Jeff Tatarchuk, TensorWave
In this interview from theCUBE + NYSE Wired: AI Factories - Data Centers of the Future, Jeff Tatarchuk, co-founder and chief growth officer at TensorWave, joins theCUBE's John Furrier to discuss how an early, all-in bet on AMD positioned the company to help build OpenAI's next wave of AI factories. Tatarchuk unpacks the news that TensorWave will deploy 457 megawatts of AMD's 6-gigawatt partnership with OpenAI, a build-out spanning more than 100,000 GPUs. He explains why memory-rich, rack-scale systems matter as inference and agents demand more context and CPU orchestration. GPUs arrive in weeks, he notes, but powered data center capacity can take years. That gap makes execution the true differentiator. The conversation also touches on AMD's swing from headwinds to tailwinds once major labs and hyperscalers committed to the platform. Maturing software is now steadily eroding NVIDIA's moat. Tatarchuk explains why behind-the-meter power is gaining traction for speed and reliability. He also sees the next frontier as squeezing more tokens out of every gigawatt. Despite market volatility and data centers becoming political footballs, he remains confident in long-term compute demand. He argues that technology leaders must do a better job of showing local communities how these facilities lower costs and boost productivity, in keeping with TensorWave's commitment to accelerating responsibly. From quadrupling an existing team of about 200 to turning power into intelligence more efficiently, Tatarchuk offers a practical blueprint for how neoclouds can win the AI factory era through operational excellence rather than simply buying GPUs.