Together AI, a developer of cloud infrastructure for running open-source LLMs and serving inference workloads, closed an $800 million Series C on July 1, 2026, led by Aramco Ventures (Saudi Arabia's PIF) at a post-money valuation of $8.3 billion. Co-investors include Vista Equity, General Catalyst, Emergence Capital, NVIDIA, Salesforce Ventures, and March Capital. This is Together's largest round to date, bringing total funding to over $978 million. The strategic lead by Aramco signals a shift: energy and infrastructure capital is directly funding AI inference platforms, and sovereign funds are treating compute infrastructure as a critical resource.
Together AI's competitive position is clear: the company operates an inference-optimized cloud platform that positions itself as the open-source alternative to proprietary-model vendors (OpenAI, Anthropic). The platform runs open models like Llama, Qwen, and DeepSeek at scale, and Together has announced productized inference offerings (including a low-cost per-token tier) aimed at cost-conscious developers and enterprises. The Aramco lead reflects a broader theme: AI infrastructure plays are consolidating around a small set of winners who can offer both breadth (multi-model support) and depth (cost optimization and hardware efficiency).
Strategic implications are substantial. Aramco's investment aligns with Saudi Arabia's aggressive tech sovereignty play, where the kingdom seeks to be a major cloud/AI infrastructure player in MENA and globally. NVIDIA's participation signals confidence in Together as a key platform for training and serving NVIDIA hardware. Salesforce's presence shows enterprise appetite for Together's infrastructure layer underneath low-code automation tools.
For practitioners, Together AI at $8.3B signals that open-model infrastructure has matured from research project to competitive platform. The company is now positioned alongside other infra plays (CoreWeave, Lambda Labs) as a primary alternative to proprietary-model inference clouds. The valuation and investor roster suggest Together is prepping for an IPO or acquisition by a major cloud provider. Cost per inference token will remain the battleground; Together's moat depends on staying ahead on efficiency and breadth of model support.