Together AI Raises $800 Million at $8.3 Billion Valuation to Lead the Open Source AI Cloud

The shift from closed frontier models to open source alternatives has been building for two years, and it is now moving fast enough to attract very serious capital. Together AI, the neocloud platform built specifically to host and serve open source AI models, has raised 800 million dollars in a Series C round that values the company at 8.3 billion dollars. The announcement came on July 1, 2026 and represents one of the largest single raises in AI infrastructure this year.

The round was led by Aramco Ventures, the venture arm of Saudi Aramco, which had already backed the company at its Series B through Prosperity7. Vista Equity Partners, General Catalyst, Emergence Capital, Nvidia, March Capital, Pegatron, and SentinelOne's S Ventures all participated alongside the lead. Nvidia's continued presence on the cap table after backing the company at Series A is a signal worth noting: the world's dominant GPU maker has a clear financial interest in seeing the compute‑intensive open source market grow.

A Valuation That Has More Than Doubled in 16 Months

Together AI's trajectory as a business makes the valuation jump easier to understand. The company raised a 305 million dollar Series B at a 3.3 billion dollar valuation in February 2025, led by General Catalyst and Prosperity7. Before that, its 102.5 million dollar Series A in November 2023 was led by Kleiner Perkins, again with Nvidia and Emergence Capital participating. This Series C more than doubles the Series B valuation in just over a year, and it comes backed by actual revenue performance rather than speculative positioning.

The company says its annual bookings crossed 1.15 billion dollars as of its most recent quarter. Thousands of paying customers are now on the platform, with Cursor, Cognition, and Decagon among the names the company has cited publicly. In March, The Information had reported that Together AI was seeking one billion dollars in funding at a 7.5 billion dollar valuation, which suggests the company's commercial progress in the intervening months pushed both the raise target and the valuation higher than originally planned.

What Together AI Actually Offers and Why It Is Winning

Neoclouds, as a category, rent GPU compute to developers and companies building AI applications. Most of them are essentially compute rental businesses, marketplaces for Nvidia GPU time that exist because hyperscalers like AWS, Google Cloud, and Azure cannot keep up with demand. Together AI is doing something more differentiated. It pairs the compute with its own inference optimization software, which it says can reduce the cost of running popular open source models by as much as 80 percent. That software layer is the moat it is building around what would otherwise be a commoditized infrastructure business.

The underlying market dynamic driving Together AI's growth is the rapid adoption of open source models by companies that previously assumed they would need to rely entirely on closed frontier APIs from OpenAI or Anthropic for production workloads. Open source models have improved dramatically in capability, and for a wide range of use cases they now perform well enough to be the economically rational choice. Open source model usage on Together AI's platform has tripled in the past year, a figure the company supports with data from OpenRouter, an AI gateway that tracks cross‑platform model usage trends.

Co‑founder and CEO Vipul Ved Prakash framed the company's mission as making intelligence abundant rather than expensive, a positioning that resonates directly with enterprise buyers watching their AI inference costs compound every quarter. Prakash is a serial entrepreneur who previously founded and sold Topsy, a social network search platform, to Apple for more than 200 million dollars. His co‑founders bring academic credibility that balances the commercial emphasis: Percy Liang is a professor at Stanford who leads the Center for Research on Foundation Models, and Ce Zhang is an associate professor at ETH Zürich with a research focus on systems for machine learning.

The Capital Goes Toward Capacity, Not Just Operations

Together AI has secured commitments for more than 500 megawatts of compute capacity from investors as part of this round, a detail that sets this raise apart from a typical growth‑stage funding event. The 500 megawatt commitment is designed to fund roughly 50‑fold capacity growth over five years, which signals that the company is not treating this as runway capital but as the foundation for a significant expansion in physical infrastructure. That scale of capacity build requires the kind of long‑term, patient capital that sovereign‑backed funds like Aramco Ventures are better positioned to provide than most traditional venture firms.

The compute capacity question is the central constraint for every company in this category right now. Demand for GPU infrastructure continues to outrun available supply, and neoclouds that can secure reliable, large‑scale access to compute are in a structurally better position than those that depend on spot purchasing or shorter‑term contracts. Together AI's investor relationships with Nvidia and the TSMC‑linked investors that have backed adjacent companies give it stronger supply‑side positioning than most of its category peers.

A Category Heating Up Fast

Together AI is operating in a neocloud market that has attracted significant capital in recent months. Upscale AI closed a 500 million dollar raise at a 2 billion dollar valuation. TensorWave, which focuses specifically on AMD GPU clusters as an alternative to Nvidia‑dependent infrastructure, raised a 350 million dollar Series B at a 1.55 billion dollar valuation. Groq and RunPod are competing for similar developer and enterprise demand, and hyperscalers are not standing still either, with AWS, Google, and Microsoft all investing in AI‑native infrastructure improvements of their own.

What distinguishes Together AI from the pack is its explicit bet on the open source layer, not just the hardware beneath it. The inference software and the developer tooling it has built around open source models give it a different relationship with its customers than a pure compute rental business would have. Developers who optimize their workflows around Together AI's inference engine and tooling face meaningful switching costs, which translates into the kind of retention profile that justifies the valuation multiple the market is now assigning to the business.

The 8.3 billion dollar valuation also makes Together AI a credible IPO candidate over the next 18 to 24 months, assuming the revenue trajectory continues. With annual bookings already past 1.15 billion dollars and the infrastructure capacity to grow significantly from here, the company is reaching the scale at which public market investors tend to start paying close attention. Whether Together AI chooses that path or continues to grow privately under its current investor base is the next strategic question its leadership will have to answer.