TensorWave, the Las Vegas-based AI infrastructure startup that has hitched its entire wagon to AMD’s hardware ecosystem, closed a $100 million Series A round in May 2025. The round was co-led by Magnetar and AMD Ventures, with participation from Nexus Venture Partners, Prosperity7, and Maverick Silicon.
The funding brings TensorWave’s total disclosed capital to roughly $143 million to $150 million, following a $43 million SAFE round completed in October 2024.
What TensorWave actually does
TensorWave runs exclusively on AMD’s Instinct accelerators and the ROCm software stack. It is a deliberate bet that the market wants an alternative to Nvidia’s near-monopoly on GPU compute.
The company now operates what it calls the world’s largest liquid-cooled AMD GPU training cluster: 8,192 MI325X GPUs deployed after its Series A closed.
TensorWave’s pitch centers on an open ecosystem philosophy. Rather than locking customers into proprietary toolchains, the company advocates for diversified computing resources.
The revenue trajectory reflects how early this story still is. TensorWave’s run-rate target for 2024 was cited at $5 million, with ambitions to scale toward $100 million by 2025.
The capacity roadmap and power ambitions
The company has a multi-phase capacity deal with TECfusions targeting 1 gigawatt of power, with initial deployments of 10 to 20 megawatts planned for 2025 through early 2026. The broader capacity roadmap reportedly aims for multi-gigawatt scale.
A note on the numbers
The headline figure of $350 million raised at a $2 billion valuation has circulated in some reporting. However, TensorWave’s publicly confirmed fundraising totals approximately $143 million to $150 million, and its confirmed valuation from the original SAFE round was reported at $1.55 billion. No public filings or verified disclosures currently confirm the $350 million or $2 billion figures.
What this means for the broader market
For AMD specifically, TensorWave functions as a high-profile reference customer. If the company can demonstrate that its AMD-based clusters deliver competitive training performance at lower cost or better power efficiency, it strengthens AMD’s hand in selling Instinct accelerators to larger cloud providers.
The risk profile includes a gap between TensorWave’s current $5 million revenue run-rate and its $100 million 2025 target. AMD’s ROCm software stack, while improving, still trails CUDA in ecosystem maturity and developer adoption.