TypeSafe valued at $7.5 billion three weeks after leaving stealth

San Francisco AI lab TypeSafe closed an $870 million Series A led by Andreessen Horowitz, with Sequoia Capital and existing investor DCVC. The post-money valuation is $7.5 billion. Bloomberg first reported the round, which the company and a16z confirmed. Martin Casado joined the board.

TypeSafe left stealth only three weeks earlier with a $40 million seed round led by DCVC, reportedly priced at about $200 million, so its valuation rose roughly 37-fold. The company says about a third of the Fortune 500 use its product but has not named customers.

Its model, Jev, returns structured output that software can consume directly instead of conversational text, for example pulling the amount, date and tax number from an invoice in exactly the expected format. The name refers to type safety, guaranteeing data is of the expected type.

Reliability is a major barrier to enterprise AI. A model feeding accounting or logistics systems must produce correctly formatted output every time. Investors are betting that agents and automation will drive demand for such specialised models.

TypeSafe is not alone. Ramp reached a $60 billion valuation, evaluation platform Arena was valued at $3.1 billion and agent startup Manus raised more than $500 million. Investors are backing models that solve one problem very well rather than competing head-on with big labs, though such rapid step-ups recall the 2021 startup bubble.

OpenAI, Anthropic and Google keep improving their own structured output features, which could narrow TypeSafe's edge. Unverified customer claims and no disclosed revenue mean the valuation rests on expectations, while public AI stocks fell sharply the same week.

The first real signal will be named customers and usage figures.

Pilot automated extraction from invoices, orders and forms, add a validation step for every field sent into your systems, and define a fixed schema so models return structured output instead of free text.