Capital & Industrial Strategy

Anthropic's confirmed close of a $65 billion funding round at a post-money valuation of $965 billion is the largest private AI financing in history and the first time the company has surpassed OpenAI on a valuation basis. The round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, each contributing over $2 billion according to Bloomberg. The scale of the commitments from crossover and growth-stage funds — not just strategic corporates — indicates genuine conviction in Anthropic's revenue trajectory rather than defensive positioning.

The strategic context is critical. Anthropic simultaneously confirmed that its 'Mythos' model, described as having advanced cyber capabilities, is imminent. This has already triggered EU-US diplomatic engagement, with a senior EU official telling CNBC that Brussels is seeking to 'intensify' talks with Washington over the model's governance. For investors, Mythos represents both a capability unlock for enterprise security and defence contracts and a regulatory overhang that could constrain deployment markets. Semafor notes that Anthropic's capital structure now teeters between the risk of scaling too fast and the risk of scaling too slowly — a reflection of the brutal unit economics at the frontier, where compute costs grow in step with capability.

Dell's fiscal quarter delivered AI server revenue growth of 757% year-over-year, driving a 30-32% single-day stock surge — the company's best trading day since its 2018 return to public markets. Annual sales guidance far exceeded analyst consensus, per CNBC and Reuters. Price increases on AI server configurations contributed alongside volume, suggesting Dell has pricing power in a supply-constrained environment.

This result is a significant enterprise adoption signal. Dell's customer base skews heavily toward large enterprises and government, meaning this is not hyperscaler capex — it is distributed enterprise infrastructure build-out. The data corroborates the broader thesis that AI infrastructure spend is now in a second wave: hyperscalers built first, and enterprises are now following with their own on-premise and hybrid deployments. NetApp also beat estimates on the same day, with analysts highlighting strong storage demand as another downstream indicator of enterprise AI workload growth.

Two hardware funding events this week illustrate how venture capital is now probing specific bottlenecks in the AI infrastructure stack rather than making broad compute bets. Groq is reportedly raising $650 million in an internal round as it pivots from chip manufacturing toward AI inference services, per TechCrunch. This repositioning follows Nvidia's $20 billion arrangement with Groq's talent and IP — a deal that effectively validated Groq's architectural approach while hollowing out its standalone hardware competitive position. The pivot to inference-as-a-service is a logical response: Groq retains the software and systems expertise while outsourcing the capital-intensive chip fabrication risk.

Separately, South Korean startup XCENA raised $135 million at a $570 million valuation on the thesis that memory bandwidth — not raw compute — is the binding constraint for large-scale AI inference, per TechCrunch. The Wall Street Journal frames this as part of a broader VC turn toward hardware and 'physical AI,' driven by the view that AI threatens to commoditise software but that infrastructure retains pricing power. The XCENA raise is notable geographically: South Korean deep-tech is attracting capital in memory specifically because of the country's existing semiconductor expertise at Samsung and SK Hynix.

OpenAI has held discussions with Citigroup and JPMorgan about joining its IPO banking syndicate, expanding beyond its previously reported advisors, per Bloomberg. Adding bulge-bracket distribution capacity suggests the offering is sizing up — both in terms of deal scale and the breadth of institutional and retail investor base OpenAI intends to reach. The timing pressure is real: with Anthropic now carrying a $965 billion private valuation, OpenAI's IPO pricing will face direct public comparison to a private comp that has surpassed it.

In parallel, China's MiniMax has filed regulatory paperwork for a domestic A-share or Hong Kong IPO, per Bloomberg, positioning itself against DeepSeek and other domestic frontier model providers. A Chinese frontier AI IPO would be a first-of-kind public market test for that cohort and would establish a domestic valuation benchmark that Beijing's capital markets currently lack. The two listings, if they proceed concurrently, will create an unusual moment of simultaneous public price discovery for US and Chinese frontier AI — with significant implications for how investors price the competitive dynamic between the two ecosystems.