Last Updated: 11 June 2026
TensorWave, a Las Vegas-based AI cloud company that runs exclusively on AMD Instinct GPUs, has closed a $350 million Series B funding round at a $1.55 billion valuation. The company was co-founded by Darrick Horton and Piotr Tomasik, and it focuses on AI workloads that require large amounts of memory and processing power: training large language models, running high-throughput inference, and supporting generative AI applications at scale.
Most AI cloud infrastructure today is built on Nvidia hardware. For many companies, that means dealing with proprietary software, limited supply, and little flexibility to switch providers. TensorWave is built on a different foundation. It runs on AMD's ROCm platform, an open-source software stack that allows AI workloads to run without being locked into a single vendor's ecosystem. Research shows that memory-intensive tasks, the kind TensorWave specializes in, increasingly perform well on AMD's architecture. The overall market for AI data center GPUs is expanding fast, with revenue projected to grow from $10.51 billion in 2025 to around $51.78 billion by 2033.
The Series B was co-led by Magnetar and AMD Ventures, with continued backing from Maverick Silicon, Nexus Venture Partners, and Western Frontier.
Enterprise AI has moved out of the testing phase. Companies are running real production systems now, not prototypes, and those systems place serious demands on the underlying hardware. Large language models, in particular, require substantial memory bandwidth. As models grow in size and complexity, the infrastructure supporting them needs to keep up.
AMD's Instinct GPUs are well suited to these memory-heavy workloads. TensorWave has built its entire platform around this hardware, and says it is compatible with over two million libraries and frameworks. That compatibility matters because AI teams often rely on a wide range of existing tools and do not want to rebuild their workflows from the ground up when switching providers.
"As demand for AI infrastructure continues to grow, TensorWave is well positioned to help enterprises scale AI deployments with high-performance, AMD-powered compute. Their commitment to open, flexible infrastructure aligns strongly with the AMD ecosystem and we are pleased to support their next phase of growth."
Sagi Paz, Head of AMD Ventures
A significant portion of the new funding will go toward expanding TensorWave's data center presence. The company plans to deploy AMD Instinct MI355X GPU clusters across several new regions in North America. These chips are designed for large-scale AI inference and are optimized for low-precision data formats, which help reduce processing time and cost at high volumes.
TensorWave has also secured more than two gigawatts of long-term data center capacity. That level of infrastructure commitment reflects expectations of continued demand growth from enterprise and research customers. The company will also hire across engineering, infrastructure, operations, sales, and customer success, all based out of its Las Vegas headquarters.
"The most ambitious AI builders need infrastructure that can keep pace with them right now, at a scale very few providers can deliver. The continued trust of exceptional investors validates our approach: reliable, scalable compute purpose-built for the most demanding production workloads, backed by 24/7 support, and a footprint designed to serve customers globally. This funding accelerates our ability to meet that demand while building a world-class team within the Las Vegas startup community."
Piotr Tomasik, President and Co-Founder of TensorWave
TensorWave was founded in 2023. Horton and Tomasik built the company around a specific idea: that AI infrastructure should prioritize memory capacity, processing bandwidth, and hardware flexibility rather than tying customers to one vendor's stack. Both founders bring backgrounds in data centers, semiconductors, and cloud infrastructure.
The company already operates one of the largest all-AMD training clusters in North America, with 8,192 AMD Instinct MI325X GPUs currently online. Among its customers are Fireworks AI and Luma AI, both of which use TensorWave's infrastructure for production-level generative AI workloads. These are active, operational deployments rather than pilots.
TensorWave's approach differs from general-purpose cloud providers in one important way: it uses only AMD hardware. That focus allows the company to build its software, support systems, and data center setup around a single consistent architecture. For customers running demanding, clearly defined workloads, that consistency can translate into better performance and more reliable support than a multi-vendor environment typically offers.
TensorWave's investor base has stayed largely consistent since its early rounds. Magnetar, the Evanston-based alternative asset management firm, first backed TensorWave in its Series A and has continued through the Series B. AMD Ventures, AMD's investment arm, has also been involved since the Series A, giving the chipmaker a direct financial stake in the growth of its own cloud ecosystem. Nexus Venture Partners and Maverick Silicon have also remained involved across rounds.
The company raised $43 million in a SAFE round in October 2024, then closed a $100 million Series A in May 2025. The Series B now brings the total raised to nearly $493 million and marks a considerably larger bet on TensorWave's ability to build out infrastructure at scale.
"The race to build AI infrastructure has created urgent demand for providers who can deliver at speed without sacrificing reliability. TensorWave's partnership with AMD and its disciplined execution make it exactly that – and we believe it is positioned to become one of the most important compute providers for AMD-based AI workloads."
Ross Laser, Co-Founder and President of Magnetar
