nakedcapitalism.com
Two Reasons Why the AI Industry Is Calling for Potential Nationalization
_This is Naked Capitalism fundraising week. 1009 donors have already invested in our efforts to combat corruption and predatory conduct, particularly in the financial realm. Please join us and participate via ourdonation page, which shows how to give via check, credit card, debit card, PayPal, Clover, or Wise. Read about why we’re doing this fundraiser, what we’ve accomplished in the last year, and our current goal: funding bonuses for our terrific writers._
There are two possible reasons why the AI industry is beginning to prepare the ground for a potential nationalization. The first is to socialize the losses created by the AI financial bubble. The second is to merge AI companies with the state and become an official part of the power structure.
The recent AI doomer scenarios pushed by leading AI executives and amplified by the media are very suspicious. As Nat has been covering, it sounds very much like an orchestrated effort by frontier AI labs to justify anti-competitive regulatory capture and potential bailouts. Without the part about the bailouts, that is also the opinion of Yanis Varoufakis.
In a recent conversation with Wolfgang Münchau on their podcast _Econoclast_ , Varoufakis described Dario Amodei and Sam Altman’s pleas for government regulation as a “cynical, dreadful fraud”. By hyping bogus “10% extinction” scenarios, frontier labs attempt to establish cartels and use antitrust controls against smaller startups while escaping direct corporate liability.
This would also be the motivation behind the recent AI Accord on Superintelligence signed at the White House. As Nat has previously covered, it is a cynical, voluntary public-relations exercise designed to achieve regulatory capture, preempt real legal liability, and protect the incumbent AI labs, rather than provide genuine oversight.
There are two dynamics here. The first is media-created hype about potential doomsday scenarios regarding AI. The second is a public relations exercise to accept government supervision, which in practical terms will be basically negligible, but will establish regulation barring other companies, especially foreign ones, from entering the market and hence create a monopoly.
This could be read as the AI industry creating hype about why their technology is so important—so much that it could kill us all—and pretending to accept government regulation so that when the bubble bursts, they are deemed “too big to fail” and rescued with public funds. This scenario seems plausible and ruthlessly clever.
The financial woes hanging over these companies are well known to readers of this blog. Once again, Nat’s excellent coverage has shown that financial analysts and informed commentators are warning that excessive debt, unbuilt capacity, and speculative narratives are approaching a wall of economic reality.
For example, a leaked draft of Anthropic’s IPO prospectus shows that while 2025 revenue surged over 1,000% to $4.6 billion, the company posted an $8.06 billion operating loss and a staggering $42 billion net loss (partly attributed to liability write-downs). Anthropic faces over $518 billion in future cloud and compute commitments and relies heavily on just two customers for nearly 25% of its revenue.
In another example, while reports cite OpenAI’s revenue run rate nearing $70 billion, audited 2025 figures show it lost $38.5 billion on $13.07 billion in revenue. The company is seeking a $30 billion bridge round at a $1.4 trillion valuation after pushing its IPO to 2027.
In this scenario, what do we make of Alex Karp’s statement: “The only way to deal with this kind of liability is to go to the government and say, ‘Nationalize us, please’”?
Karp can say it because Palantir has been basically a CIA protégé since its conception, and is so enmeshed in the surveillance and defense industry in the U.S. that it’s difficult to see where one ends and the other begins. However, what he’s saying here is different. He says that the only way to protect frontier models from intellectual property theft allegations and uncapped legal liabilities is to eventually seek or require nationalization.
That is in line with his vision, duly explained in his book _The Technological Republic_. He proposes that rather than viewing the state as an adversary, a regulatory nuisance, or a passive client, technology companies have an existential, patriotic duty to serve the national interest.
One of his key proposals is a return to the “Cold War Alliance” (The New Manhattan Project). Karp advocates returning to the civic-industrial paradigm of the mid-20th century, when top scientists, engineers, and government officials worked in tight alignment to counter “totalitarian threats”.
He likens the current artificial intelligence race to the Manhattan Project, arguing that the survival of liberal democracies depends on whether the West or its adversaries achieve and deploy algorithmic superiority.
Both OpenAI and Anthropic founders have made similar claims. Dario Amodei’s vision has gone from a utopian scenario where a nation could have thousands of Nobel Prize-level agents working for it to one where AI could possibly destroy life as we know it. Sam Altman has followed a similar trajectory. It is also possible that both speak of different possible scenarios depending on the audience and their interests, which seems more plausible.
This has led them to argue that the U.S. should treat AI as a new Manhattan Project, echoing Alex Karp’s vision in his book, which was published in February 2025, before alarms of a financial bubble caused by the AI industry began to be widely heard and before doomsday scenarios were in everyday circulation. Is this a coincidence?
The tech industry, the “Magnificent Seven”, and especially the AI industry, function as a cartel. They compete with each other, but they lobby and push for regulation that benefits them. Furthermore, because of their dominion over hardware and software, they have become a new kind of industrial elite.
I have argued before that this elite know each other, talk to each other, share common spaces, and watch and read similar things (accounting for personal peculiarities). This became more evident during the trial between Elon Musk and Sam Altman and with the “uncovering” of Peter Thiel’s “secret society”. And this new elite is claiming its place in the upper echelon of systemic power. That’s one of the reasons why J.D. Vance, who was a Peter Thiel employee, is where he is today.
They have a similar agenda, though each specializes in different fields. One of their core understandings is that “democracy is no longer compatible with freedom”, as Thiel wrote in 2009 (whose freedom, one may ask?). Thiel, during his time at university, founded a magazine called _The Stanford Review_ with the support and money of Irving Kristol, considered one of the fathers of the neocons.
The technology developed by Thiel, the rest of the PayPal Mafia (including Elon Musk), and, through their influence, many other technology companies, follows an agenda towards technocratic government, massive surveillance, total financial control, and automated warfare. I’m not saying that every Silicon Valley founder is behind this nor that the technology they develop is only useful for that. What I’m saying is that as a group and an elite, those who are part of it have similar ideas that revolve around those themes.
And what does this have to do with nationalization? I think that it can be read in two ways. The first is fencing their companies from the storm looming around the AI and tech industry and potentially receiving a bailout. The second is taking another step toward integrating their technology into the fabric of the state—not only as a tool the state uses, but as part and parcel of the state, solidifying their position in the upper echelons of systemic power and preventing a take over that would push them out (and which fouders of OpenAI feared). I think both can be true at the same time.
This entry was posted in AI, Politics on October 1, 2026 by Curro Jimenez.
### Post navigation
← The Fun of Curating for Naked Capitalism
##
Subscribe to Post Comments
19 comments
1. **DJG, Reality Czar** October 1, 2026 at 2:31 pm
Well, this sentence is clarifying: “Anthropic faces over $518 billion in future cloud and compute commitments and relies heavily on just two customers for nearly 25% of its revenue.”
Oh. Anthropic is teetering. Something is wrong with management and the product as it is being portrayed.
Notes:
–Alex Karp Is the Hillary Clinton of Silicon Valley. He’s a useless appendage, like her, wrong about everything. The term “simulated intelligence” applies to both of them.
–Simulated intelligence is big tabulation. It has no consciousness. It is a software problem.
–Those “existential” (damn, how I have come to hate that word) break-ins to other software, those escapes from the sandbox, recall Microsoft Windows and its endless glitches and crashes. It Is a software issue, related to quality control, not a Hallmark rom-com about mad scientists at a country inn battling the apocalypse.
Let them fail.
Reply ↓
2. **Deschain** October 1, 2026 at 2:56 pm
Yeah this isn’t about a government takeover of AI, it’s about an AI take-under of government.
Reply ↓
3. **Ignacio** October 1, 2026 at 3:11 pm
The nationalization of US AI would trigger 1) worldwide intensified fear about US-nationalized projects as imperial instruments of domination 2) the acceleration of other existing AI projects all around the world, some of them public, others private, and 3) national regulations of all kinds.
Reply ↓
1. **Ignacio** October 1, 2026 at 3:31 pm
Forgot to add hypothetical before nationalization.
Reply ↓
2. **hazelbee** October 1, 2026 at 3:53 pm
there will be plenty in europe that would be horrified by this and want to switch to local alternatives. Mistral perhaps. or locally hosted open weight models.
us gov and AI combined? with the push to put AI into all workflows? no no no no no. thats a huge data capture by the US.
Reply ↓
4. **Henry Moon Pie** October 1, 2026 at 3:44 pm
I don’t doubt that these lunatics will try to grab money and energy anywhere they can get it, but the lack of a viable business plan to make back the extraordinary sums of money being spent is not dispositive. AI agents are force mulitpliers. They can hack the previously unhackable. They can coordinate everything from denial-of-service attacks to drones more effectively than has been possible heretofore. They can sort through masses of data collected by the ubiquitous cameras faster than has ever been possible. They can flood information streams with their slop misinformation and propaganda more effectively than an army of hasbara agents. For many, they are becoming an Oracle of truth.
These guys are not out to make a profit. They’re out to control the world, to shape the future toward their twisted conceptions, and they will not surrender their control to either the government or some collective of frontier labs. I would expect that the first one to achieve the model power he thinks is sufficient to achieve his goals will quickly set his sights on the elimination of his competitors.
So I’m setting down my marker. Listen to Klein’s interview of Jensen Huang. Huang knows what the situation is with respect to carbon in the atmosphere, yet he’s advocating a “burn everything” policy for the sake of AI advancement. That will put the last nails in the coffin for our children’s and grandchildren’s future. It will kill tens of millions if not billions within a decade or two. Sure Huang wants to sell more chips, but there’ more to it than that. These people are sacrificing human civilization, such as it is, for the sake of their bizarre dreams, which are nightmares for us and our progeny.
Reply ↓
5. **flora** October 1, 2026 at 3:46 pm
I’ll go with reason number 1: They’re broke, can’t pay their enormous debts, and want a bailout disguised as a public service.
Reason number 2 is good, but since they’re already part of govt, aka the military, it’s a less compelling reason imo.
My reason number 3: if they’re nationalized they can’t be sued by states for any damages caused and can’t be regulated by the states. They might even escape the “regulated public utility” oversight run by state utility boards.
Reply ↓
1. **flora** October 1, 2026 at 3:54 pm
Listen to this clip of Bill Gates on the Ezra Klein show.
It sounds like he’s really angling for nationalization without trying to sound like he’s angling for nationalization. To keep us ‘safe’. Because he cares about humanity. Sure he does. (And nationalizing the industry will make a whole lot of his problems go away.) But it won’t stop data centers buildouts. It would probably increase them, with no citizen recourse.
https://www.youtube.com/shorts/ubtWvzfZusA
Reply ↓
1. **flora** October 1, 2026 at 4:07 pm
ooohhh, there’s that word ‘cyberattack’ again. Bill has been saying for a while we’ll have a massive cyberattack soon. I remember him predicting a pandemic was right around the corner. And he was right. Is he psychic or something? / ;)
All of his predictions are couched in dire fear terms. Be afraid, etc.
He fought govt regulation hammer and tong when he was head of Microsoft. I do not think this leopard has changed his spots.
Reply ↓
2. **Henry Moon Pie** October 1, 2026 at 7:27 pm
flora, there are times that maxxing out on cynicism can leave you quite vulnerable. If I remember correctly, you were very skeptical/cynical about the threat posed by Covid as well as by our breaching 7 of 9 planetary boundaries, so this approach to the potential threat posed by AI is not surprising. I viewed the entire interview, including, and perhaps especially, the clip you linked, and I don’t see a clever Machiavellian spiel to pump Gates’s position. The guy has, heretofore, had a stunning contempt for Nature with his Frankenplants and robot bees, and his Epstein “participation” is horrifying. And it was only recently that he was, like Huang, advocating abandoning climate goals for the sake of AI development, but he gives me some evidence in this interview that he is not as nihilistic as Huang or Musk. Listen to him describe the current, specific threats that led him to give this interview.
“Damn the torpedoes, full speed ahead, is not a strategy for dealing with pandemics, rendering the biosphere inhospitable toward human civilization, or the threat posed by powerful machines whose inner workings are a mystery to us. Even someone like Bill Gates, who has done a lot of damage, may have limits on how far he’s willing to foster catastrophe. Maybe his vision for humanity, while anti-Nature, is not completely anti-human. Whatever it is, I don’t hear him talking his book. I hear genuine agitation, maybe even fear, in his voice, and the guy does have some idea of what he’s talking about.
Reply ↓
3. **The Rev Kev** October 1, 2026 at 7:32 pm
_‘with no citizen recourse’_
Because those data centers would be a matter of “national security” so their build-out over rides any other concern.
Reply ↓
2. **Judith** October 1, 2026 at 4:13 pm
Gary Marcus interviews Zephyr Teachout, who begins the interview by saying : “I joined the current AI debate because I saw way too many people treating repeat evidence of lawbreaking as wholly unreachable by current law.”
https://garymarcus.substack.com/p/can-companies-like-openai-keep-getting
Reply ↓
1. **flora** October 1, 2026 at 4:29 pm
Thanks for this link. I have seen the same thing. e.g. copyright doesn’t apply to digital text, only to text on paper copied to paper, etc. That was the rational. The tool modifies or determines the law instead of the law determining the legal use of the tool. etc.
Reply ↓
2. **flora** October 1, 2026 at 5:45 pm
Great read. I nominate this link for tomorrow’s NC links.
I think several lawyers read this site. This info could give them some good ideas.
Reply ↓
3. **ChrisFromGA** October 1, 2026 at 5:48 pm
She gets it … if you or I build a dangerous machine, say one that has hazardous materials, and a propensity to explode, and we leave it near a school, we don’t get to say “”oops, it broke out of its containment and killed people, sorry …”
We end up in a federal penitentiary if we’re lucky … Gitmo if we’re unlucky.
Reply ↓
6. **voislav** October 1, 2026 at 5:41 pm
I feel this is about competition, For these companies to justify trillion dollar valuations they have to capture a sizable portion of the US GDP. The biggest problem for them is not each other, it is Chinese and open source models that run at a fraction of the cost and compute.
So these have to be banned, creating a walled garden where their profits are secure. Nationalization is a vehicle to pull those profits out, these valuations are so high that there is no market mechanism to sell the shares without crashing the price. But if the government just writes a check there is no market to react to the selloff.
Reply ↓
7. **ChrisFromGA** October 1, 2026 at 6:03 pm
Don’t forget that because Congress wet the bed and failed to pass any legislation to regulate AI, the States are free to regulate as strictly as they want to without being preempted by Federal Supremacy.
This may be motivating some of the hysteria emanating out of the likes of Altman, Anthropic, and Gates. They want the Federal Leviathan to serve their interests by preempting state laws.
Fortunately, Congress is so wantonly corrupt and inept, they cannot even pass legislation anymore, only trade stocks and have affairs with their staff (see, Boebert, Lauren, and Sinema, Kyrsten.)
And let’s not forget about the vegetable in the Senate being wheeled around like “Weekend at Bernies”!
Reply ↓
8. **John Merryman** October 1, 2026 at 8:19 pm
So we have a government that is bordering on bankruptcy coming to the rescue of a speculative bubble?
What part of the West isn’t circling the drain?
Reply ↓
9. **Hank Linderman** October 1, 2026 at 8:43 pm
“By hyping bogus “10% extinction” scenarios…”
Bogus? How so?
https://www.facebook.com/reel/2538854556526162
Sorry for the FB link but the discussion is concerning if only half of it is true.
Best…H
Reply ↓
### Leave a Reply Cancel reply
Your email address will not be published. Required fields are marked *
Comment *
Name *
Email *