Together AI, a neocloud provider that rents GPU clusters and open-source AI infrastructure, has raised an $800 million Series C in a massive Together AI funding round announced on Wednesday, July 1, 2026. The pitch is pretty simple: developers want cheaper, more flexible AI compute than closed-model APIs and hyperscaler queues often give them. Founded in 2022 by Vipul Ved Prakash with Percy Liang and Ce Zhang, the company is betting that open models — not just proprietary ones — will power a lot of production AI.
What is Together AI and how does it work?
Together AI is basically a full-stack platform for running, tuning, and serving open-source models. A developer can start with serverless inference through an OpenAI-compatible API, move heavier traffic onto dedicated endpoints, and then graduate to single-tenant GPU infrastructure when latency, privacy, or throughput matters more than convenience.
That workflow is a big part of the appeal. You can pick a model from Together AI’s model library and call it through standard API tooling. You can also upload your own weights from Hugging Face or S3 for dedicated deployment, then keep everything on the same platform instead of stitching together separate vendors for experimentation and production. For teams building agents, batch jobs, or internal copilots, that continuity saves a lot of ugly platform work.
The customization layer is more than a checkbox. Together AI supports fine-tuning jobs from the command line, including LoRA and preference tuning methods like DPO, RPO, and SimPO. It then lets users deploy those tuned models for inference. It also offers an evaluations API and a sandbox product, so developers can test model behavior and run code in isolated environments without bolting on extra infrastructure.
And if a customer needs raw horsepower, Together AI sells that too. Its GPU clusters are built on Kubernetes and support Slurm-style job scheduling. They use high-speed InfiniBand for multi-node workloads and offer hardware configurations including Nvidia H100, H200, and B200 systems. Clusters can be provisioned in minutes. They can scale in real time and pair with persistent storage — exactly the sort of plumbing most AI teams don't want to build themselves.
Who founded Together AI and why are investors betting $800M?
The founding story
Together AI started in 2022 with a blunt thesis: foundation models were getting centralized because the compute bill was getting absurd, and open alternatives needed their own infrastructure stack. That founding logic still runs through the company now. It isn't just selling cloud capacity. It's selling the idea that open-source AI can be production-grade if someone builds the right rails around it.
Why these founders actually fit the problem
Ved Prakash brings the startup operator résumé. Before Together AI, he founded social media search company Topsy, which Apple bought in 2013 for a reported $200+ million. That matters because Together AI isn't a research project pretending to be a business — it has a CEO who's already built and exited a data-heavy platform company.
Liang is the academic heavyweight in the mix. He's a Stanford computer science professor and director of the Center for Research on Foundation Models, and his work has focused on making foundation models more accessible, understandable, and benchmarked in a rigorous way. If your company thesis is that open models should be usable in the real world, that's strong founder-market fit.
Zhang is the systems builder. He's a Neubauer Associate Professor at the University of Chicago, previously taught at ETH Zurich, and has spent years working on machine learning platforms and the infrastructure bottlenecks behind them. That's the kind of background you'd want if your product sits between model research and ugly production compute.
Traction, fundraising, and the neocloud race
Together AI now has thousands of paying customers, with names including Cursor, Cognition, and Decagon. Annual bookings topped $1.15 billion as of the last quarter. That's the line that makes this round feel less like speculative hype and more like a capacity deal for a business already under serious demand.
Aramco Ventures led the Series C, which came in at $800 million on an $8.3 billion valuation. The round also included Vista Equity Partners, General Catalyst, Emergence Capital, Nvidia, March Capital, Pegatron, SentinelOne’s S Ventures, and others. Before that, Together AI raised a $305 million Series B at a $3.3 billion valuation about 16 months earlier, and a $102.5 million Series A in 2023 led by Kleiner Perkins with Nvidia and Emergence Capital. Back in March 2026, it was seeking $1 billion at a $7.5 billion valuation, so this final outcome looks like less money but a richer price.
Competition is getting crowded fast. Direct rivals include other neocloud and AI infrastructure providers such as CoreWeave, Lambda, TensorWave, Fluidstack, and newer entrants chasing inference-heavy workloads. The older alternative is still the same trio everyone knows — AWS, Google Cloud, and Azure. The other substitute is just paying premium token prices to closed-model providers and living with the margin hit. Together AI’s edge is that it wraps open-model inference, fine-tuning, sandboxing, and GPU clusters into one stack. It sells that on cost, control, and speed rather than raw capacity alone.
Why does the Together AI funding round matter?
This round matters because AI infrastructure is brutally capital-intensive. Software startups can fake it with a small burn and a good demo. GPU clouds can't. If Together AI wants to keep winning enterprise workloads, it needs actual hardware access, real capacity planning, and enough balance-sheet strength to support customers that don't tolerate outages or latency spikes.
That's why the size of this deal stands out. Together AI didn't just raise equity. It also disclosed commitments for more than 500 MW of compute capacity to be capitalized independently by new investors. Pair that with thousands of customers already on the platform, and the message is obvious: investors aren't just betting on open-source AI as an ideology. They're betting that the open-model production stack is becoming a real infrastructure business.
There's also a timing angle here. If open models keep improving, the vendor that can deliver solid inference economics and custom deployment without forcing companies into a closed ecosystem becomes a pretty attractive choke point.
How big is the AI infrastructure market behind Together AI?
The macro numbers are huge, and frankly a little wild. Grand View Research pegs the global AI data center market at $147.3 billion in 2025, with an estimate of $180.6 billion for 2026 and a jump to $810.6 billion by 2033. That implies a 23.9% compound annual growth rate. It helps explain why investors keep pouring giant checks into compute suppliers instead of only into model labs and app startups.
North America held the largest share of that market in 2025 at 37.5%, which lines up with where a lot of enterprise AI demand is landing first. The structural shift is broader than data center buildout alone. Together AI says open-source model usage across the industry tripled over the last year, echoing a change in buyer behavior: companies still want top-tier model quality, but they also want lower inference costs, customization, and less vendor lock-in than closed APIs usually offer.
Final take on Together AI funding
The easy read is that Together AI just raised a lot of money because AI is hot.
The better read is that this Together AI funding round is a bet on a specific future: one where enterprises run a mix of open and custom models, care obsessively about inference cost, and don't want their entire AI stack locked inside one provider. If Together AI can turn its bookings into durable margins while keeping GPU supply ahead of demand, it could end up being a lot more important than just another cloud startup.
Read how Dovetail Capital raised ₹100 crore in a Series A led by Elev8 Venture Partners to expand its asset servicing platform, strengthen global operations, and simplify fund administration, compliance, and cross-border investing for institutional investors.
FAQ
- What is the latest Together AI funding round? Together AI raised an $800 million Series C announced on July 1, 2026, at an $8.3 billion valuation. Aramco Ventures led the round, and the investor list included names such as Vista Equity Partners, General Catalyst, Emergence Capital, Nvidia, March Capital, and Pegatron.
- How does Together AI work for AI developers? It gives developers one platform to run open-source models, fine-tune them, evaluate them, and deploy them on shared or dedicated infrastructure. A team can start with API-based inference, then move to custom endpoints, sandboxed development environments, or full GPU clusters as workloads get bigger and more sensitive.
- Who founded Together AI? Together AI was founded in 2022 by Vipul Ved Prakash, Percy Liang, and Ce Zhang, with Ved Prakash serving as CEO and Zhang as CTO. Ved Prakash previously sold Topsy to Apple, Liang is a Stanford professor who leads CRFM, and Zhang is a University of Chicago computer science professor with deep systems and ML infrastructure expertise.
- Is Together AI a cloud company or an AI startup? It's really both, but the cleaner label is an AI neocloud. Together AI sells the underlying compute, inference, tuning, and deployment stack for open-source and custom models, which puts it closer to infrastructure than to a typical application-layer AI startup.
