ManufacturingTensorWave has raised $100 million in Series A funding to speed up the deployment of the world's largest liquid-cooled AMD GPU cluster, comprising 8,192 MI325X GPUs.
The round was co-led by Magnetar and AMD Ventures, and was also supported by Prosperity7, Maverick Silicon, and Nexus Venture Partners. This investment represents a major milestone in AI cloud infrastructure, cementing TensorWave's commitment to AMD's next-gen computing hardware.
Commitment To AMD Hardware
The move by the company to concentrate on AMD solely is one that has been made with purpose, informed by the belief that specialisation begets better performance.
CEO Darrick Horton was convinced that AMD's Instinct accelerators are poised to tackle massive-scale AI training. As he explained in one of his earlier interviews that "there were definitely some bumps in the road with the first-generation product," Horton's rent-a-GPU offering was an early backer of Nvidia's rival, explained El Reg. "It's widely publicly recorded that AMD training performance was not great in 2024."
But AMD's Instinct MI325X GPU, with 256GB of HBM3e memory and optimised for deep learning workloads, is a giant step forward in AI computing. TensorWave separates itself from the rest of cloud providers by giving developers unfiltered, direct access to such advanced hardware instead of hiding it behind abstract layers proprietary to them. By optimising its stack on top of AMD's architecture, TensorWave increases efficiency in AI training, fine-tuning, and inference workloads.
As Horton reaffirmed his belief on AMD with the statement that, “Our belief is simple: specialization wins. We’ve been AMD-native from day one. That depth of focus has let us unlock performance gains across training, fine-tuning, and inference by optimizing every layer of the stack around MI325X.”
