The AI-inference platform that turned 'running models in production' into its own infrastructure category.
Updated · Analysis by Trace Cohen · baseten.co
Series F, Jun 2026
Series F, Jun 2026
Mar 2026 — ~+1,900% YoY, up from $200M in Dec 2025
across 87 clusters, 18 clouds
Jun 2026
From ~$200M annualized in Dec 2025 to ~$600M by Mar 2026 — a ~20x YoY jump.
Source: Company reports / press (2025–26)
~$13B — a $1.5B Series F in June 2026 (two tranches, $11B and $13B) led by Altimeter, Conviction, and Spark Capital. Just five months earlier it was a $5B company.
Baseten runs usage-based AI inference: customers pay either for API consumption on open-source models or for GPU minutes/hours, with Baseten packaging autoscaling, observability, billing, and developer tooling around it.
Growth is driven by companies moving workloads onto open-source and custom models instead of closed APIs — Baseten is the production layer that makes that switch operationally viable, which is why its annualized revenue tripled in a single quarter.
Baseten raised a $300M Series E in January 2026 at $5B (IVP, CapitalG, Nvidia), then a $1.5B Series F in June 2026 at up to $13B (Altimeter, Conviction, Spark, with Sands, Wellington, IVP, and Greylock).
Open-model inference and fine-tuning platform.
Fast inference for open-source models.
Developer-first model-serving and compute platforms.
AWS Bedrock, Azure, GCP inference plus Nvidia's own stack.
Baseten is the purest bet on the 'inference is the real AI infrastructure' thesis: a $5B company in January, ~$13B by June, with annualized revenue up ~20x year over year. The growth is real because the tailwind is real — every team moving off closed APIs onto open-source models needs a production layer. The risk is that inference is the most contested layer in AI, with hyperscalers and Nvidia sitting directly upstream of Baseten's margins.
Baseten is valued at roughly $13 billion after a $1.5B Series F in June 2026 (priced across two tranches at $11B and $13B), led by Altimeter, Conviction, and Spark Capital — up from $5B just five months earlier.
Baseten reached roughly $600 million in annualized revenue by March 2026 — about +1,900% year over year — up from $200 million in December 2025.
Baseten runs a usage-based AI inference platform: customers pay for API consumption on open-source models or for GPU minutes/hours, and Baseten handles autoscaling, observability, billing, and tooling. It serves 1B+ inference calls per day across 87 clusters and 18 clouds.
As companies move workloads off closed APIs and onto open-source and custom models, they need a reliable production layer to serve them at scale — making inference one of AI's most contested infrastructure layers, with platforms like Baseten growing ~20x year over year.
Analysis by Trace Cohen · @Trace_Cohen · t@nyvp.com. Figures are as of the update date; verify before relying on them.