Solana Founder Trolls AI Slowdown Push With Trillion-Dollar Market Cap Jab

Anatoly Yakovenko has four words for the AI safety debate: profitability at a trillion-dollar market cap

When three of the most powerful figures in artificial intelligence lined up behind a proposal to slow frontier model development this weekend, the loudest skeptical voice came not from Washington or Brussels but from the crypto industry. Solana co-founder Anatoly Yakovenko responded to the AI slowdown push with a four-word post on X: “Profitability at 1 trillion USD mcap.”

The remark, published Sept. 13, was framed as a sarcastic observation rather than an accusation, and Yakovenko provided no revenue, valuation or profitability calculations to back it. But it distilled a suspicion spreading through tech circles: that calls for restraint from companies allegedly preparing initial public offerings may be inseparable from their financial interests.

What the slowdown proposal actually says

The debate was started by Anthropic CEO Dario Amodei, who published an essay titled “We Must Pace the Frontier” calling for slower advances in frontier AI capabilities. Amodei was careful to distinguish pacing from halting: “We must slow the pace at which we improve the capabilities of AI models,” he wrote, arguing that progress can remain fast while developers spend more time testing safeguards, studying model behavior and allowing outside evaluators to examine their work.

The proposal unfolds in three stages. First, Anthropic plans to give third-party evaluators ongoing access comparable to that of internal risk-assessment staff, naming the Model Evaluation and Threat Research organization, METR, as an example. Second, frontier AI companies in democratic countries would coordinate on common safety requirements and limits on unchecked capability growth — coordination Amodei concedes could raise antitrust issues and might require narrow government authorization. Third, and most ambitiously, international coordination including possible agreements with China, ranging from restrictions on AI-assisted biological weapons to eventual limits on rapid, automated model improvement.

Amodei himself describes a comprehensive pause as the least likely form of international agreement because verification would be so difficult, and he argues any arrangement must preserve the technological position of the United States and its allies.

Altman and Musk pile on

OpenAI CEO Sam Altman endorsed the proposal directly in a Sept. 12 post. “I agree with Dario that we need to pace the frontier,” Altman wrote, specifically embracing the embedded-evaluator component: “Committing to having independent evaluators with employee-like access is a great idea, and we will do the same.” OpenAI promised more information but supplied no implementation date, evaluator name or access terms.

Elon Musk responded more briefly, reposting Amodei’s essay with the words “Dario is right,” according to Reuters and the Financial Times. His endorsement did not specify which parts of the three-stage plan xAI would adopt. Reuters also reported that both Anthropic and OpenAI are preparing for potential IPOs — context that gives Yakovenko’s jab its sting, even though the report established no link between either company’s valuation goals and its safety posture.

Notably, none of the three executives endorsed Yakovenko’s suggested connection between pacing and trillion-dollar valuations. His line remains his own characterization of their motives, and no named company has confirmed that its support for slower development depends on reaching any particular valuation. Because the companies remain private, references to market capitalization can only describe estimates from private transactions or proposed listings.

Sacks counters: slow down alone

The sharpest rebuttal to the coordination plan came from David Sacks, the venture capitalist and former crypto policy adviser, who argued that Anthropic and OpenAI could simply slow their own development without dragging competitors or lawmakers along. “You guys are the frontier,” Sacks wrote, addressing Amodei and Altman directly, describing the two companies as holding a “duopoly on frontier intelligence.”

Sacks said he would support a voluntary slowdown if unreleased systems presented risks serious enough to concern their own developers, but he disputed the need for an antitrust exemption, a regulatory approval system or evaluators with authority over competing laboratories. He went further, accusing the companies of pursuing possible “regulatory capture” — while presenting no evidence of any agreement between Anthropic and OpenAI to exclude competitors.

He also raised product liability as an alternative explanation for corporate caution, arguing that customers already punish models that behave in unauthorized or unreliable ways, and he questioned whether China would participate in any global AI agreement at all — a verification problem Amodei’s own essay acknowledges.

Why a blockchain founder is in this fight

Yakovenko’s intervention highlights how the AI safety debate now cuts across industry lines. Crypto founders have watched regulators apply scrutiny to tokens and decentralized protocols while frontier AI labs — whose systems have been implicated in cyberattacks, surveillance and social engineering incidents — largely self-regulate. The optics of the industry’s largest labs coordinating on pace, with potential antitrust cover, invites exactly the kind of incentive-based skepticism the Solana co-founder delivered.

No Solana Foundation statement has connected the network or its SOL token to the AI policy dispute, and Yakovenko’s post represents personal commentary rather than a disclosed financial projection from any company involved.

What happens next

The concrete next step is outside evaluation. Anthropic has committed to inviting an external review team into its internal safety process, with reviewers receiving company laptops, office access and permissions similar to internal risk staff, subject to legal and confidentiality restrictions. Those reviewers would be allowed to publish findings without Anthropic controlling their conclusions, with limited redactions for legally privileged, security-sensitive or commercially sensitive material. No evaluator contract or start date has been announced, and OpenAI has promised a similar system of its own.

Whether the trillion-dollar jab sticks depends on what those evaluations find — and on whether the companies calling for a slower frontier can convince the market that safety, not IPO math, is doing the talking.