- SeriesCUSD 800M — Amount: USD 800M; Post-money valuation: USD 8.3B; Led by: Aramco Ventures (Saudi Aramco's venture capital arm) | Valuation rose from USD 3.3B to USD 8.3B, a 2.5x increase in 16 months. Notably, in March 2026, The Information reported the company was negotiating to raise approximately USD 1B at a pre-money valuation of USD 7.5B—ultimately closing at a post-money valuation of USD 8.3B, indicating that commercial progress during the interim (annualized bookings exceeding USD 1.15B) drove up the fundraising scale and valuation. ⚠️ The 500MW compute commitment is the core of this round. Investors committed to more than 500 megawatts of compute capacity, which is capitalized separately from the USD 800M equity, to support approximately 50x compute expansion over the next five years. For inference clouds, supply guarantees are scarcer than cash—the reason large enterprises...
- SeriesBUSD 300M — Amount: USD 305M; Post-money valuation: USD 3.3B; Led by General Catalyst (lead) + Prosperity7 Ventures (co-lead). Valuation rose from USD 1.25B (2024) to USD 3.3B — a 2.6x increase in 14 months. The same month, the company launched its GPU Cluster product, running NVIDIA Blackwell B200 and GB200 hardware. NVIDIA has consistently appeared in seed, Series A, Series A extension, and Series B rounds, making it Together's most frequent strategic investor across rounds. This model is identical to xAI, CoreWeave, Perplexity, and Cohere: locking in customers at the equity level for future inference compute demand. The 2024 infrastructure...
- Series A ExtensionUSD 100M — Amount: not separately disclosed (bringing cumulative financing over USD 100M); not separately disclosed; led by Prosperity7 Ventures (under Saudi Aramco) | NVIDIA's first entry marked Together being recognized as a key inference layer partner. Prosperity7 Ventures (under Saudi Aramco), entering simultaneously, later co-led Series B, with Series C led by Aramco Ventures—Gulf capital increased stakes across three rounds, representing the clearest main thread in this page's capital structure. Abhishek Shukla of Prosperity7 once stated their logic: 'Building AI infrastructure over the next decade will be the largest infrastructure project in human history'—this was a bet on compute capacity.
- SeriesAUSD 100M — Amount: USD 102.5M; Post-money valuation: Not officially disclosed; Led by: Kleiner Perkins | In April 2023, the company released RedPajama-V1 (a 1.2 trillion token open-source corpus replicating LLaMA training data), and closed its Series A the following month. In October 2023, it released RedPajama-Data-V2, expanding to 30 trillion tokens of multilingual web data—then one of the largest publicly available pre-training datasets globally. That same month (October 2023), Tri Dao (inventor of FlashAttention) officially joined as Co-founder and Chief Scientist. After graduating with a PhD from Stanford, he chose to join Together rather than founding an independent lab or joining a hyperscale cloud provider, reflecting the company's research...
- Seed RoundUSD 20M — Amount: Approx. USD 20M; Led by Lux Capital. Seed funding was used for the first version of the inference API platform and early research agenda, including the RedPajama open-source training dataset project—collaborating with ETH DS3Lab, Stanford CRFM, Hazy Research, and MILA to produce 1.2 trillion tokens replicating LLaMA pre-training data. This open-source contribution was Together's key strategic move: it established credibility as a genuine contributor to the open-source ecosystem rather than just a commercial consumer, forming the basis for attracting strategic investors like NVIDIA later. Source: benched.ai / Sacra
Physix Frontier