account for two-thirds of the entire 2008 bailout. And that would just be the “tip of the iceberg” should the market implode.
After less than a day of this news cycle, OpenAI pivoted. CEO Sam Altman “clarified,” — or walked away from — the remarks with a 15-paragraph response on X.
“We do not have or want government guarantees for OpenAI datacenters,” he wrote.
But it didn’t take long for critics to call out these claims. Marcus and others like More Perfect Union published a smoking gun, of sorts — a filing that showed that OpenAI had asked the White House, the Office of Science and Technology Policy to be exact, to consider federal loan guarantees in an 11-page letter written just a week before.
The document called for the White House to “Counter the PRC [China] by de-risking US manufacturing expansion.”
Note the jargon! What is de-risking, you ask?
“To provide manufacturers with the certainty and capital they need to scale production quickly, the federal government should also deploy grants, cost-sharing agreements, loans, or loan guarantees to expand industrial base capacity and resilience,” the letter read.
Because China. Which has now become a mantra for OpenAI and a cudgel that the company clearly believes is effective in Washington. Marcus and others called out what they saw as Altman’s dishonesty, an issue which increasingly trails the CEO, reputation-wise. “Even for him, his latest whopper is something,” Marcus wrote.
“You can see a lot of Sam in his child [ChatGPT],” one commenter noted on the blog. “It’s kind of hard to tell where it ends and he begins. They’ll both just say anything confidently with some ridiculous objective, and backpedal as soon as you ask for clarity.”
Writer Emile Torres went further, extensively tracing what he called Altman’s history of disingenuousness.
“Sam Altman is a profoundly duplicitous person who’s willing to say whatever is necessary to acquire increasingly obscene amounts of wealth and power,” Torres wrote.
Some of these allegations come up in Karen Hao’s Empire of AI, as well.
Here’s what else we’re checking out this week:
Great interview with Michael Lewis on the Morning Brew Daily podcast, in which he looks back over his career and compares the housing bubble in 2008 and ensuing financial crisis that prompted his book, “The Big Short,” to the AI boom today.
Home prices are up more than 56 percent since 2020. And the Trump administration has an idea to increase mortgage terms to 50 years, which would require legislation. But even the WSJ’s Editorial Board is not down with the proposal, noting that among other issues, it will increase demand for housing, which would not be conducive to lowering prices. It will however, benefit any friends in the home-building and mortgage-banking industries, they note.
Am I the only one who felt like the GQ interview with Sydney Sweeney was kinda lame? Or at least the magazine’s decision to clip its spiciest moment for social media. Aren’t there bigger fish to fry? Yes, culture is important. But drilling down on these controversies only gives grist to outrage merchants online and imo, seems to either be a cause or effect of the left not being able to build an identity around more than semiotics, vibes and micro-aggresions. As we noted previously here, a study found that outrage around Sweeney’s American Eagle “good genes” ad was largely a “conservative media creation,” with right-wing media accounts using a mostly inflated bogeyman of liberal overreaction about the ad to drum up real and widespread anger online. This is how the content cycle works.
404 Media takes us in a “downward spiral down the AI slop toilet” via AI-made fake deportation videos, shared into the information wastefields on Facebook, and generating confused, credulous, and chaotic responses. “A YouTube page linked from the Facebook account shows a real video uploaded of a car in Cyprus nearly two years ago before any other content was uploaded, suggesting that the person behind the account may live in Cyprus…” Fun times.
Investor and Marshall Project co-founder Neil Barsky has a message to fellow one-percenters in New York after Zohran Mamdani’s victory: “The menace [to New York] does not come from Mr. Mamdani or from Wall Street,” he wrote in the NYT. “It comes from the president of the United States, who is hellbent on inflicting real and lasting damage to the city that made him famous.” Still, he’s mad that Mamdani said billionaires shouldn’t exist.
Aaron Zamost, a former head of comms at Square, writes about how the social media and search giants look a lot like the financial services industry. “Google, Apple and their peers now act like the self-preservation-obsessed incumbents they once disrupted,” he says. “They move slower, talk safer and patrol the moat. They’ve traded risk for complacency — too afraid of offending the president, losing access or inviting a subpoena. Big Tech now serves power before it serves its users.”
Outkast’s Rock and Rock Roll Hall of Fame Induction. Speeches here and performance here.
See you all next week.