Accel reportedly in talks to lead $1B round for Thinking Machines at $40B valuation

Thinking Machines, Mira Murati’s AI lab, is reportedly in talks to raise $1B at a valuation of at least $40B, with Accel potentially leading. The implied revenue multiple is extremely high given a reported $100M+ annual revenue run-rate. The new valuation would still undercut the ~$50B level the company sought late last year, and comes after it launched the Inkling open-weight model on its Tinker platform.

Analysis

This is less about one startup and more about how much optionality capital is still willing to pay for frontier AI exposure. A valuation that embeds years of future monetization at only low-eight-figure revenue run-rate implies the market still believes compute scarcity, not software scarcity, is the bottleneck; that is structurally supportive for NVDA, GPU supply chain vendors, and the hyperscalers funding the arms race.

The second-order effect is capital concentration. If these mega-rounds become the only way to finance frontier labs, smaller AI entrants will have to build on top of incumbent models rather than challenge them, which can slow app-layer competition while reinforcing winner-take-most dynamics at the infrastructure layer. That is mildly negative for broad private-market dispersion, but positive for the few names that can monetize every incremental inference cycle.

Near term, the real catalyst is whether this pricing clears and is followed by more copycat rounds; if it does, it will likely reinforce 2025 capex guidance into the next earnings season. The contrarian read is that the lower target versus prior expectations and the talent churn are early signs of private-market discipline, meaning the bubble narrative may be right on valuation but wrong on the public-market trade: a funding reset hurts late-stage startups more than it hurts the chip vendor selling picks-and-shovels. Falsifiers would be any sign of hyperscaler capex deceleration, GPU order pushouts, or a pullback in AI training spend in the next 1-3 months.

Key Decisions for Investors

• Buy NVDA on a 2-4% pullback over the next 3-5 sessions via a 1-3 month call spread; thesis is continued scarcity pricing for training/inference compute, with upside into the next hyperscaler capex prints.
• Pair trade: long NVDA / short AMD over the next 1-3 months; if frontier funding remains robust, the market should keep paying a premium for the most entrenched CUDA/vendor lock-in exposure. Stop if AMD commentary shows faster-than-expected share gains or pricing is stabilized by inference demand.
• Prefer SMH over QQQ on any broad-tech weakness for a 1-3 month tactical trade; this is a cleaner expression of AI infrastructure demand than chasing private-market headlines. Risk/reward improves if the next earnings cycle confirms capex discipline and backlog conversion.
• No trade in ISC.TO based on this item alone; there is no identifiable read-through. Reassess only if the article is a proxy for a broader financing shock in Canadian tech/private markets.