The AI Native Cloud
HQ: San Francisco, CA | Founded: 2022 | Employees: 400-450 | Stage: Series C | Website: https://together.ai
Together AI, founded in 2022 in San Francisco, is the AI Native Cloud — a full-stack platform combining state-of-the-art open-source models, high-performance inference infrastructure, and frontier research in AI efficiency and scalability. The company was co-founded by Vipul Ved Prakash (CEO, serial entrepreneur behind Cloudmark and Topsy), Ce Zhang (CTO, Stanford professor), Percy Liang (Chief Scientist, Stanford HAI director), Chris Ré (Stanford professor, Snorkel AI co-founder), and Tri Dao (Chief Scientist, FlashAttention creator). Together AI powers over a million developers and thousands of paying customers with production-scale inference, training, fine-tuning, and reinforcement learning.
The platform spans serverless inference across 200+ open-source models, dedicated GPU endpoints, batch processing, fine-tuning (full and LoRA), GPU clusters (on-demand and reserved), code sandboxes, and managed storage. Its research-to-production pipeline includes FlashAttention-4 (achieving 71% GPU utilization on Blackwell), RedPajama datasets (100T+ tokens used by Snowflake Arctic, AI2 OLMo, and Salesforce XGen), and ATLAS inference optimizations. Customers report 6x to 60x cost savings compared to closed-model pricing, with annual bookings crossing $1.15B.
Together AI has secured over 500 MW of compute capacity from investors and partners, including a landmark 250 MW deal with Saudi Arabia's HUMAIN (projected $5B gross annualized revenue), a $240M multi-year agreement with IBM Cloud for NVIDIA B300 clusters, and partnerships with Y Combinator for dedicated startup GPU clusters.
Together AI is a research-driven AI infrastructure company that has rapidly become a leading cloud platform for running open-source AI models. With $1.3B+ raised, $1.15B annual bookings, and a customer base including Cursor and Decagon, it occupies the broadest position in the open-model inference stack. Its FlashAttention research lineage and aggressive compute expansion (500+ MW secured) position it as a credible contender to hyperscaler AI clouds.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki