Together AI closed an $800 million Series C round on July 1, 2026, valuing the company at $8.3 billion. Aramco Ventures led the round, with NVIDIA, Vista Equity Partners, General Catalyst, and Emergence Capital joining. The company’s valuation has grown from $3.3 billion at the start of 2025 — a 2.5x increase in 18 months.

The headline number matters less than what it signals. Open-source AI infrastructure has become a serious enterprise bet, not a scrappy alternative to OpenAI and Anthropic.

What Together AI Does

Together AI runs a cloud platform that lets businesses train and deploy AI on open-source models — Llama, Mistral, Qwen, and others — at significantly lower cost than using proprietary closed systems. The pitch is straightforward: you get similar capability without the lock-in, and you keep control over your data and model weights.

Their paying customers now number in the thousands and include AI-native companies like Cursor, Cognition, and Decagon — precisely the class of businesses building the next wave of enterprise software. Annual bookings crossed $1.15 billion in the most recent quarter.

The new capital comes with something concrete: investors have committed over 500 megawatts of compute capacity to Together AI’s infrastructure buildout, which the company says will support roughly 50-fold capacity growth over the next five years. That’s not a vision slide — it’s physical infrastructure backing.

Why an Oil Giant Led the Round

Aramco Ventures leading this round is the detail most enterprise leaders should sit with. Saudi Aramco is not a typical AI investor. What they are is one of the largest industrial enterprises on the planet, with enormous compute-intensive workloads and a deep strategic interest in not being wholly dependent on US-headquartered AI providers.

That calculus applies well beyond energy. Regulated industries, government-adjacent businesses, and multinationals operating across jurisdictions are all navigating similar questions about AI supply chain concentration. Together AI’s open-source model gives them an option.

NVIDIA joining the round is also notable — they have a vested interest in workloads running on GPU infrastructure they build, and an open-source AI platform at scale drives exactly that demand.

The Closed vs. Open Divide Is Widening

The market for enterprise AI is splitting into two lanes:

Closed providers (OpenAI, Anthropic, Google) offer the highest-performing models, managed APIs, and tight integrations with enterprise software. You pay more per token and accept their roadmap.

Open-source infrastructure (Together AI, plus cloud-hosted alternatives from AWS and Azure) offer flexibility, lower costs, data control, and the ability to fine-tune models on your own data. The capability gap versus closed providers has narrowed substantially.

Until recently, most enterprises defaulted to closed providers for reliability reasons. That’s changing. Together AI’s $1.15B+ in annual bookings tells you this isn’t a fringe choice anymore.

What This Means for Business

If your AI spend is growing and most of it goes to closed providers, this round is worth paying attention to. The open-source infrastructure market is maturing fast enough that alternatives are worth evaluating on their own merits, not just as cost-cutting measures.

A few considerations for businesses thinking through their AI architecture:

Data control. If your AI use cases involve proprietary data — customer records, internal documents, financial models — where that data goes during inference matters. Open-source deployment on your own or a specialist cloud gives you more control than sending data to a shared API.

Cost trajectory. Together AI’s pricing advantage exists today, and the infrastructure investment suggests it’s built to last. The 500MW compute commitment is a signal that they expect to compete on cost at scale for years, not quarters.

Capability parity. Open-source models have closed most of the gap with proprietary models on structured business tasks. For workflows like document extraction, classification, summarization, and data transformation, you don’t need a frontier closed model. An open-source model running on purpose-built infrastructure often does the same job at a fraction of the cost.

The vendor concentration risk is real. The Anthropic Fable 5 export restrictions earlier this year were a reminder that geopolitical dynamics can affect AI supply chains. Diversification isn’t paranoia — it’s planning.

None of this means every business should abandon managed APIs. For most small and mid-sized businesses, the simplicity of a managed API still wins. But if you’re running AI at scale or in a regulated context, Together AI’s growth is a signal that the open-source route is production-ready in a way it wasn’t two years ago.

---

If you’re deciding where to start with agents, start here. The free Working With Claude field guide walks through the ecosystem, Claude Code, and a real rollout plan. Get your copy.