Capital & Industrial Strategy

OpenAI completed a $122 billion funding round at an $852 billion pre-money valuation, led by Amazon, Nvidia, and SoftBank, according to Bloomberg and The Wall Street Journal. The round is the largest in Silicon Valley history and represents an increase from the initially announced $110 billion commitment. Notably, $3 billion came from retail investors through a money manager planning to include OpenAI in exchange-traded funds, marking the company's first direct access to individual capital, per TechCrunch and Financial Times.

The capital is explicitly earmarked for chips, data centres, and talent acquisition, signalling OpenAI's intent to maintain infrastructure parity with hyperscale cloud providers as compute costs rise. The retail tranche and ETF inclusion suggest preparation for a public offering, creating liquidity pathways for late-stage investors while broadening the investor base beyond institutional participants. The valuation implies a trailing revenue multiple north of 40x if OpenAI's rumoured $20 billion annual run rate is accurate, pricing in multi-year revenue growth assumptions that exceed current monetisation trajectories.

Nvidia invested $2 billion in Marvell Technology as part of a partnership allowing Marvell to integrate custom AI chips and networking equipment onto Nvidia's platform, per Bloomberg and Financial Times. The deal focuses on silicon photonics to accelerate data centre interconnect speeds. Marvell's stock surged 13% on the announcement, according to CNBC. This follows similar $2 billion investments Nvidia made in CoreWeave, Nebius, and Coherent in recent months, bringing the total strategic deployment to approximately $8 billion.

The pattern reveals Nvidia's strategic pivot from pure supplier to equity participant in its ecosystem, ensuring that custom chip designers and infrastructure providers building on its platform remain aligned and dependent on Nvidia's architecture. By taking stakes in compute providers (CoreWeave, Nebius), component suppliers (Coherent for optics), and now custom silicon partners (Marvell), Nvidia is vertically integrating without formal M&A, avoiding antitrust scrutiny while locking in distribution and design-in wins. The Marvell deal is particularly significant because it opens Nvidia's closed ecosystem to third-party accelerators, but only those Nvidia has equity influence over, effectively creating a controlled marketplace.

CoreWeave secured an $8.5 billion loan to expand AI cloud infrastructure, according to Reuters, while Nebius committed $10 billion to build an AI data centre in Finland, per Reuters. Microsoft announced over $1 billion in Thailand cloud and AI infrastructure investment over two years, per The Wall Street Journal. However, Bloomberg reports that Asian bankers financing AI infrastructure deals are expressing caution due to energy cost volatility driven by the Iran conflict, clouding $800 billion in regional data centre financing.

The capital deployment suggests infrastructure providers are building capacity ahead of confirmed enterprise demand, betting that AI workload migration will justify the capital expenditure. CoreWeave's $8.5 billion loan, likely secured against future revenue contracts with hyperscalers or corporate customers, indicates debt markets remain open for AI infrastructure despite broader credit tightening. However, the energy shock introduces a material risk: rising power costs could render recent data centre economics unviable if electricity prices remain elevated, particularly in Asia where many facilities rely on imported energy. Microsoft's Thailand investment signals big tech's geographic diversification strategy to access cheaper land and power, though geopolitical risk in Southeast Asia remains a variable.

Runway launched a $10 million fund and startup program to back companies building on its AI video models, aiming to accelerate adoption of interactive, real-time video intelligence applications, per TechCrunch. Meanwhile, Yupp, a crowdsourced AI model feedback startup that raised $33 million from a16z crypto's Chris Dixon, shut down less than a year after launch, according to TechCrunch. ThinkLabs AI raised $28 million in a round led by EIP and including Nvidia's and Edison's venture arms, per Axios.

The divergence is clear: vertical infrastructure plays with enterprise distribution (Runway's video API for developers, ThinkLabs' industrial AI applications) are attracting follow-on capital, while consumer-facing or crypto-adjacent models lacking clear monetisation (Yupp's decentralised feedback platform) are failing. Runway's $10 million fund is modest but strategically important—it's ecosystem development spending designed to create switching costs by funding startups that build on Runway's models, similar to how Stripe and Plaid funded fintech startups. Yupp's failure underscores that crowdsourcing and decentralised AI infrastructure remain solutions in search of problems, unable to compete with centralised hyperscale model training and evaluation workflows.

Zhipu's shares surged 35% in Hong Kong despite reporting 60% higher losses in 2025, driven by investor optimism about agentic AI prospects, per Bloomberg. Chinese AI stocks are now among the most volatile in Asian equity markets, with newly listed model developers and chip designers swinging on retail flows uncorrelated to financial performance, according to Bloomberg. In contrast, a Taiwan-focused fund investing in smaller AI supply chain firms beat 99% of peers by diversifying across component manufacturers rather than concentrating in headline names, per Bloomberg.

The Chinese AI equity market is exhibiting speculative characteristics similar to Western meme stocks, with retail investors driving valuations based on narrative momentum (agentic AI, national champion status) rather than unit economics or path to profitability. Zhipu's 35% rally despite widening losses indicates that public market participants are pricing in future monopoly rents from consumer AI dominance in China, a bet that assumes regulatory protection and elimination of domestic competition. The Taiwan supply chain fund's outperformance suggests that disciplined capital allocation into diversified manufacturing exposure (optics, PCBs, cooling systems) is generating superior risk-adjusted returns compared to concentrated bets on model developers or marquee chip designers like TSMC.