Together AI has raised $800 million in a Series C financing round, valuing the San Francisco-based AI infrastructure company at an $8.3 billion post-money valuation. Announced on July 1, the funding was led by Aramco Ventures and represents a sharp step-up from Together AI’s $3.3 billion valuation in February 2025. The transaction underscores accelerating investor demand for companies providing the computing and software layer behind enterprise adoption of open-source artificial intelligence.
Series C Financing
The round also included NVIDIA, Vista Equity Partners, General Catalyst, Emergence Capital, Schneider Electric, Pegatron, Salesforce Ventures, March Capital, DTCP Growth, Lux Capital, Geodesic, PSP Partners, and other backers. Together AI said it has additionally secured commitments for more than 500 megawatts of computing capacity, which new investors will capitalize independently to support expected infrastructure growth. That combination of equity funding and compute commitments gives the company fresh resources to scale its platform while addressing the capital-intensive requirements of large AI workloads.
Founded in 2022, Together AI operates an AI cloud platform for training, fine-tuning, and running models in production. Its offering centers on open and customizable models, including DeepSeek, Nemotron, MiniMax, Kimi, and GLM, alongside accelerated compute and inference services. The company’s position is based on the premise that businesses can reduce dependence on proprietary large language models without sacrificing performance for many applications.
Building Production AI Infrastructure
Together AI argues that inference, the process of running trained models, is increasingly becoming a core production cost as organizations deploy AI agents for software development, customer support, document analysis, and workflow automation. The company says it is optimizing the wider stack that determines the economics of those deployments, including models, kernels, compilers, training systems, hardware utilization, and infrastructure software. Its recent product and research work has included FlashAttention-4 for NVIDIA Blackwell hardware, Together Megakernel, together.compile, and expanded post-training capabilities for tool calling, reasoning, and vision-language models.
The company says customers using open models through its platform can achieve cost reductions of between six and 20 times compared with closed-model alternatives, while retaining comparable or better performance. It cited AI customer-service company Decagon as an example, saying the customer cut inference costs sixfold after moving to Together AI. These figures are company-reported, but they illustrate the commercial argument behind the company’s open-model strategy as AI workloads move from experimentation into sustained operational use.
Growth and Market Position
Together AI said annual bookings exceeded $1.15 billion in the latest quarter and that it serves thousands of paying customers, including Cursor, Cognition, Decagon, ElevenLabs, and Suno. Reuters reported that its computing capacity and infrastructure are expected to expand roughly 50-fold over the next five years, reflecting the company’s ambition to become a larger provider of production inference. The growth plan will place Together AI in increasingly direct competition with hyperscale cloud platforms and providers of proprietary model APIs, while also exposing it to the execution risks that accompany rapid infrastructure buildouts.
The Series C gives Together AI both financial scale and additional compute commitments at a moment when enterprise AI spending is shifting toward production-grade deployments. For co-founder and CEO Vipul Ved Prakash, together with founders Percy Liang and Ce Zhang, the central objective is to make high-performance AI more accessible through open models and infrastructure rather than concentrate it in a small group of providers. Whether the company can convert its reported demand into durable margins will depend on its ability to expand capacity efficiently, maintain technical performance, and defend its position in a fast-moving AI infrastructure market.