Together AI $1B Series C Marks Infrastructure Shift

How Fast Is Together AI Growing?

From $100 million annualized revenue at the start of 2025 to $300 million by September—a tripling in nine months. That's the threshold where venture investors stop reading metrics and start writing checks.

Together AI's trajectory reads like a compression of what usually spans three or four funding rounds. Founded in 2022 by CEO Vipul Ved Prakash, the company started by providing developers access to Nvidia GPUs and open-source AI models. It now positions itself as the "AI Native Cloud," offering production-scale inference, pre-training, and model development. That shift from commodity access to integrated platform is what turned $100M revenue into $300M in the span of nine months.

The funding round reflects this momentum. The previous Series B—$305 million at a $3.3 billion valuation—closed just over a year ago. This new $1 billion Series C values the company at $7.5 billion, according to The Information. That's not a normal step function. That's accelerating returns on invested capital.

When a company can raise 3.3x the capital at 2.3x the valuation in just 13 months, it signals one thing: the market is not skeptical. It's convinced. Investors watched Together AI move from Series B to Series C and saw not a scaling company but a foundational layer. The 10x year-over-year annual contract value growth confirms it—that's the kind of expansion that happens when you're solving a problem nobody expected would compress this fast.

Why Enterprises Like Salesforce and Zoom Chose This Infrastructure

Enterprise adoption tells you more than growth rate alone. Together AI serves over 1 million developers and thousands of customers, with annual contract value growing 10x year over year.

Here's what matters: the company has closed 27 deals exceeding $1 million. And one exceeding $1 billion. That kind of deal concentration—where three to seven enterprise contracts carry the weight of thousands of smaller ones—signals platform density. This is the infrastructure layer everyone discovered they actually needed.

Cursor, Decagon, and Cartesia count as core developer-facing customers. But the enterprise roster matters more: Salesforce, Zoom, The Washington Post. These aren't startups shopping for cheaper-than-OpenAI inference. They're public companies with legacy systems embedding AI at scale. They wouldn't migrate infrastructure unless Together had proven it cheaper, faster, and less disruptive than alternatives.

For context, most enterprise AI projects fail because infrastructure decisions compound. Teams pick a platform, bake it into production workloads, and then find themselves stuck. When Salesforce commits a seven-figure annual contract, they've already spent months validating inference reliability and cost-per-token at production scale. The deal value includes not just commitment but institutional conviction.

Together AI's research lab, led by Chief Scientist Tri Dao (creator of FlashAttention—a foundational optimization for inference speed), translates research into deployment advantage. CEO Prakash noted at AI Native Conf: "The same researchers who contribute foundational studies are the ones implementing them in the production systems relied upon by our clients." That's rare. Most companies separate research from operations. Together fused them.

How Much Did Together AI Grow in One Year?

Together AI's valuation jumped from $3.3 billion to $7.5 billion in just 13 months, while revenue tripled from roughly $100 million to $300 million annualized.

Why Did Infrastructure Companies Suddenly Become Valuable?

Capital finally distributed across the AI stack in 2026—investors moved beyond betting on just model labs or end-user apps to funding the infrastructure layer that runs all AI workloads.

Together AI's $1 billion raise is smaller than Anthropic's $30 billion or xAI's $20 billion mega-rounds. But Anthropic and xAI are foundation model companies, building the base models that infrastructure sits on top of.

Together's round signals something different: investor capital has finally distributed across the AI stack. For most of 2023 and 2024, venture money clustered at the top (model labs) and bottom (applications). Infrastructure—the dangerous middle layer of chips, platforms, and orchestration—was underfunded relative to its strategic importance.

The $305 million Series B in April 2024 hinted at changing appetite. This $1 billion round confirms it. With Anthropic and xAI securing mega-scale funding, investors recognized that foundation models alone don't extract value. The companies that run inference efficiently, manage training workloads, and keep enterprise customers' models compliant and fast become the compounders. Together AI's valuation jump reflects that shift in perception.

What Makes Infrastructure Companies Different from Models or Apps?

Infrastructure companies control more economic leverage than either models or applications because they become embedded and repeatable across multiple use cases.

A foundation model (like Claude or Grok) is singular—you either use it or you don't. An application (like ChatGPT) is end-user-facing, but vulnerable to switching if a better app emerges. Infrastructure, by contrast, becomes critical as the cloud provider itself.

Together AI achieved this by combining three things: first, it built the platform (not just a service). Second, it embedded research directly in operations (FlashAttention in production). Third, it diversified its customer base across both developers (who try it first) and enterprises (who bet their revenue on it). That combination—platform + research + diversification—is exactly what venture capitalists bet on in growth-stage infrastructure rounds.

The $1B Series C validates that these bets are paying off. With revenue tripling year-over-year and customer deals stacking up, the company has moved from "promising startup" to "foundational infrastructure company" in the investor mental model. That distinction justifies the valuation jump.