Two AI Labs, valued at parity on the secondary market. Similar IPO timelines. Both under fire publicly. Both trying to win retail investors before Q4 2026. And neither able to afford the price war that comes with commoditising models. Last week, they made their moves — and picked opposite sides of the emotional spectrum.
They have more in common than their PR suggests
Last week, Anthropic overtook OpenAI on ARR. Real milestone — but not obviously a durable one. Anthropic is compute-constrained, and just struck a deal with Google and Broadcom for 3.5 gigawatts of TPU capacity starting in 2027. The revenue lead arrived before the infrastructure to support it.
Altman's position has its own problems. This week, The New Yorker published an 18-month investigation — over 100 sources, internal memos, Slack messages — titled "Sam Altman May Control Our Future — Can He Be Trusted?"
The sharpest quote is from an anonymous board member:
"He's unconstrained by truth. He has two traits that are almost never seen in the same person. The first is a strong desire to please people, to be liked in any given interaction. The second is almost a sociopathic lack of concern for the consequences that may come from deceiving someone."
Beyond the personal drama, both companies share two structural problems that no marketing campaign fully solves.
The first is consumer trust. A recent Quinnipiac University poll found that 76% of Americans trust AI-generated information either hardly ever (27%) or only some of the time (49%). The researchers' summary: "Americans are clearly adopting AI, but they are doing so with deep hesitation, not deep trust."
The second is moats — or the lack of them. Benedict Evans spent his February 2026 essay "How Will OpenAI Compete?" applying every strategy framework he knows and came up empty. Roughly half a dozen organisations are now shipping competitive frontier models with equivalent capabilities. Two years into the model wars: no fundamental moat, no meaningful barriers to entry, no network effect, no winner-takes-all.
The moves
Against that backdrop, both companies made bold plays last week — and the emotional logic of each is worth paying attention to.
Sam chose greed. OpenAI released a 13-page policy paper, "Industrial Policy for the Intelligence Age" — pitching a sweeping social contract: a public wealth fund giving every citizen a stake in AI-driven growth, portable benefits that follow workers across jobs, AI access treated like broadband, formal worker voice in AI deployment. It's a genuinely interesting document. It also landed the same day as the New Yorker piece, which made "how much do we trust the motives behind this" a harder question to shake.
Dario chose fear. Anthropic announced that Claude Mythos — its most powerful model — is too dangerous to release. The company says it used Mythos to find "thousands of zero-day vulnerabilities" across every major OS and browser. Rather than a public launch, Anthropic is giving access to 40+ organisations — Apple, JPMorgan, the Linux Foundation — to test their own systems. The message: we built something extraordinary, and we're responsible enough not to unleash it on everyone.
Whether Mythos is genuinely that dangerous is almost beside the point. For a compute-constrained company, it's a smart move — claim frontier leadership without the infrastructure cost of a full release, while doubling down on the safety story that wins enterprise deals. The Google/Broadcom TPU announcement the same week backstops the credibility.
Do we need to choose?
Honestly, both narratives make sense. A new social contract and responsible AI development aren't mutually exclusive. Institutional investors are already hedging — secondary pricing puts both at a 17-18x multiple, essentially valuing them at parity.
The real question is whether investor appetite holds for both at IPO. Both narratives are being written for that audience first. The next few months will tell us which story the market finds more compelling — or whether, as the secondary pricing already suggests, it simply wants exposure to the category and isn't picking sides yet.
Enjoyed this article?
Subscribe for private market insights delivered to your inbox.