Together AI's $800M Series C Tests Whether Open-Source Inference Can Scale to an Enterprise Default

A round led by Aramco Ventures at an $8.3 billion valuation marks Together AI's most significant bid to move from developer-friendly inference hosting toward a durable enterprise compute platform.

Together AI closed an $800 million Series C on Wednesday, pushing its valuation to $8.3 billion in a round led by Aramco Ventures, the investment arm of Saudi Arabia's national oil company. The deal is one of the largest single raises for an AI-infrastructure company focused exclusively on open-source model serving, and it arrives at a moment when enterprise demand for compute alternatives to the major hyperscalers has moved from a niche preference to a procurement imperative. Multiple sources confirmed the terms, making this one of the more legibly priced milestones in the neocloud segment this year.

Founded in 2022, Together AI built its platform around a straightforward thesis: that the market for running large language models at scale would eventually fragment away from three dominant cloud providers, and that developers willing to trade managed convenience for lower unit cost would represent a large and durable customer base. The company hosts hundreds of open-source models — including Meta's Llama family and Mistral's catalog — on its own GPU clusters, pricing inference well below AWS Bedrock or Azure OpenAI Service equivalents. By late June 2026, the company reported its platform was processing 400 trillion tokens — a figure that, while not independently audited, suggests meaningful throughput scale, and one whose growth has accelerated as enterprise teams have adopted open-weight models as a concrete cost-control lever rather than a research experiment.

The choice of Aramco Ventures as lead investor warrants attention beyond the headline number. Saudi Aramco has been systematically deploying capital into AI infrastructure as part of a national strategic pivot away from hydrocarbon dependency, and a lead position in a frontier-adjacent compute platform aligns with that agenda. For Together AI, the relationship carries potential beyond pure financing: Aramco's commercial network spans industrial operators across the Middle East and Asia that represent a largely underpenetrated enterprise AI customer base. The New York Times noted on July 1 that Together AI has been actively pitching companies seeking cheaper AI options and alternatives to Nvidia-dependent supply chains — a message that resonates in regions where cloud spend accountability is tightening and where sovereign AI ambitions are driving domestic infrastructure investment.

The structural risks, however, are significant and should not be understated. The open-source inference market is already crowded: Fireworks AI, Groq, Cerebras, and Replicate all compete on similar price-performance vectors, and the hyperscalers themselves have responded with spot-instance pricing and serverless inference tiers that continue to narrow the cost gap. Together AI's advantage has historically rested on model breadth and developer familiarity, but neither constitutes a durable moat — any well-capitalized entrant can replicate a model catalog within months. The company's infrastructure also carries a notable internal tension: it relies substantially on Nvidia GPU supply, which means that the cheaper-than-hyperscaler narrative it cultivates applies to its customer pricing but not yet to its own procurement costs. The gap between its public positioning and its supply-chain reality is one that $800 million may help address, but will not resolve overnight.

At an $8.3 billion valuation, Together AI is priced for a transition from pure inference hosting toward something more proprietary — likely custom fine-tuning services, enterprise deployment tooling, or a platform layer that aggregates demand across multiple clouds. Whether it can execute that transition before better-capitalized competitors converge on the same space, or before open-source inference pricing approaches commodity levels, is the central question the round leaves open. Three concrete signals are worth watching: first, how Together AI allocates the $800 million — specifically whether capital flows toward proprietary silicon partnerships or toward geographic expansion into Middle Eastern and Asian markets where the Aramco relationship creates natural distribution leverage; second, whether token volume growth translates into improving gross margins, which would indicate genuine pricing power rather than volume-at-loss customer acquisition; and third, the pace at which enterprise customers move from evaluation pilots to production contracts, a conversion dynamic the company will likely be pressured to disclose as it positions toward a potential public offering.