Together AI $800 million round doubles value as Aramco leads

Together AI, a five-year-old firm that rents out clusters of Nvidia chips so businesses can run open-source artificial intelligence models, announced on July 1, 2026, that it had closed an $800 million Series C funding round at an $8.3 billion post-money valuation. The Together AI $800 million round instantly makes the company one of the most valuable private players in enterprise AI infrastructure, even though most consumers have never heard of it.

The deal is notable not only for its size but for who wrote the largest check. The round was led by Aramco Ventures, the investment arm of Saudi Aramco, the state-controlled oil giant, signaling how deeply Persian Gulf capital has moved into the plumbing of the American AI economy. In roughly 16 months, Together AI has more than doubled its valuation, a pace of appreciation that underscores how quickly money is chasing the companies that supply computing power to the AI boom rather than the consumer-facing chatbots that dominate headlines.

Together AI $800 million round

The financing brought a crowded roster of investors to the table. Alongside lead investor Aramco Ventures, the round drew participation from Vista Equity Partners, General Catalyst, Emergence Capital, chipmaker Nvidia, March Capital, Taiwanese electronics manufacturer Pegatron, and S Ventures, the venture arm of cybersecurity company SentinelOne. The mix of financial sponsors, a strategic chip supplier, and a hardware manufacturer reflects the various interests converging on AI infrastructure.

With this financing, Together AI's total funding since its 2022 founding reaches roughly $1.3 billion. That is a substantial war chest for a company that operates in a capital-intensive corner of the industry, where growth depends on the ability to buy or lease enormous quantities of expensive graphics processing units. The presence of Nvidia as an investor is telling, since the chipmaker has strong incentives to seed the ecosystem of firms that turn around and purchase its hardware.

The round arrives after a period of speculation about the company's fundraising plans. Earlier reports in March 2026, via The Information, suggested Together AI was seeking $1 billion at a $7.5 billion valuation. The round that ultimately closed was smaller in raw dollar terms but landed at a higher valuation than those earlier reports had indicated, a sign that investor demand pushed the price up even as the company took in less cash than first floated.

Neocloud model for renting GPU clusters

Together AI belongs to a category that industry insiders have taken to calling the "neocloud." Rather than building the sprawling, general-purpose data center empires of Amazon Web Services or Microsoft Azure, neoclouds specialize narrowly in renting out Nvidia GPU clusters optimized for training and running AI models. The pitch to customers is speed, specialization, and cost.

The company's specific niche is open-source AI. Together AI's platform lets businesses train and run workloads on freely available models such as DeepSeek, MiniMax, Kimi, and Nvidia's Nemotron. For many enterprises, running an open model on rented infrastructure costs meaningfully less than paying per-token fees to closed systems from OpenAI or Anthropic. That cost gap is the engine of Together AI's growth.

By handling the technical complexity of provisioning and orchestrating GPU clusters, Together AI positions itself as a middle layer between the raw hardware and the companies that want to deploy AI without becoming infrastructure experts themselves. It is a business model that thrives precisely when open-source models improve to the point where they rival proprietary alternatives, and that is exactly what has been happening.

Enterprise economics behind open model adoption

The core thesis behind Together AI's rising valuation is a shift in how businesses think about AI spending. For the first wave of generative AI adoption, companies reached for the most capable closed models, largely from OpenAI and Anthropic, and paid a premium for them. As open-source models have narrowed the quality gap, the calculus has changed.

According to figures the company shared, industry-wide usage of open-source AI models has roughly tripled over the past 12 months. That is a striking rate of adoption, and it maps directly onto Together AI's own momentum. The company said its annualized bookings crossed $1.15 billion last quarter, a figure that suggests the neocloud is capturing real revenue and not merely paper valuation.

For finance chiefs scrutinizing AI budgets, the appeal is straightforward. An open model running on rented GPUs can deliver comparable results for a fraction of the ongoing cost of premium closed APIs, particularly at high volumes. As inference workloads scale, small per-query savings compound into large numbers, and that arithmetic is pulling more enterprise traffic toward platforms like Together AI.

Cursor and Decagon as customer proof points

A funding announcement is only as convincing as the customers behind it. Together AI pointed to companies including Cursor, the fast-growing AI coding tool, and Decagon, a customer-service automation startup, as examples of clients using its platform to move workloads onto open models.

Those references matter because both Cursor and Decagon are themselves high-profile, heavily used AI products. When companies of that profile shift portions of their AI workloads to open models running on a neocloud, it validates the argument that open-source infrastructure is ready for demanding, production-grade use rather than experimentation alone.

The presence of such customers also helps explain investor enthusiasm. Backers are not simply betting on a thesis about open-source economics in the abstract; they are looking at a company that has already signed the kind of sophisticated clients whose engineering teams could build their own infrastructure but choose to rent Together AI's instead.

Fifty-fold infrastructure expansion plan

Together AI has laid out an aggressive plan for the new capital. The company said it intends to expand its inference product line and scale its computing infrastructure roughly 50-fold over the next five years. That is an enormous ambition, and it reflects a wager that inference, the process of actually running trained models to generate answers, will become the dominant cost center in enterprise AI.

For much of the AI boom, attention centered on training, the compute-hungry process of building models in the first place. But as models mature and deployment spreads, the recurring cost of inference at scale increasingly dwarfs one-time training expenses. Together AI is positioning itself for that shift, betting that the future of AI spending lies in serving predictions cheaply and continuously.

Scaling infrastructure 50 times over five years is not cheap, which helps explain both the size of this raise and the company's openness to raising more. Chief executive Vipul Ved Prakash has indicated the company may seek additional funding later in 2026 as inference demand grows, suggesting the $800 million is a milestone rather than an endpoint in Together AI's capital needs.

Founders and the funding trajectory since 2022

Together AI was founded in 2022 by chief executive Vipul Ved Prakash, Stanford professor Percy Liang, and researcher Ce Zhang. The founding team blended entrepreneurial and academic pedigree, pairing Prakash's operating experience with Liang's standing as one of the more prominent academic voices in machine learning.

The valuation history charts a steep climb. A March 2024 round valued the company at $1.25 billion. By February 2025, a $305 million Series B had more than doubled that figure to $3.3 billion. Now, roughly 16 months later, the Series C has more than doubled the company again to $8.3 billion. Each step has compressed the previous one, a compounding pattern that speaks to how rapidly the market has re-rated infrastructure providers.

That trajectory has also put an exit on the horizon. Prakash has not ruled out an initial public offering as soon as 2027, which would make Together AI one of the first pure-play neoclouds to test public markets. For a company that was a research-flavored startup just a few years ago, a 2027 IPO timeline would represent a remarkably fast maturation.

Saudi capital in the vendor-financed AI buildout

Perhaps the most consequential detail of the Together AI $800 million round is the identity of its lead investor. Aramco Ventures, tied to the Saudi state oil company, now sits atop the cap table of a critical American AI infrastructure provider. That reflects a broader pattern of Gulf sovereign and quasi-sovereign wealth flowing into the AI supply chain, from chips to data centers to the neoclouds that stitch them together.

Observers have described the current phase of the AI buildout as increasingly "vendor-financed," a dynamic in which the suppliers and beneficiaries of AI spending help fund the very companies that will buy from them. Nvidia's participation as an investor in a firm that purchases Nvidia GPUs is one example; oil-rich sovereign capital seeking exposure to the AI economy is another layer of the same phenomenon.

That structure raises questions worth watching as the sector matures. When the capital funding an infrastructure boom is intertwined with the vendors and geopolitical actors that benefit from it, the line between organic demand and financed demand can blur. For now, Together AI's booking figures and marquee customers suggest genuine underlying growth. But the composition of its investor base is a reminder that the economics of the AI infrastructure race are as much about who is willing to finance the buildout as about who is using the technology.

Whether Together AI can execute a 50-fold expansion, sustain its booking momentum, and reach public markets by 2027 will determine if the $8.3 billion valuation proves prescient or premature. What is already clear is that the money underwriting the AI era is increasingly concentrated in the companies that own the compute, and Together AI has just made itself one of the largest among them.