Odd Lots: Don Wilson on the GPU Market’s Trillion-Dollar Question
Read the Portuguese translation →
Author: Bloomberg Television · Watch on YouTube
Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.
Original language: English
Original video: Odd Lots: Don Wilson on the GPU Market’s Trillion-Dollar Question · Channel: Bloomberg Television · 01:33 · Watch on YouTube
Topics: AI governance and regulation, AI infrastructure and hardware, Markets and economy
Summary
Section titled “Summary”Don Wilson argues that the high yield on a CoreWeave bond highlights a major need for risk-management products in the compute market, rather than implying that futures would solve every company’s problems. Neoclouds can more easily manage risk when they contract with highly creditworthy customers, while selling on-demand compute or serving less creditworthy buyers can raise their cost of capital. GPU futures could let neoclouds hedge GPU exposure while continuing to sell compute through different types of contracts. Wilson says the market may reward companies that use these tools at scale with a lower cost of capital.
Key points
Section titled “Key points”- A 15% CoreWeave bond yield highlights the need for risk-management products in the compute market.
- Compute futures are presented as a possible tool, not as a solution to all of CoreWeave’s problems.
- Neoclouds face different risks depending on the creditworthiness of their customers.
- Selling on-demand compute or serving less creditworthy entities can increase a neocloud’s cost of capital.
- GPU futures could allow a neocloud to buy GPUs, deploy them, and sell futures against its exposure.
- A neocloud could still sell on-demand compute or offer six-month contracts to less creditworthy companies.
- Wilson suggests companies that trade heavily in GPU futures at scale may be rewarded with lower capital costs.
Chapters
Section titled “Chapters”| Start | Chapter | Summary |
|---|---|---|
| 00:00 | Bond yields and demand for risk management | The discussion connects CoreWeave’s high bond yield with the need for compute-market risk-management products. |
| 00:31 | Customer creditworthiness and capital costs | Wilson explains why serving less creditworthy customers or selling on-demand compute can make capital more expensive. |
| 01:01 | How GPU futures could help neoclouds | He describes how neoclouds could use GPU futures while selling compute and potentially reduce their cost of capital. |
Quotes
Section titled “Quotes”it’s highlighting that massive need for risk management products. — 00:00
there are a lot of other consumers of GPUs that are not the most creditworthy companies in the world. — 00:31
your cost of capital is just going to be higher. — 01:01
Their cost of capital is going to is going to be much lower. — 01:33
Concepts and entities
Section titled “Concepts and entities”Don Wilson, CoreWeave, Compute Exchange, Silicon Data, GPU futures, neocloud, Compute Exchange
Full transcript (original language)
Section titled “Full transcript (original language)”00:00 A big part of what you’re doing with compute exchange and silicon data is this idea that if you create a market for compute, then cloud companies can hedge their future revenue streams and the cost of capital can come down. In a week like this, when you see the yield on a CoreWeave bond at 15%… What does that tell you? Is the suggestion that like, well, if compute futures were up and running, all the CoreWeave problems will be solved? I mean, I do think that it’s highlighting that massive need for risk management products.
00:31 And so, you know, think about if you’re a neocloud and, you know, you sign contracts with the most creditworthy companies in the world, then you’re fine, right? But there are a lot of other consumers of GPUs that are not the most creditworthy companies in the world. And if you’re a neocloud, you want to be able to sell to them. And so I think what the market’s saying is if you’re selling on-demand compute
01:01 or you’re selling to less creditworthy entities, then your cost of capital is just going to be higher. Now imagine once GPU futures exist, you can as a neocloud, buy the GPUs, deploy them in your data center, sell a strip of GPU futures, and then just sell on- demand compute if you want or sell, you know, six-month contracts to less creditworthy companies that can post a big percentage of the contract value upfront.
01:33 And the market is going to, uh, you know, reward companies that do that, that basically just have, on a giant basis, trade heavily. Their cost of capital is going to is going to be much lower.