Altman, Musk and Amodei align on calls to slow AI development
Anthropic CEO's warning about loss of human control wins backing from rivals who have historically competed fiercely on model speed
Briefing
- The leaders of OpenAI, Anthropic and xAI have publicly aligned on slowing frontier AI model development, a rare show of unity among direct competitors that could accelerate regulatory and legislative pressure on the sector.
- Anthropic CEO Dario Amodei published a call titled 'We Must Pace the Frontier', arguing the industry is moving faster than safety guardrails can accommodate.
- Sam Altman and Elon Musk, whose companies compete directly with Anthropic, both endorsed Amodei's position despite longstanding commercial and personal rivalries.
- A former Anthropic researcher told the BBC that AI staff are 'genuinely frightened' for humanity's future, adding insider weight to what might otherwise be dismissed as executive posturing.
- If the alignment among lab chiefs translates into a coordinated industry proposal, chip demand assumptions and deployment timelines across the sector could face revision.
Analysis
- AI infrastructure capex assumptions face downside revision if the Altman-Amodei-Musk alignment produces a coordinated industry proposal for voluntary slowdowns. Chip demand forecasts for Nvidia, TSMC, and hyperscaler buildout timelines are the most direct casualty: if frontier model training cycles lengthen or pause, GPU reservation queues thin and cloud AI capacity additions slow. This directly contradicts the demand confirmation Oracle's 121% cloud infrastructure growth just provided, creating a conflicting signal for AI infrastructure equities.
- The OpenAI IPO delay to 2027, now explicitly linked by Altman to safety concerns, gains a harder policy rationale as all three major lab CEOs publicly align on slowing development. This makes a quick reversal of the delay implausible and removes the most anticipated AI liquidity event from the 2026 calendar, prolonging valuation uncertainty for secondary market OpenAI holders and late-stage AI venture funds that marked positions to an anticipated 2026 listing.
Historical Context
- 2024-2025
The EU AI Act's tiered risk framework was shaped in part by voluntary industry input from frontier labs, establishing the precedent that coordinated lab positions influence binding regulatory text. A joint proposal from OpenAI, Anthropic, and xAI would carry substantially more weight than individual submissions in any equivalent US legislative process.
- 2023
The March 2023 open letter calling for a six-month pause in AI training, signed by Musk among others, generated substantial press coverage but produced no binding commitments and was followed by accelerated model releases. The current alignment is structurally similar but now includes active lab CEOs rather than external signatories, giving it more credible policy traction.
- 2021-2022
The semiconductor shortage demonstrated how quickly AI and consumer electronics capex assumptions can misfire when demand signals shift. Nvidia's data center order book in 2022 collapsed before recovering sharply; a voluntary AI training slowdown could replicate the demand air pocket that preceded that correction.
Connected Events
OpenAI IPO will not happen in 2026, Sam Altman confirms
Altman explicitly cited safety concerns when confirming OpenAI will not IPO in 2026, and this briefing shows that same safety framing now has public buy-in from Musk and Amodei, making the IPO delay rationale structurally harder to unwind.
Oracle Q1 revenue grows 30% as AI cloud infrastructure demand surges
Oracle's 121% cloud infrastructure growth and Larry Ellison's cancelled stock sale both reflect a market pricing in sustained AI training demand. A coordinated lab slowdown directly contradicts that demand thesis and sets up a fundamental conflict with the Oracle-driven AI infrastructure optimism from this week.
Further reading
See Indexa more often on Google
Mark Indexa as a preferred source — your Top Stories will surface more Indexa coverage.