# Baseten

> Baseten is a San Francisco-based AI inference infrastructure company that provides dedicated and serverless GPU compute for running AI models at scale. Founded in 2019 by four ex-Gumroad engineers, the company has grown into a unicorn with a $5B valuation and $585M in total funding, backed by NVIDIA and other top-tier investors. Baseten powers inference workloads for 100+ enterprises including Cursor, Notion, HeyGen, and Clay, offering an inference stack with near-zero cold starts, proprietary networking, and open-source tooling like Truss for model packaging.

- **Founded:** 2019
- **Headquarters:** San Francisco, CA, USA
- **Founders:** Tuhin Srivastava (CEO & Co-founder), Amir Haghighat (CTO & Co-founder), Philip Howes (Co-founder), Pankaj Gupta (Co-founder)
- **Team size:** ~200 employees as of early 2026 (133% YoY headcount growth through mid-2025)
- **Products:** Inference Stack, Model APIs, Truss, Baseten Chains, Baseten Delivery Network (BDN)
- **Notable:** Achieved unicorn status in 2025 with $5B valuation at Series E, Total funding of $585M across six rounds, 100x growth in inference volume year-over-year through 2025

## Products & services

- **Inference Stack** — Dedicated and serverless GPU-backed inference infrastructure with proprietary networking and near-zero cold starts, enabling high-throughput and low-latency model serving.
- **Model APIs** — Pre-built API endpoints for popular open-source models, enabling teams to call models without managing infrastructure.
- **Truss** — Open-source framework for packaging, testing, and deploying ML models. The flagship OSS project with 1.1K+ GitHub stars across the basetenlabs org.
- **Baseten Chains** — Compound AI pipeline system enabling developers to chain multiple models and logic steps together in production workflows.
- **Baseten Delivery Network (BDN)** — Proprietary global network layer optimizing GPU routing, scheduling, and data transfer to minimize latency across regions.
- **Training Infrastructure** — GPU cluster access and tooling for model fine-tuning and training workloads, launched in 2025.

## Achievements

- Achieved unicorn status in 2025 with $5B valuation at Series E
- Total funding of $585M across six rounds
- 100x growth in inference volume year-over-year through 2025
- Recognized on Enterprise Tech 30 list
- 95% claimed win rate in head-to-head competitive evaluations
- Near-zero customer churn among enterprise accounts
- Open-source Truss framework with 1.1K+ GitHub stars
- 91+ public repositories on GitHub under basetenlabs org
- Acquired RL startup Parsed in December 2025
- NVIDIA strategic investment underscores deep GPU infrastructure partnership

## Latest updates

- **2026-01** — Closed $350M Series E led by NVIDIA with $150M strategic investment; valuation reached $5B.
- **2025-12** — Signed AWS Strategic Collaboration Agreement (SCA) for preferred cloud infrastructure and joint enterprise go-to-market.
- **2025-12** — Acquired Parsed, a reinforcement learning startup, to bolster training and optimization capabilities.
- **2025** — Launched Training Infrastructure product for fine-tuning and training workloads on GPU clusters.
- **2025** — Surpassed 100 enterprise customers; reported 100x YoY inference volume growth.
- **2025** — Grew headcount by 133% year-over-year to approximately 200 employees.

## Links

- Website: https://baseten.co
- LinkedIn: https://www.linkedin.com/company/baseten
- Twitter/X: https://twitter.com/basetenco
- GitHub: https://github.com/basetenlabs

---

Profile page: https://yespress.io/baseten
Published by YesPress — https://yespress.io
Last updated: 2026-04-02