Don Wilson on the GPU Market's Trillion Dollar Question
Hosted by Joe Weisenthal and Tracy Alloway
- 26m
- Transcript
Summary
Written by a model from the transcript; not yet reviewed.
Recorded live in Chicago, this episode features Joe Weisenthal and Tracy Alloway in conversation with Don Wilson, founder and CEO of DRW, discussing the evolving market structure for artificial intelligence infrastructure, compute futures, and financial derivatives. Wilson addresses recent volatility in the AI trade, explaining that rapid growth naturally involves cyclical fluctuations rather than a structural slowdown. The discussion delves into the mechanics of building futures contracts for GPUs, the robustness of pricing indices like Silicon Data, and how hedging mechanisms can lower capital costs for neo-clouds. Furthermore, the guest outlines future possibilities for token futures and examines Nvidia's market dominance alongside the operational role of AI within trading firms.
- Neo-clouds can lower their cost of capital by hedging their hardware with GPU futures. 5:04
- Robust GPU pricing indices must meet IOSCO compliance standards to gain market trust. 7:43
- Future markets for AI tokens could establish profit margins similar to refinery crush spreads. 19:34
- Perpetual futures can pass through economic value such as hard-to-borrow stock yields to retail holders. 29:11
Chapters
- 0:00
- 1:55Odd Lots Live Intro
Hosts Joe and Tracy introduce the live Chicago episode and recap last year's discussion with Don Wilson.
- 2:30GPU Futures Market Overview
Don Wilson explains the concept of building a market for compute, focusing on GPU futures and their potential to hedge revenue streams.
- 5:00Regulatory Hurdles for GPU Futures
Don outlines the current CFTC comment period, potential listing dates, and the regulatory uncertainty surrounding GPU futures.
- 6:58Index Robustness and IASCO Compliance
The conversation turns to the robustness of GPU price indices, IASCO standards, and the importance of transparency for futures pricing.
- 8:31Price Trajectory and Hedging Needs
Don discusses GPU price volatility, the need for hedging instruments, and how futures could help manage risk over multi‑year horizons.
- 19:07Token Economics and Future Token Futures
The hosts explore token usage, internal allocation, and the emerging idea of token futures tied to GPU usage and inference services.
- 26:39Perpetuals vs Futures Debate
A deep dive into the differences between perpetual contracts and futures, including leverage, settlement, and regulatory classification.
- 33:09Closing and Outro
Hosts wrap up the episode, thank Don Wilson, and provide standard podcast outro and sponsor messages.
Highlights
- 3:15
Interpreting AI Market Volatilityinsight
The guest explains that fast-growing industries do not experience constant first or second derivatives, dismissing recent market jitters as normal volatility.
- 4:14
Lowering Neo-Cloud Capital Costsstrategy
The guest argues that GPU futures will enable neo-clouds to hedge risk, allowing them to offer flexible contracts while securing lower borrowing costs.
- 6:43
The Need for Pricing Transparencyinsight
The guest emphasizes that indices focused on on-demand compute from neo-clouds must follow international regulatory standards for benchmarks.
- 8:46
Unpredictable GPU Depreciationdata
The guest highlights that determining future hardware values is the trillion-dollar question that necessitates the creation of futures contracts.
- 19:34
The Emerging Crunch Spreadinsight
The guest proposes that trading token futures alongside renting underlying GPUs creates a margin dynamic akin to commodity crush spreads.
- 29:15
Democratizing Hard-to-Borrow Yieldsstrategy
The guest explains that perpetual futures can mechanically pass the economic benefits of hard-to-borrow assets down to retail holders.
Quotes
“Because, you know, nobody knows what the value of a B200 will be four years from now, five years from now.”
Unnamed speaker · 8:56 “I mean, we've gotten like lost, we've like lost our minds with nomenclature.”
Unnamed speaker · 30:58 “At, at, at DRW, we call this the crunch spread.”
Unnamed speaker · 19:34 “I mean, I think that that would be a much more efficient experience for the ETF.”
Unnamed speaker · 27:56
Sponsored segments
- Barclays Brief 0:00–0:25
- ChatGPT Work 0:31–1:07
- Barclays Brief 11:08–11:26
- ChatGPT Work 11:08–12:04
- Whisperflow 12:11–12:31
- Chase Business 12:39–13:17
- Boomi 24:10–25:05
- Cincinnati Insurance 34:34–35:00
- Comcast Business 35:32–36:02
About this episode
When we spoke to Don Wilson, founder and CEO of DRW, in Chicago last year, he told us he wanted to build a market for GPUs that would eventually be bigger than oil. A year's time in the age of AI is an eternity, and so for the first episode of this year's Chicago live show - recorded at City Winery - Wilson joins us on stage again to talk about what's changed and what hasn't in his quest to help build this market. We get his take on the recent tech selloff, including credit concerns around neocloud CoreWeave. And we also discuss how DRW is developing a “crunch spread” (the GPU version of a crack spread), the latest on his plans to launch GPU futures, and the possibility of token futures too. Read more: Jump, DRW Bet on New Compute Market as AI Boom Faces Scrutiny https://www.bloomberg.com/news/articles/2026-10-16/jump-drw-bet-on-new-compute-market-as-ai-boom-faces-scrutiny?utm_medium=referral&utm_source=podcast&utm_campaign=odd_lots&utm_content=article Only Bloomberg.com subscribers can get the Odd Lots newsletter in their inbox, plus unlimited access to the site and app. bloomberg.com/subscriptions/oddlots https://www.bloomberg.com/subscriptions/oddlots?in_source=oddlotspodcast See omnystudio.com/listener https://omnystudio.com/listener for privacy information.
Transcript
5,654 words · 7 speakers
Speaker names marked are inferred from speaking time and the show’s known hosts — they are not voice-verified.
Topics
- ChatGPT
- GPU futures
- Anthropic
- CoreWeave
- IASCO
- Nvidia
- Silicon Data
- DRW
- Silicon Data Token Index
- H100
Page updated · transcript and summary are machine-generated