Baseten Raised $1.5B Series F

Blog publsihed by Baseten Juen 22, 2026

We are thrilled to announce Baseten’s $1.5B Series F, led by Altimeter Capital, Conviction Partners, and Spark Capital, co-led by Sands Capital and Wellington Management, with participation from Battery Ventures, Blackbird, D.E. Shaw Ventures, Durable Capital Partners, Greylock, IVP, Verified Capital, and 01A.This is our fourth fundraise in 18 months, and we are grateful for our investors’ conviction and support and, most importantly, for the trust and partnership of our customers.

Over the last year, our revenue has grown 20x, and inference volume has grown 40x. This growth reflects the new market reality — inference is the most important layer in the AI stack, and Baseten has become the partner of choice for companies looking for support with inference and post-training.

We are proud to work closely with the teams defining this era, including Cursor, Notion, Lovable, Harvey, HubSpot, OpenEvidence, Abridge, Decagon, Parallel, and many more. These companies have built products where intelligence is core to the user experience, and their businesses are booming. They are also at the forefront of a wave of companies that are more motivated than ever to own their intelligence. As Satya Nadella shared last week, durable advantage will come from the learning systems companies build around their own data, judgment, workflows, and feedback loops. We see the same thing every day: our customers are building systems shaped by proprietary data, evals, domain-specific harnesses, post-training, and custom RL environments.

The opportunity ahead has never been so apparent: open-weight models have become strong enough that enterprises can now use them as serious alternatives to closed APIs. Baseten’s work is to help companies take advantage of this progress with confidence. Our researchers and engineers embed with our customers, post-train and optimize specialized models, and serve those models with exceptional latency, reliability, observability, and cost at production scale.

This financing allows us to invest aggressively in the compute, software, and talent required to support our customers as AI becomes central to their products and businesses.

If you want to post-train custom models and run inference on the fastest, most reliable platform, we’d love to help. If you want to build the infrastructure layer of AI, we are hiring.

Comments

We took time to study this one as Baseten doesn't immediately fit in categories we usually follow.

This new round is the company's fourth raise in 18 months. The company says revenue grew 20x and inference volume 40x over the past year, and it names Cursor, Notion, Lovable, Harvey, HubSpot and OpenEvidence as customers. Management argues that open-weight models are now good enough for enterprises to treat them as real alternatives to closed APIs. Baseten's pitch is to embed its engineers with customers, post-train specialized models, and serve them at production scale. The new money goes to compute, software and hiring.

The trajectory of this investment is interesting with Nvidia among investors. The round was split into two tranches, priced at $11B and $13B. Five months earlier, the $300M Series E in January 2026, led by IVP, CapitalG and Nvidia, valued the company at $5B. Annualized revenue run-rate reportedly tripled from $200M to $600M in a single quarter. For comparison, the Series D in September 2025 was $150M at $2.15B.

Their solutions have seriously evolved and address key markets needs. The portfolio covers dedicated inference, Model APIs, training/post-training, multi-cloud capacity management, and cloud, self-hosted or hybrid deployment. It spans LLMs, transcription, TTS, image generation and embeddings. The business is usage-based: customers pay per API call or per GPU minute or hour. Since the round, Baseten has also announced it is acquiring Blaxel, moving toward agentic infrastructure.

Market and competition. The inference layer is now the most heavily funded part of the AI stack below the model labs, and the other players have raised even bigger rounds since June:

Fireworks AI raised $1.505B at a $17.5B valuation on July 16, 2026, with ARR above $1B
Together AI raised $800M at $8.3B on July 1
Modal, Replicate and RunPod compete directly
AWS Bedrock, Azure, Google Cloud and Nvidia's own software stack are the bigger threat
Custom-silicon players such as Groq compete indirectly

The company has chosen a different approach. Baseten is betting that software optimization, not proprietary chips, drives down cost per token. That leaves it squeezed between hyperscalers that own the cloud relationship and silicon vendors below. Its main edge is price, so if competitors undercut it, the valuation becomes harder to defend. Baseten is growing faster than Fireworks but is still smaller, at roughly 0.6x its revenue and a lower multiple. Consolidation in this segment looks likely within 12–24 months. We'll see and there is a good chance...