Exclusive: Google deepens Thinking Machines Lab ties with new multi-billion-dollar deal

The more consequential detail buried in the headline is that Google is routing capital through Cloud infrastructure contracts rather than direct equity, which lets both parties claim independence while Google secures a major committed workload on its platform before Murati's models are even publicly benchmarked.

This fits a pattern Modelwire has been tracking since mid-April: infrastructure position is becoming the actual competitive moat, not model quality. The MIT Technology Review piece we covered on April 16 argued exactly this, that enterprise AI advantage lives in controlling the operational layer, not in capability benchmarks. The Cerebras IPO story from April 18 adds another data point: specialized hardware companies are gaining leverage precisely because hyperscalers and well-funded independents like Thinking Machines Lab are competing to lock in compute relationships before the next generation of training runs begins. Upscale AI raising at a $2B valuation after just seven months, also from our April 16 coverage, suggests capital is flooding infrastructure broadly, not just to marquee names.

Watch whether Thinking Machines Lab announces a public model release or enterprise access program within six months of this deal closing. If the compute commitment precedes any public product, that confirms this is primarily a platform lock-in play by Google rather than a bet on Murati's near-term model output.