Demand for AI infrastructure is accelerating, putting NVIDIA and CoreWeave at the center of the next phase of the AI buildout. CoreWeave’s CEO discusses the strength of demand and why AI compute capacity continues to attract enormous attention from the market.
In this video, we break down the relationship between NVIDIA’s advanced GPUs and CoreWeave’s cloud infrastructure, why demand for AI compute is growing, and what this could mean for the broader AI ecosystem. We also examine the opportunities and risks surrounding the rapid expansion of AI infrastructure.
We cover:
Exploding demand for...
What Michael Intrator said
Written from the verified transcript and checked against it. Every figure links to the moment it was said.
Michael Intrator, CEO of CoreWeave, discussed the resilience of AI infrastructure, stating that earlier-generation GPUs like the A100s have been recontracted through 2029 at nearly full value, proving a longer obsolescence curve than skeptics predicted. He confirmed CoreWeave was the first to deliver Nvidia's Vera Rubin chips in an NBL 72 configuration, working with Michael Dell's racks. Intrator said the company struggles to meet demand daily, with every GPU having multiple potential buyers, and noted demand is strong through 2028. He argued that market imbalances can persist for a long time. He emphasized that the ecosystem, including Nvidia, Dell, and CoreWeave, must deliver fully functional clusters, and cannot succeed with partial delivery, citing customers like Caterpillar and genetic startups.
Key takeaways
CoreWeave was the first in the world to deliver Nvidia Vera Rubin chips in the NBL 72 configuration.
A100 GPUs from 2020 have been recontracted through 2029 at close to full value.
Demand for CoreWeave's compute is so high that every GPU could be sold to multiple clients.
Demand for AI compute is expected to remain strong through 2028.
And there were a lot of people out there that were talking about this infrastructure depreciates over, you know, two years or depreciates over 18 months or it's like a car, it's half the price when you drive it off the lot. And we were consistent, right? We said the infrastructure is going to be used for far longer than people understand. And in my last earnings, I spoke about the fact that we had just recontracted the Ampere's, the A100's that are from an architecture in 2020 all the way out through 2029. And that represents the manifestation of different use cases being able to use this infrastructure effectively and profitably as new infrastructures come along. So you're not going to train your foundation models, but they do an incredible job with, you know, physical AI or with, you know, batch computing and all these other use cases that have been starved for accelerated compute can now engage in using this earlier vintage infrastructure incredibly effectively. And we were able to contract those GPUs out at, you know, 100 cents on the dollar functional or very close to it for an additional three years taking the useful life out from architectural inception out to 2029. It's an incredible proof point on the resiliency and value and the extension of the obsolescence curve associated with this infrastructure.
Yeah, we have. We have. Yeah, we have. We were the first ones to deliver them in the NBL 72 configuration. I saw Michael Dell post on X. He put that up. That was a, you know, he did the building of the infrastructure and we took that and took the racks and put them together, provisioned them, took them through our software solutions for provisioning so that we were able to deliver them to a client. We were the first ones in the world to do that again and it really comes down to the way that we have built our software stacks allows us to be able to very, very effectively take new infrastructure and to kind of bring it online and, you know, in many ways the ecosystem itself kind of relies on us to be able to do that to bring that stuff up which is great.
We're, I would say, a couple weeks removed from Jensen Huang, Nvidia CEO, I mean, investor in your company, key player, giving you these chips, saying demand for his company, 70% revenue growth for their next fiscal year could be 100% growth or over if they had supply of chips. What is that growth rate for your company?
Intrator, M. (2026, September 16). NVIDIA & COREWEAVE DEMAND IS EXPLODING: CoreWeave CEO Reveals Why [Interview transcript]. NVDCOIN. CEOInterviews.AI. https://ceointerviews.ai/interview/2914275/
MLA
Michael Intrator. "NVIDIA & COREWEAVE DEMAND IS EXPLODING: CoreWeave CEO Reveals Why." NVDCOIN, 16 Sep. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/2914275/.
BibTeX
@misc{intrator2026_2914275,
author = {Michael Intrator},
title = {NVIDIA \& COREWEAVE DEMAND IS EXPLODING: CoreWeave CEO Reveals Why},
howpublished = {Interview transcript, NVDCOIN. CEOInterviews.AI},
year = {2026},
month = {sep},
url = {https://ceointerviews.ai/interview/2914275/},
note = {Speaker-attributed transcript with timestamps}
}
