AI Tools & SaaS

Together AI Raises $800 Million Betting on Cheaper, Not Better

Finn · The Tech Rundown ·

Together AI raised an $800 million Series C at an $8.3 billion valuation, led by Aramco Ventures with participation from Vista Equity, General Catalyst, Nvidia, and Salesforce Ventures, according to the company's own release and coverage from PYMNTS and Data Center Dynamics. The round more than doubles the company's prior $3.3 billion valuation from a February 2025 raise, bringing total funding to $1.3 billion.

Together AI's product is infrastructure for running open-source models, DeepSeek, MiniMax, and Kimi among them, more cheaply than businesses could get equivalent capability from closed frontier APIs like OpenAI's or Anthropic's. The company reports annual bookings topped $1.15 billion last quarter, and says open-source model usage has roughly tripled industry-wide over the past twelve months.

The bet is on a specific kind of customer, not a better model

Together AI isn't trying to out-build OpenAI or Anthropic on capability. Its pitch is that a large and growing set of enterprise workloads don't need frontier-model performance, they need adequate performance at a fraction of the cost, and that gap is where an $8.3 billion company can be built without ever training a state-of-the-art model from scratch.

That's a meaningfully different business than the one most AI headlines are about. The frontier labs are locked in a spending race measured in hundreds of billions of dollars a year on the theory that being best matters more than being cheap. Together AI's valuation jump is a signal that investors think the opposite bet, being reliably cheaper on models that are merely good enough, is also a real, fundable business, and possibly a less capital-intensive one, since it doesn't require Together AI to fund its own frontier training runs.

Whether that thesis holds depends on whether the gap between "frontier" and "good enough" keeps being wide enough for price to be the deciding factor, or whether open-source models close that gap so completely that the closed labs' price premium stops mattering anyway. Either way, more than tripling open-source usage industry-wide in a year is a fast enough shift that Together AI's bet looks better funded than speculative at this point.

Sources: Businesswire · PYMNTS · Data Center Dynamics