Skip to main content

Baseten

Category: AI Infrastructure

A serverless infrastructure platform designed for machine learning engineers to deploy, manage, and scale AI models in production with high performance and minimal operational overhead. Baseten was founded in 2019. The company is led by Tuhin Srivastava. Based in San Francisco, USA. Team size: 150-250. Total funding raised: ~$585M total across 6 rounds. Latest round: Series E ($300M, Jan 2026) at $5B valuation. Key investors include ["IVP","CapitalG","NVIDIA","Spark Capital","Greylock","BOND","Conviction","01A","Altimeter","Battery Ventures","South Park Commons","Lachy Groom","BoxGroup","Blackbird Ventures","Premji Invest","Scribble Ventures"].

Founded
2019
Headquarters
San Francisco, USA
Team size
150-250
Total funding
~$585M total across 6 rounds

Value proposition

Eliminates the complexity of managing GPU infrastructure by providing a developer-centric, Python-first environment that reduces model deployment time from weeks to minutes while optimizing for high-throughput inference.

Products and solutions

["Truss: An open-source model packaging standard for seamless transition from development to production.","Serverless GPU Inference: Scalable infrastructure for running models like Llama 3, Mistral, and Whisper.","Baseten Chains: A framework for building and deploying multi-step, complex AI workflows and model pipelines.","Model Monitoring & Observability: Integrated tools for tracking model performance, latency, and resource utilization.","Custom Runtimes: Support for specialized hardware requirements and optimized execution environments."]

Unique value

Baseten bridges the 'deployment gap' by offering an open-source packaging standard (Truss) that ensures models run identically in local environments and on production-grade cloud GPUs.

Target customer

Machine learning engineers, data scientists, AI startups, and enterprise engineering teams moving LLMs and generative AI into production.

Industries served

["Technology & Software (SaaS)","Healthcare & Life Sciences","Financial Services","E-commerce & Retail","Media & Entertainment"]

Technology advantage

Proprietary GPU orchestration layer that significantly reduces 'cold start' times for large language models (LLMs) and optimizes autoscaling based on real-time inference demand, offering better cost-performance ratios than generic cloud providers.

How they differentiate

Baseten differentiates through its open-source model packaging standard, Truss, which ensures parity between local development and production environments. It also offers 'Baseten Chains' for complex, multi-model workflows and a highly customizable serverless GPU orchestration layer that minimizes cold starts.

Main competitors

["Replicate","Modal","Together AI","Anyscale"]

Key partnerships

["NVIDIA (Strategic investor — $150M in Series E, deep GPU optimization partnership)","Cloud Infrastructure Providers (AWS, GCP for underlying compute)","Open-source communities (Lead maintainers of Truss)","Model Providers (Optimized hosting for Meta's Llama, Mistral AI, DeepSeek, Qwen, GLM models)","Parsed (Acquired Dec 2025 — reinforcement learning / post-training optimization)","Inferless (Acquihired Mar 2026 — inference infrastructure team)"]

Notable customers

["Abridge","Clay","Cursor","OpenEvidence","Notion","Gamma","Writer","Sourcegraph","Zed Industries","Wispr","ClickUp","Mercor","Bland","Mirage","Descript"]

Major milestones

["Launched Truss, an open-source standard for model packaging","Raised $40M Series B to scale global GPU infrastructure in March 2024","Raised $75M Series C in February 2025 at $825M valuation","Raised $150M Series D in September 2025 at $2.15B valuation led by BOND","Acquired Parsed, a reinforcement learning startup, in December 2025","Raised $300M Series E in January 2026 at $5B valuation led by IVP and CapitalG with NVIDIA investment","Acquihired Inferless team in March 2026 for inference infrastructure expertise"]

Growth metrics

Achieved a $5 billion valuation in January 2026; 100x inference volume growth in 2025; 10x revenue growth year-over-year; billions of weekly LLM calls; serves hundreds of organizations.

Market positioning

Enterprise-grade AI inference infrastructure provider focusing on high-performance deployment of open-source and custom models.

Geographic focus

Global (Cloud-based), with a primary business focus on North America and Europe.

Patents and IP

No specific public patents disclosed; focuses on proprietary orchestration logic and open-source contributions to the ML ecosystem.

About Tuhin Srivastava

Tuhin Srivastava is the Co-founder and CEO of Baseten. He was previously the Co-founder of Shape (a 3D body scanning company), which was acquired by Carbon Health in 2021. His professional background includes serving as an early Software Engineer at Gumroad and an Investment Banking Analyst at Macquarie Group. He holds a BS in Electrical Engineering from the University of Southern California.

Official website: