Together AI raises $800M Series C at $8.3B

Customers include Cursor, Cognition and ElevenLabs, and the valuation is 2.5x the $3.3B mark from 16 months earlier, on $1.15B in annualized bookings. The round also secured separate commitments for over 500 megawatts of compute capacity, structured outside the company's balance sheet and capitalized independently by investors. Other participants include Vista Equity Partners, General Catalyst, Emergence Capital, NVIDIA, March Capital, Pegatron and SentinelOne's S Ventures. The funding is meant to scale open-source inference infrastructure roughly 50-fold over the next five years. The pitch is cost. Together lets enterprises run models such as DeepSeek, Nemotron and Kimi at competitive performance and a fraction of the cost of closed systems, and claims savings of up to 60 times on inference. That figure is a company claim. Three inference infrastructure companies are now collectively valued above $40 billion, so rivals are well funded. For AI founders, the signal is that open-weight inference has become a large, investable layer. Teams building on open models now have a better-capitalized supplier of GPU capacity, which may help with pricing and availability. The 500 MW commitment, financed off the balance sheet, also shows how providers are paying for compute without diluting equity as much. Founders who depend on a single closed-model vendor have a credible alternative to price against.