Together AI closed a $106 million funding round led by Salesforce Ventures. The round values the two-year-old startup at $1.25 billion and pushes total disclosed backing past $228 million. The company had raised a $102.5 million Series A in November 2023.
Coatue Management, Lux Capital, Emergence Capital, and Nvidia joined the round. Together AI will direct the fresh capital toward expanding its cloud platform, which trains and deploys open-source models. The platform competes with hyperscale providers such as Amazon Web Services and Microsoft Azure, but it specializes in running models no single vendor owns.
CEO Vipul Ved Prakash, a former Apple engineer who co-founded the messaging platform Kik, leads the company. Nvidia's presence on the cap table links the investor to the hardware inside Together AI's compute clusters: Nvidia supplies the GPUs that power the platform.
The Open-Source Infrastructure Pitch
Together AI operates a cloud platform where developers and enterprises train, fine-tune, and run open-source generative models such as Llama, Mistral, and Stable Diffusion. Because the platform targets models with publicly available weights and architecture, it appeals to teams that need to customize models on their own data without being locked into a closed API like OpenAI's.
Valuation and Funding Trajectory
Together AI reached a $1.25 billion valuation just eight months after its Series A, a sharp mark-up that signals strong demand for its services. The company did not disclose revenue. Still, the rapid succession of rounds arrived in a market where competitors such as CoreWeave, Lambda, and RunPod also raised large sums to build the compute layer for generative AI.
Nvidia's participation stands out. The chipmaker has invested in several GPU-dependent startups, including CoreWeave and Inflection AI, to widen the ecosystem around its hardware. Its involvement suggests that Together AI's platform is optimized for Nvidia silicon, which dominates AI workloads.
Competitive Landscape
Together AI competes in a crowded compute market. The largest players remain the hyperscale trio: Amazon Web Services, Microsoft Azure, and Google Cloud. A new tier of specialized AI cloud providers has emerged alongside them. CoreWeave raised $1.1 billion in debt and equity during 2023. Lambda pulled in $500 million in early 2024. Both offer GPU clusters that undercut or outmaneuver the big three on cost and flexibility.
Together AI sets itself apart by staying exclusive to open-source models. It bundles managed training jobs, model hosting, and collaboration tools. The company has styled itself as a compute-first platform for the open-source AI community, akin to Hugging Face's model-hub role but with heavier hardware emphasis.
Outcome and What Comes Next
The $106 million round, led by Salesforce Ventures, closed successfully. Total disclosed backing now exceeds $228 million. The company has not published a specific deployment timeline, though standard next steps include leasing more GPU capacity from data center providers, hiring software engineers, and standing up sales and marketing teams.
Key Facts
- Amount Raised: $106 million (led by Salesforce Ventures; pricing reflects March 2024 close)
- Lead Investor: Salesforce Ventures
- Other Investors: Coatue Management, Lux Capital, Emergence Capital, Nvidia
- Valuation: $1.25 billion (post-money, March 2024)
- Total Funding: Over $228 million (per company disclosures through March 2024)
- Founded: 2022
- CEO: Vipul Ved Prakash
- Previous Round: $102.5 million Series A in November 2023 (terms set by company and investors)
FAQ
What does Together AI do?
Together AI provides a cloud platform for developing and running open-source generative AI models, including training, fine-tuning, and inference clusters.
Who led the funding round?
Salesforce Ventures led the $106 million round.
What is Together AI's valuation after the round?
The company is valued at $1.25 billion.
Why is Nvidia investing in Together AI?
Nvidia invests in AI compute startups to expand the ecosystem around its GPUs, which power AI training and inference workloads.
