{"slug": "together-ai-hits-8-3-billion-valuation-as-enterprises-flee-closed-ai-models-for", "title": "Together AI hits $8.3 billion valuation as enterprises flee closed AI models for open source", "summary": "Together AI announced an $800 million Series C led by Aramco Ventures, with Nvidia, Vista Equity Partners, General Catalyst, Emergence Capital, and Salesforce Ventures participating, valuing the company at $8.3 billion as enterprises shift from closed AI models to open-weight alternatives. The San Francisco-based company, founded in 2022, reported annual bookings exceeding $1.15 billion in its most recent quarter and claims open-source model usage has tripled in the past year, citing OpenRouter research. Together AI says companies using open models can see costs 6 to 20 times lower with equal or better performance, with Decagon cutting inference costs sixfold after moving to the platform.", "body_md": "*Together AI's $800 million Series C is less about one startup getting expensive and more about enterprise buyers deciding that open-weight AI is now ready for production.*\n\nTogether AI just put a large number on a simple market shift. On July 1, the San Francisco company announced an $800 million Series C led by Aramco Ventures, with Nvidia, Vista Equity Partners, General Catalyst, Emergence Capital, Salesforce Ventures and others joining. Reuters and TechCrunch both reported the new valuation at $8.3 billion. That is real money chasing a real change in behavior.\n\nThe company was founded in 2022 by Vipul Ved Prakash, Ce Zhang, Chris Ré and Percy Liang. It has built open AI research over the years, including projects such as RedPajama, but the business investors are now funding is the infrastructure layer: GPU cloud, inference, training and access to open-weight models through an API that developers can use without rewriting their stack. That's the bet. Not one model to rule them all, but the plumbing underneath many of them.\n\nTechCrunch reported that Together AI's annual bookings crossed $1.15 billion in its most recent quarter, and that the company says open-source model usage across the industry has tripled in the past year, citing OpenRouter research. Bookings are not the same thing as recognized revenue, so don't blur the two. But bookings at that level still tell you customers are committing budget, not merely testing demos.\n\n## The open-model cost gap is now a budget problem\n\nTogether AI says companies building with open models can see costs that are 6 times to 20 times lower while keeping equal or better performance, and its announcement points to Decagon cutting inference costs sixfold after moving to the platform. That is a number worth sitting with. TechCrunch also reported that Together says its platform can cut inference costs by as much as 60 times in some cases. Even the lower end of that range is enough to get a finance team interested.\n\nFrankly, the cost argument is hard to dismiss. If you're running AI inside a product instead of showing it in a board deck, token prices become margin. A customer support agent or coding assistant that calls a model all day - a research workflow does too - doesn't care whether the model came with the loudest brand name. It cares whether the answer is good enough and the bill doesn't wreck the unit economics.\n\nMcKinsey's 2025 research with the Mozilla Foundation and the Patrick J. McGovern Foundation found that 76 percent of surveyed organizations expected to increase their use of open source AI technologies in the years ahead. The same report said more than half of respondents were already using open-source AI somewhere across data, models or tools. That gives Together AI's funding round a firmer base than a normal venture story. The demand was already visible before the check cleared.\n\nClosed-model providers are not finished. OpenAI, Anthropic and Google still set much of the pace at the frontier, and plenty of enterprises will keep paying for managed systems when reliability, compliance and model quality matter more than cost. But the old assumption that serious AI automatically flows through a closed API is gone. You can now have another conversation.\n\n## Nvidia is backing the layer that buys its chips\n\nNvidia's participation is not a side note. Together AI became an Nvidia Cloud Partner in March 2025, and the company said at the time that it had more than 200 megawatts of data center and power capacity to support large AI workloads. That is the unglamorous detail that matters most. AI infrastructure is not only software. It is power, cooling, networking and racks full of GPUs that need to be kept busy.\n\nThe competitive field is already crowded. Fireworks AI announced a $250 million Series C at a $4 billion valuation in October 2025, then announced a $1.505 billion Series D at a $17.5 billion valuation on July 15, 2026, saying it had surpassed $1 billion in annualized revenue run rate. Runpod announced a $100 million growth investment led by Summit Partners on June 24, reaching a $1 billion valuation and more than one million developers. The market is moving quickly because the need is not subtle.\n\nTogether AI's advantage is that it sits at the intersection of open-model demand and hard infrastructure. Reuters reported that the company plans to expand its computing capacity and infrastructure roughly 50-fold over the next five years. That kind of growth will not be clean. Power commitments are expensive, GPU supply is political - and every big cloud provider has its own answer through products like Amazon Bedrock and Google Vertex AI.\n\nStill, the direction is plain. Together AI was valued at $1.25 billion in March 2024 and $3.3 billion in February 2025, before this $8.3 billion round. Investors are paying up because the model layer is no longer the only place value can collect. The companies that make AI cheap enough to run everywhere may end up with just as much power as the companies that make the models famous.\n\n**Also read:** [Eliyan raises $145 million at a $1 billion valuation as AI's real bottleneck moves between chips](https://startupfortune.com/eliyan-raises-145-million-at-a-1-billion-valuation-as-ais-real-bottleneck-moves-between-chips/) • [NextEra and Brookfield are turning a Cold War uranium site into a $100 billion AI data center](https://startupfortune.com/nextera-and-brookfield-are-turning-a-cold-war-uranium-site-into-a-100-billion-ai-data-center/) • [GrubMarket files for IPO at $4.5 billion valuation as AI rewires the food supply chain](https://startupfortune.com/grubmarket-files-for-ipo-at-45-billion-valuation-as-ai-rewires-the-food-supply-chain/)", "url": "https://wpnews.pro/news/together-ai-hits-8-3-billion-valuation-as-enterprises-flee-closed-ai-models-for", "canonical_source": "https://startupfortune.com/together-ai-hits-83-billion-valuation-as-enterprises-flee-closed-ai-models-for-open-source/", "published_at": "2026-07-29 16:09:37+00:00", "updated_at": "2026-07-29 16:24:16.573393+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-infrastructure", "ai-startups", "ai-products", "generative-ai"], "entities": ["Together AI", "Aramco Ventures", "Nvidia", "Vista Equity Partners", "General Catalyst", "Emergence Capital", "Salesforce Ventures", "OpenRouter"], "alternates": {"html": "https://wpnews.pro/news/together-ai-hits-8-3-billion-valuation-as-enterprises-flee-closed-ai-models-for", "markdown": "https://wpnews.pro/news/together-ai-hits-8-3-billion-valuation-as-enterprises-flee-closed-ai-models-for.md", "text": "https://wpnews.pro/news/together-ai-hits-8-3-billion-valuation-as-enterprises-flee-closed-ai-models-for.txt", "jsonld": "https://wpnews.pro/news/together-ai-hits-8-3-billion-valuation-as-enterprises-flee-closed-ai-models-for.jsonld"}}