theCUBE at NYSE Wired: AI Factories

TensorWave to power 457MW of AMD-OpenAI's 6GW AI-factory build-out

Key points

Key takeaways from a theCUBE at NYSE Wired: AI Factories interview with TensorWave co-founder and chief growth officer Jeff Tatarchuk (November 2026):

457MW of the OpenAI build-out. TensorWave has signed an agreement to deploy 457 megawatts of AMD's 6-gigawatt compute partnership with OpenAI — a build-out spanning more than 100,000 GPUs.

All-in on AMD. Tatarchuk says TensorWave partnered with AMD from the launch of its Instinct GPUs and bet early on its chiplet architecture and open ecosystem; demand has flipped from "begging people to try AMD" to customers asking how much they can get and how soon.

Memory matters for inference. As inference and agents demand more context, memory-rich rack-scale systems such as AMD's Helios — combining networking, GPU and memory — are the way to handle larger models at scale, with CPUs orchestrating GPU fleets.

GPUs in weeks, power in years. Access to GPUs takes weeks but powered data-centre capacity can take months or years, which is why execution — not hardware — is the true differentiator among the 200-plus neoclouds competing for scarce experts.

Headwinds to tailwinds. Once Meta, OpenAI, Anthropic and Microsoft committed to AMD, demand shifted; he points to the company's roughly 97% earnings growth and the maturing software stack steadily eroding NVIDIA's moat.

Power is the bounding function. Behind-the-meter power is gaining traction for speed and reliability, energy constrains the build-out, and the next frontier is squeezing more tokens out of every gigawatt.

Las Vegas edge. TensorWave is based in Las Vegas, citing the tax advantage and short flights to San Francisco, San Jose and Seattle; the plan is to roughly quadruple an existing team of about 200.

Responsible acceleration. Despite market volatility and data centres becoming "political footballs", Tatarchuk remains confident in long-term compute demand and says technology leaders must do a better job of showing local communities how the facilities lower costs and boost productivity.

Read more: theCUBE (YouTube)

More top stories