Capital & Industrial Strategy

The Trump administration directed Anthropic to disable access to its Fable 5 and Mythos 5 models for all foreign nationals, citing a discovered jailbreak method as a national security risk. Anthropic complied but issued a pointed public dissent, stating it disagreed 'that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.' The action appears to be grounded in export control authority rather than any finding of deliberate wrongdoing by Anthropic. Confirmed sources include Wired, Financial Times, WSJ, and CNBC.

The strategic implications are significant and cut in several directions simultaneously. For Anthropic, a company whose brand is built on safety-first AI development, the episode creates a painful irony flagged by TechCrunch: its own transparency about model risks may have handed regulators the justification for an intervention it considers disproportionate. For the broader industry, this establishes that the US government will exercise commercial AI export controls in real time, not just at the chip or infrastructure layer. Any frontier model company with international users must now treat government-ordered access revocation as an operational and commercial risk to price into its business model — a factor that will bear directly on Anthropic's forthcoming IPO valuation and investor due diligence.

SpaceX raised $75 billion in the largest IPO on record, debuting on Nasdaq at a $2.1 trillion valuation with shares up nearly a fifth on the first day of trading, according to Financial Times and WSJ. The smooth execution — characterised by WSJ as a 'Goldilocks debut' — demonstrates that public market appetite exists for mega-cap technology offerings even in an environment where demand exceeds supply of quality listings. Bloomberg's Wall Street Week framed the IPO as a stress test for capital markets, with TechCrunch coining the 'MANGOS' acronym — Meta/Microsoft, Anthropic, Nvidia, Google, OpenAI, SpaceX — to describe the cohort.

The deal has direct read-across for AI capital markets. Semafor notes the boost is a positive signal for Anthropic and OpenAI's own anticipated mega-offerings. However, the government-ordered shutdown of Fable 5 and Mythos 5 complicates the timing and narrative for Anthropic's IPO. Meanwhile, a Bloomberg Opinion piece Bloomberg argues against S&P 500 inclusion exceptions for SpaceX, Anthropic, or OpenAI — a signal that institutional index-driven demand may be more constrained than headline valuations imply. Worth noting: SpaceX's decision to rent out its Colossus 1 Memphis data centre to Anthropic — after internal teams struggled to operationalise it for Grok model development — reveals an unexpected infrastructure dependency between the two companies that adds complexity to both firms' public market narratives, per Bloomberg.

Mistral is reported to be raising €3 billion at a €20 billion valuation, nearly double its Series C, according to TechCrunch. This remains unconfirmed — TechCrunch characterises it as a rumour. In parallel, Mistral CEO Arthur Mensch told CNBC in a confirmed interview CNBC that the company is focused on agentic AI and enterprise adoption. Separately, Semafor reports that an AI price war is underway as users increasingly prioritise value alongside capability.

The combination of a potential Mistral mega-round with a declared price war illustrates the structural tension facing frontier AI companies: capital requirements are increasing (more compute, more talent, larger context windows) at exactly the moment that pricing power is under pressure. Mistral's European positioning may offer some insulation — European enterprises and governments have shown willingness to pay a sovereignty premium for non-US-controlled AI infrastructure, a dynamic that could support the valuation multiple implied by the rumoured round. However, investors should treat the €20 billion figure as unconfirmed market speculation until Mistral makes a formal announcement.

A coalition of US state attorneys general has formally opened an investigation into OpenAI, requesting information across a wide range of topics, per Bloomberg and confirmed by Reuters. OpenAI says it is engaging 'constructively' with the AGs, per CNBC. The investigation's scope has not been disclosed, but a multi-state coalition implies coordinated concerns around consumer protection, data practices, or the company's ongoing for-profit restructuring — all of which would be material to IPO disclosure requirements.

At Meta, internal dysfunction in the AI division is now a confirmed public story. Wired Wired reports executives and employees are struggling with a chaotic AI strategy, while Zuckerberg acknowledged 'mistakes' in the AI workforce shift in public remarks reported by Reuters. For a company spending tens of billions annually on AI infrastructure and model development, execution incoherence at the organisational level is a direct threat to return on capital — and a signal that scale of investment alone does not translate into competitive advantage.