Solana co-founder Anatoly Yakovenko issued a terse public response on September 13 questioning the financial drivers behind recent proposals to slow advances in frontier artificial intelligence. His four-word comment targeted statements from Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and xAI founder Elon Musk supporting a more measured approach to model capability growth.

Yakovenko posted “Profitability at $1 trillion mcap” in reply to the emerging discussion. He offered no supporting calculations, named no specific company, and provided no revenue or valuation data. The remark framed the slowdown advocacy as potentially tied to commercial considerations once firms approach or reach that scale, though it remained a personal observation without evidence of coordinated financial strategy among the AI laboratories.

Amodei’s Proposal for Paced Frontier Progress

Amodei outlined his position in an essay arguing that developers should reduce the speed at which they improve AI model capabilities. He distinguished this pacing from any halt to training or research. Progress could continue at a substantial rate, he stated, while teams devote additional time to testing safeguards, studying behavior, and enabling external review.

The plan advances in stages. First, frontier labs would grant third-party evaluators ongoing access comparable to internal risk-assessment staff. Amodei cited the Model Evaluation and Threat Research organization as one possible participant. Reviewers would receive company resources and permissions subject to legal and confidentiality limits, with authority to publish findings under restricted redaction rules covering privileged, security-sensitive, or commercial information.

A second stage seeks coordination among frontier AI companies in democratic countries on shared safety standards and limits to unchecked capability expansion. Amodei acknowledged potential antitrust complications and the possible need for limited government authorization. The third stage envisions international arrangements, including discussions with China, ranging from controls on AI-assisted biological risks to eventual constraints on rapid automated improvement. He described a comprehensive pause as least likely due to verification challenges and stressed that any framework should maintain the technological position of the United States and its allies.

Support from Altman and Musk

Altman endorsed the core idea of pacing the frontier. He specifically committed OpenAI to independent evaluators holding employee-like access and said the company would release further details soon. No evaluator names, access terms, or implementation timeline were provided.

Musk offered a shorter affirmation, stating that Amodei is right. Available reports indicate he reposted the essay but did not specify which elements of the three-stage framework xAI would adopt. None of the three executives addressed or endorsed Yakovenko’s profitability characterization.

Counterarguments on Voluntary Action and Competition

Former White House AI and cryptocurrency adviser David Sacks challenged the necessity of coordinated industry or regulatory measures. Addressing Amodei and Altman directly, he noted that the two firms occupy a leading position in frontier intelligence based on model performance, revenue, and market presence. Sacks argued they could slow development voluntarily if unreleased systems posed serious risks, without requiring competitors, antitrust exemptions, or external evaluators with authority over rival labs.

He raised the possibility that calls for broader rules could amount to regulatory capture and limit open-source efforts and faster-moving startups. Sacks also questioned whether laboratories in China would adhere to any global arrangement, potentially disadvantaging compliant U.S. companies. He pointed to existing market discipline through product liability and customer rejection of unreliable systems as alternative incentives for caution.

Neither Anthropic nor OpenAI has publicly accepted descriptions of a duopoly or confirmed liability concerns as drivers of their positions. Amodei’s essay itself recognizes verification difficulties in international settings while still favoring negotiations on narrowly defined risks.

Context Around Valuations and Industry Position

Reports have noted preparation by Anthropic and OpenAI for potential public offerings, alongside rising infrastructure and energy costs that increase pressure for operational efficiency. Because the companies remain privately held, any market-capitalization figures represent estimates rather than disclosed valuations. Yakovenko did not clarify whether his $1 trillion reference applied to one firm, another, or the sector overall. No statement from Solana Labs or the Solana Foundation connected the blockchain network or its token to the policy debate.

Outside evaluations represent the most immediate concrete step. Anthropic has committed to embedding reviewers with substantial internal access and publication rights. OpenAI has signaled a parallel approach. Broader legislative ideas, including proposals for temporary suspensions of advanced work or permanent limits on artificial superintelligence, remain under discussion among U.S. lawmakers but have not advanced to formal introduction in the form summarized by sponsors.

The exchange leaves open questions about the practical timeline for evaluator programs, the scope of any industry coordination, and the feasibility of international participation. Yakovenko’s intervention added a skeptical commercial lens to a conversation centered on safety and technical risk management.