Baseten привлёк $1,5 млрд при оценке $13 млрд — ставка на дешёвые AI-модели

American startup Baseten raised $1.5 billion in a Series F round at a valuation of $13 billion. This is one of the largest funding rounds of the year in the AI infrastructure segment — and a clear signal about where the market is heading.

What is Baseten

Baseten is not another language model and not an LLM startup in the conventional sense. The company operates at the infrastructure layer: it provides software and computing power to organizations that want to independently deploy and scale AI models — especially cheaper open-source options. In other words, Baseten solves a problem that more and more companies face: how to take an open model — Mistral, Llama, Gemma and others — and quickly launch it in production without needing to build your own infrastructure from scratch. The platform provides:

• Ready-made environment for deploying AI models
• Optimized computing power for inference
• Scaling tools for variable load
• API layer for integration into client products

Who invested and why this matters

The Series F round was led by Altimeter — one of the most active technology funds in the AI segment. The participation of Altimeter specifically is notable: the fund is known for cautious but large bets on infrastructure companies with clear and proven business models. The $13 billion valuation makes Baseten one of the most expensive startups in the AI infrastructure category — alongside names like Together AI, Modal and Replicate. The market is clearly consolidating around a few major platforms.

Why cheap models became a strategic asset

Baseten's main bet is on the trend toward using economical models. Over the past 12–18 months, the market has undergone a clear shift: companies that initially built products on top of GPT-4 or Claude are beginning to migrate to more accessible alternatives. There are several reasons:

• The cost of inference on open models is many times lower than proprietary options
• The quality of Llama 3, Mistral Large and analogs has reached a level sufficient for most business tasks
• Local deployment allows companies to keep data within their own perimeter — fundamental for corporations and regulated industries

"Companies are beginning to understand that they don't need the most expensive models for most tasks," noted

Baseten CEO Tushin Srivastava in an interview with Bloomberg.

This is exactly where Baseten stands: between model developers and companies that want to use them. Where AWS, Google Cloud and Azure push their own AI services, Baseten offers a vendor-neutral platform — deploy any model, scale to your requirements.

What this means

Baseten's round reflects the broader market logic: the center of gravity in AI is shifting from model development to their efficient exploitation. The infrastructure layer — hosting, inference optimization, compute management — is becoming a key competitive advantage. Funds, judging by Altimeter's bet, are willing to pay $13 billion for those who build this infrastructure best.