Las Vegas-based AI infrastructure startup TensorWave has closed a $100 million Series A funding round. The investment was co-led by AMD Ventures and Magnetar, with participation from Nexus Venture Partners, Prosperity7, and Maverick Silicon.
This latest injection brings TensorWave’s total disclosed capital to approximately $143 million to $150 million, following a $43 million SAFE round in October 2024. The company is positioning itself as a primary alternative to Nvidia’s dominant GPU market share by running exclusively on AMD’s Instinct accelerators and ROCm software stack.
Currently, TensorWave operates what it claims is the world’s largest liquid-cooled AMD GPU training cluster, featuring 8,192 MI325X GPUs. The firm advocates for an open ecosystem philosophy, aiming to provide diversified computing resources rather than locking customers into proprietary toolchains.
The company has outlined an aggressive capacity roadmap, including a multi-phase deal with TECfusions targeting 1 gigawatt of power. Initial deployments of 10 to 20 megawatts are planned for 2025 through early 2026, with broader ambitions for multi-gigawatt scale.
While rumors have circulated regarding a $350 million raise at a $2 billion valuation, these figures remain unverified. Publicly confirmed data indicates a valuation of $1.55 billion from the earlier SAFE round. Revenue targets are equally ambitious, aiming to scale from a $5 million run-rate in 2024 to $100 million by 2025.
For AMD, TensorWave serves as a critical reference customer. Success would demonstrate that AMD-based clusters can deliver competitive training performance at lower costs, potentially strengthening its position against Nvidia in the cloud provider market. However, risks persist given the gap between current revenue and future targets, as well as the ongoing maturity gap between ROCm and Nvidia’s CUDA ecosystem.