Together AI reaches $8.3 billion valuation as enterprises shift from closed models to open-source AI infrastructure, validating the open-source alternative narrative.
Together AI quantified a market shift in July with an $800 million Series C led by Aramco Ventures, with participation from Nvidia, Vista Equity Partners, General Catalyst, Emergence Capital, Salesforce Ventures and others. Reuters and TechCrunch reported the new valuation at $8.3 billion, reflecting real capital chasing a demonstrable change in market behavior.
The company was founded in 2022 by Vipul Ved Prakash, Ce Zhang, Chris Ré and Percy Liang. After building open AI research projects including RedPajama, investors are now backing its core business: infrastructure for GPU cloud services, inference, training and access to open-weight models through an API that developers can integrate without rewriting their systems. The bet is not one dominant model, but the underlying plumbing for many.
TechCrunch reported that Together AI's annual bookings crossed $1.15 billion in its most recent quarter, with the company citing OpenRouter research showing open-source model usage has tripled across the industry in the past year. Bookings differ from recognized revenue, but at that level they indicate customers are committing budget, not merely evaluating demos.
Together AI claims companies using open models can achieve cost reductions of 6 times to 20 times while maintaining equivalent or better performance. The company pointed to Decagon cutting inference costs sixfold after adopting the platform, and reported cases where inference costs drop by as much as 60 times. Even at the lower end, these savings are enough to capture CFO attention.
The cost argument is difficult to dispute. In production environments, token prices directly affect margins. A customer support agent or coding assistant that makes repeated model calls—as do many research workflows—cares whether answers meet requirements and whether bills remain economical. Brand prominence matters less than unit economics.
McKinsey's 2025 research with the Mozilla Foundation and the Patrick J. McGovern Foundation found that 76 percent of surveyed organizations expected to increase their use of open-source AI technologies. More than half were already deploying open-source AI in some capacity across data, models or tools. This demand preceded the funding announcement, giving Together AI's round a foundation beyond typical venture narratives.
Closed-model providers remain formidable. OpenAI, Anthropic and Google continue to lead at the frontier, and many enterprises will continue paying for managed systems when reliability, compliance and model quality outweigh cost considerations. The assumption that serious AI deployment requires closed APIs has shifted, however, opening new market dynamics.
Nvidia's participation carries weight. Together AI became an Nvidia Cloud Partner in March 2025 with more than 200 megawatts of data center and power capacity for large AI workloads—a critical operational foundation. AI infrastructure requires more than software: power, cooling, networking and densely packed GPUs demand constant utilization.
The competitive landscape is dense. Fireworks AI raised a $250 million Series C at a $4 billion valuation in October 2025, followed by a $1.505 billion Series D at $17.5 billion valuation on July 15, 2026, claiming over $1 billion in annualized revenue run rate. Runpod secured a $100 million growth investment from Summit Partners on June 24, reaching a $1 billion valuation and attracting over one million developers. Market velocity reflects acute demand.
Together AI's core advantage lies at the intersection of open-model adoption and infrastructure capacity. Reuters reported plans to expand computing capacity roughly 50-fold over five years—growth that will prove operationally complex. Power commitments are expensive, GPU supply remains constrained, and incumbent cloud providers (Amazon with Bedrock, Google with Vertex AI) now compete directly.
The trajectory is clear. Together AI valued at $1.25 billion in March 2024, $3.3 billion in February 2025, and $8.3 billion in this round, capturing investor conviction that value now accrues beyond the model layer. The companies making AI economical to run everywhere may accumulate as much power as those making models famous.