TypeSafe AI raises $870 million at a $7.5 billion valuation weeks after Jev’s launch

TypeSafe AI’s promise is to make automated decisions faster and cheaper without asking AI to write an answer. On October 9, 2026, it announced an $870 million Series A at a $7.5 billion valuation. The financing follows Jev’s September launch; the adoption and performance figures accompanying it come from the company and its lead investor.

Andreessen Horowitz led the round, with Sequoia Capital and existing investor DCVC participating. TypeSafe also named angel investors without identifying them individually. The company said Martin Casado will join its board. In its own investment announcement, Andreessen Horowitz framed the bet around making existing software more intelligent, rather than simply using AI to write more software.

A decision handed straight to code

Jev was released on September 15. It uses a transformer architecture, but it is not a large language model, according to TechCrunch. Instead of producing text, it returns probabilities that TypeSafe calls “calibrated decisions.” The company positions it for automating tasks rather than generating prose or code—a narrower job than the open-ended work associated with a language model.

Andreessen Horowitz describes the distinction as a change in how a model communicates with software. A text-producing model encodes its decision in words, leaving the application to parse and normalize the answer. Jev instead hands code a typed value: an answer in a form the software can use directly. The investor says Jev’s output is so small that TypeSafe does not charge for it.

The investor’s performance pitch is substantial. It says Jev costs roughly one-hundredth to one-five-hundredth as much as frontier models and is 100 times faster for classification tasks at comparable accuracy. Classification means assigning something to a category. That task-specific comparison is the basis of the speed claim, not a claim that Jev can perform every kind of work a text-generating model handles.

1 trillionTokens generated

Andreessen Horowitz says Jev reached 1 trillion tokens generated within three days of launch. This is the lead investor’s reported usage figure.

Two different descriptions of enterprise adoption

TypeSafe says a third of Fortune 500 companies are already using Jev. Andreessen Horowitz gives a different figure and description: 25% of Fortune 500 enterprises have “integrated” it. Those statements do not describe adoption in exactly the same terms. Neither announcement explains the difference, so they should not be treated as matching measurements of enterprise deployment.

The company also says Jev has already saved customers millions of dollars in production. That statement concerns software customers are running, rather than benefits promised from the new financing. Separately, Andreessen Horowitz says thousands of use cases appeared during Jev’s first week, including generative interfaces, interactive gaming and data analysis. It presents that developer activity as part of its investment rationale.

More models and enterprise features are plans

TypeSafe’s funding announcement sets out its intended direction, rather than introducing another model. It describes separate priorities for developers and business customers, tying the new capital to a wider range of machine-native AI and additions to its enterprise offering. The commitments remain broad:

• For developers, TypeSafe says it intends to push Jev’s existing strengths further, build more machine-native models and provide additional infrastructure for smart software.
• For businesses, it says it will add enterprise features customers have requested. The announcement does not identify those features or give delivery dates.

Founded in 2024, TypeSafe brings together former OpenAI researcher Diogo Almeida, former Meta research engineer Sasha Sheng, and engineer and entrepreneur Erik Gafni. Their company now has a large financing behind a deliberately focused product: AI that supplies decisions to software, rather than written responses to people.