Capital & Industrial Strategy

OpenAI announced a $122 billion fundraising round valuing the company at $852 billion, yet its shares have fallen out of favor on secondary markets where investors are finding them difficult to sell, according to Bloomberg. This disconnect between primary and secondary market sentiment suggests institutional investors are questioning whether OpenAI's growth trajectory justifies its valuation premium over competitors. Meanwhile, Salesforce Ventures and WestBridge Capital characterised AI valuations as 'punchy', and Microsoft's stock fell 23% in Q1 — its worst quarter since 2008 — partly reflecting investor concern over its OpenAI partnership and AI infrastructure spending.

The secondary market dynamics indicate that liquidity providers and later-stage investors are repricing AI risk more aggressively than venture funds participating in primary rounds. This divergence matters because secondary market pricing often presages future down-rounds or valuation pressure when companies eventually seek to raise again or exit. Reuters notes OpenAI faces a '$852 billion problem: finding focus', suggesting the company's sprawling product ambitions may be diluting its strategic clarity at a scale where execution missteps carry existential risk.

Anthropic's valuation reached $380 billion as investor demand on secondary markets shifted decisively toward the company and away from OpenAI, according to Bloomberg. The company's Claude Code product has become a 'key moneymaker', though it suffered a source code leak that Anthropic executives blamed on 'process errors' related to its rapid product release cycle. Despite issuing thousands of GitHub takedown notices — later retracted as accidental — and users hitting usage limits faster than expected, Anthropic's market position appears strengthened by its ability to ship revenue-generating products.

A senior Anthropic executive told Bloomberg that the company's general-purpose AI agent, Cowork, is expected to reach a wider market than Claude Code, signaling Anthropic's intent to expand beyond developer tooling into broader enterprise workflows. This product strategy — focusing on monetizable, domain-specific agents rather than general-purpose models alone — appears to resonate with investors seeking clearer paths to profitability. The contrast with OpenAI's sprawling ambitions is stark: Anthropic is winning by narrowing focus to products that enterprises will pay for today.

SpaceX has confidentially filed for an IPO targeting between $40 billion and $80 billion in proceeds to fund AI space data centers, according to The Wall Street Journal and Bloomberg, making it one of the largest public offerings ever. Elon Musk's strategy to deploy AI compute infrastructure in orbit represents a bet that terrestrial data center constraints — energy availability, cooling, real estate — will make space-based alternatives economically viable. However, Reuters notes that Microsoft's recent setbacks with undersea data center projects sound a cautionary note about novel infrastructure approaches.

Meanwhile, terrestrial infrastructure spending continues at scale: Meta is building a Hyperion AI data center powered by 10 new natural gas plants with capacity equivalent to South Dakota's entire power consumption, and Intel repurchased Apollo Global Management's 49% stake in its Ireland chip fab for $14.2 billion, signaling renewed manufacturing ambition. Cognichip raised $60 million to automate AI chip design, claiming 75% cost reduction and 50% timeline compression. Cisco's president told Semafor the tech world is 'grossly underestimating' AI infrastructure needs, while Morgan Stanley's M&A co-head noted energy volatility and uneven AI effects are shaping acquisition strategy.

Chinese chipmakers now claim nearly half of China's local AI chip market as Nvidia's share contracts under US export restrictions, according to Reuters. This market share gain reflects both forced localization and genuine capability development among Chinese semiconductor firms. Separately, Zhipu's shares surged 35% after its first post-IPO earnings report showed revenue doubling, demonstrating that China's AI 'tiger' companies are finding monetisation pathways despite US technology restrictions.

Alibaba released its third proprietary AI model in three days, signaling a strategic shift toward closed-source, profit-oriented AI services rather than open model development. This pivot mirrors broader Chinese AI ecosystem strategy: build domestic capabilities behind closed architectures to capture value rather than compete on open model benchmarks where US players maintain leads. The combination of chip localisation, rapid model iteration, and revenue-focused product strategy suggests China's AI market is decoupling faster than many Western investors anticipated, creating a parallel ecosystem with distinct competitive dynamics and capital allocation patterns.