Baseten Raises $1.5 Billion Series F at Up to $13 Billion Valuation

Baseten raised a $1.5 billion Series F led by Altimeter Capital, Conviction, and Spark Capital, valuing the AI inference startup at up to $13 billion — five months after a $300 million round at a $5 billion valuation.

Technology

uristocrat

July 2, 2026 2 min read

Baseten Raises $1.5 Billion Series F at Up to $13 Billion Valuation — Baseten. Image: Baseten

Key Points

Baseten raised a $1.5 billion Series F led by Altimeter Capital, Conviction, and Spark Capital.
The round values the AI inference platform at up to $13 billion, across two tranches priced at $13B and $11B.
Baseten processes more than 1 billion inference calls per day across 87 clusters and 18 clouds.
It comes five months after a $300 million Series E at a $5 billion valuation in January.

Baseten has raised a $1.5 billion Series F led by Altimeter Capital, Conviction, and Spark Capital, with Sands Capital and Wellington Management joining as co-leads. The round, announced June 22, values the company at up to $13 billion — structured across two tranches priced at $13 billion and $11 billion.

Baseten does not build frontier models. It runs them. The company operates inference infrastructure — the layer that serves a trained model to an application every time a user actually calls it — and it now processes more than one billion inference calls a day across 87 clusters and 18 clouds. Per estimates from research firm Sacra, its revenue run-rate climbed from roughly $200 million in December 2025 to about $600 million by March 2026, growth of nearly 1,900% year over year.

For two years, the largest AI checks followed whoever was training the biggest model. Baseten's raise is a marker that the market has repriced the other half of the stack. Inference is the cost that recurs — you pay it on every query, not once at training — and as open-source and closed models converge in capability, the durable advantage moves from building a frontier model to serving one cheaply and reliably at scale. That is why an infrastructure company that runs other people's models just raised more than many of the labs that make them.

The pace is the story as much as the price. Baseten raised a $300 million Series E at a $5 billion valuation in January, led by IVP and CapitalG with a $150 million check from Nvidia. Five months later, the Series F values it at up to $13 billion — roughly two and a half times higher — the kind of markup investors reserve for a category they believe is being defined in real time. The training side of that stack is drawing the same money: Prime Intellect raised a $130 million Series A at a $1 billion valuation for its enterprise AI training platform.

The same logic is reshaping the hardware beneath it. It is why OpenAI and Broadcom built Jalapeno, a custom inference chip, and why Meta laid out four MTIA chip generations on a six-month cadence — both aimed at the part of the AI bill that repeats every time a model runs rather than trains. Even Nvidia, whose GPUs still dominate that layer, is hedging into new markets by entering the PC chip business with RTX Spark. Inference has become its own infrastructure war, and Baseten just got funded to fight it at the software layer.

Base Power reached the same $13 billion mark in August, closing a $1 billion Series D for its home-battery fleet.

Further down the same stack, Naïve raised a $28.5 million Series A to build infrastructure for AI-run companies.

The inference layer is drawing hardware money too: London chip startup OLIX raised $312 million in Series B financing at a $3.3 billion valuation to build a photonic accelerator aimed at the decode stage of model inference.

Source: Business Wire