Together AI Valued at $8.3 Billion After $800 Million Round Led by Aramco Ventures

Together AI, the company that rents out clusters of Nvidia graphics cards and infrastructure for training and running open language models, has closed an $800 million Series C funding round. The round values the company at $8.3 billion, more than double the $3.3 billion valuation it carried just 16 months ago.

The company positions itself as the infrastructure layer for enterprises moving away from closed models such as those from OpenAI or Anthropic toward open-source alternatives. Together AI lets customers train and run models like DeepSeek, Nemotron, MiniMax, and Kimi at significantly lower cost than closed systems, while matching or beating their performance in many use cases.

Why investors are betting on open models

Aramco Ventures leading the round is notable in its own right: it is the venture arm tied to one of the world's largest oil companies, underscoring how far capital from outside traditional Silicon Valley circles has moved into AI infrastructure. Nvidia's participation as an investor further ties the chipmaker to a company that is one of the biggest renters of its GPUs.

Annual bookings climbing past $1.15 billion suggests demand for inference on open-source models is growing faster than for services built on closed models. For many companies, the cost per query on an open model running on rented infrastructure can be substantially lower than the per-token fee charged by APIs like GPT or Claude, which adds up to real budget savings at a scale of millions of queries a month.

Rising competition among neoclouds

Together AI competes in the so-called neocloud segment, providers that specialize exclusively in infrastructure for training and running AI models, as distinct from traditional clouds like AWS, Azure, or Google Cloud. CoreWeave, Lambda, and Crusoe also operate in this space, and each new mega-round raises the valuation bar for the whole category.

The size of the round also shows that investors treat access to GPU compute as a separate, high-margin business, independent of which particular language model happens to be winning the benchmark race at any given moment. Capital is flowing into the infrastructure layer regardless of whether customers choose open or closed models, as long as they are using cloud GPUs at all.

What it means for companies in Poland

For Polish companies weighing AI deployments, Together AI's round signals that cheaper infrastructure offerings for open models will keep expanding, which could lower the entry cost for smaller firms unwilling to pay per token for closed models. At the same time, growing competition in the neocloud segment means more options when choosing an infrastructure provider for teams building their own large language model based solutions.

Three weeks after the round was announced, the open question is whether revenue growth will hold up enough to justify the $8.3 billion valuation if demand for compute cools in the second half of the year.

Sources: TechCrunch (techcrunch.com), Business Wire (businesswire.com), Tech Funding News (techfundingnews.com)