Capital & Industrial Strategy

OpenAI is in early-stage discussions with investors for a new private round that would value the company at $1.2 trillion or more, according to The Information. CEO Sam Altman has already ruled out an IPO this year, and the strategic logic for staying private is straightforward: it shields the company from quarterly revenue pressure while it ramps enterprise sales from a relatively modest base. The implied valuation represents a significant step-up and would make OpenAI the most valuable private company in history, but terms are unconfirmed and the round is described as early-stage conversations, not a committed deal.

The revenue ambition behind that valuation is being backed by serious sales infrastructure investment. OpenAI has hired Brian McCarthy from SpaceX's Cursor unit as global sales chief, reporting to recently appointed CRO Dali Rajic, and has brought in two senior Snowflake sales executives, per The Information. Snowflake's GTM playbook — high-value enterprise land-and-expand in data-intensive industries — is an explicit template. Separately, OpenAI launched a legal-focused AI platform targeting law firms, per Reuters, a vertical that combines high willingness-to-pay with structured, high-value data — exactly the segment profile that justifies premium pricing and supports a $1.2 trillion revenue multiple narrative.

Huawei has pulled forward the launch of its Ascend 960DT AI chip to Q1 2027 — nine months earlier than originally planned — with a second variant, the 960PR, to follow in Q3 2027, according to Bloomberg and TechCrunch. The announcement came directly from Huawei's rotating chairman at the company's Connect conference in Shanghai, giving it credibility as an operational commitment rather than a roadmap aspiration. The explicit framing — replacing Nvidia in China and competing globally — reflects a strategic shift from defensive import substitution to offensive market ambition.

The acceleration matters structurally because it compresses the window in which US-aligned suppliers might consolidate their position in third-party markets. Chinese AI model developers, who currently generate only roughly 10% of the revenue of US peers per Rhodium Group analysis via CNBC, have been constrained partly by compute access. A credible domestic alternative at scale changes that equation. The simultaneous push by Moonshot's Kimi model into Wall Street financial data partnerships, per CNBC, suggests Chinese AI firms are also competing aggressively on the application layer despite the revenue gap.

India's updated semiconductor policy has attracted $12 billion in investment pledges from global and domestic investors within months of launch, per Bloomberg. Simultaneously, the country's leading state-run infrastructure financier has sanctioned loans exceeding 30 billion rupees ($313 million) each to at least four data centre projects — a total commitment topping $1.2 billion — reflecting active deployment of public capital rather than just policy signalling, per Bloomberg.

This twin-track approach — sovereign industrial policy for semiconductors combined with state-backed project finance for data centre infrastructure — mirrors the playbook that Taiwan and South Korea used to build their chip industries, as noted by The Economist in its analysis of how AI investment benefits are diffusing through Asian supply chains. India is positioning itself to capture both the manufacturing and the cloud infrastructure layers of the AI buildout, though the $12 billion in semiconductor pledges remains indicative intent rather than committed capital with confirmed project timelines.

Crusoe — which built OpenAI's largest data centre — closed a $3.9 billion funding round at a $30.9 billion valuation, per TechCrunch and Reuters. The strategic signal is the product pivot: Crusoe is now building both hyperscale facilities and factory-assembled modular 'AI factories' — a direct response to enterprise demand for AI compute that doesn't require proximity to a hyperscale campus. Nvidia's Jensen Huang simultaneously guided that chip sales will roughly double in the next year, per CNBC, providing the demand backdrop against which Crusoe's $30.9 billion valuation is being priced.

The infrastructure constraint is being addressed at the grid level as well. Google, Nvidia, Anthropic, and Emerald AI have formed a coalition targeting 100 GW of new grid capacity for data centres, per TechCrunch. That figure — roughly equivalent to the entire installed nuclear capacity of the United States — illustrates the scale of the physical infrastructure problem facing AI deployment. Meanwhile, Nebius has raised AI cloud prices again as compute demand outstrips supply, per Reuters, providing real-time price signal confirmation that supply remains structurally short.

SoftBank has increased its margin loan backed by Arm Holdings shares by $5 billion to $25 billion, per Bloomberg. This is a confirmed, closed financing — not a proposal — and reflects SoftBank's structural playbook of using its most liquid, appreciating asset as a funding vehicle for AI positions that are not yet generating returns. The loan scale implies SoftBank is pledging a substantial portion of its Arm stake as collateral, concentrating risk significantly.

Bridgewater's Greg Jensen offered a notable counterweight: the hedge fund holds a 'very small position' in the AI infrastructure trade because demand growth is largely priced in, per The Information. Jensen also called for regulating top AI compute holders like systemically important banks, per The Information — a framing that, if adopted by regulators, would impose capital and operational constraints on the hyperscalers most aggressively expanding their AI infrastructure positions. The divergence between SoftBank's aggressive leveraged exposure and Bridgewater's cautious positioning is a meaningful signal of valuation uncertainty at the infrastructure layer.