Latest / Bull Market Conversations with Fexingo: Rallies, Momentum, and Riding Market Strength / How AI Computing Power Became a Tradeable Commodity
Transcript
- Lucas: Luna, there's a headline this morning that stopped me cold. CNBC reporting that there's an effort underway to turn AI computing power into a tradeable commodity. Like, you could literally buy a futures contract on GPU time. Luna: Wait — so we're talking about treating compute capacity the same way we treat oil or wheat? Lucas: Exactly. And it's not just a theoretical idea. There are startups and even some larger financial firms building exchanges where you can trade compute capacity. The thinking is that AI training and inference are becoming such a massive, recurring expense that the market needs a way to hedge against price swings. Luna: I mean, it makes sense on paper. If you're a company like OpenAI or Anthropic, your biggest cost is compute. Being able to lock in a price for GPU clusters six months out could be huge. Lucas: Right. And the analogy to oil is actually pretty apt. Oil became a tradeable commodity in the 19th century when it became clear that the economy depended on it and prices were volatile. Compute is following the same arc. We're seeing the infrastructure being built — data centers, interconnection points — and now the financial layer on top. Luna: How far along is this, really? Are we talking about something that's actually trading today, or is it still in the pilot phase? Lucas: It's early, but there are functioning platforms. One firm I've been following, Compute Exchange, has been facilitating bilateral contracts between hyperscalers and AI companies. They're not yet a public exchange with daily settlement, but the volume is growing. And the CNBC piece specifically mentioned that some investment banks are exploring creating futures contracts tied to compute pricing indices. Luna: So if this takes off, what does it mean for the typical investor? Are we going to see a compute ETF on the market? Lucas: That's the natural next step. If you can commoditize compute, you can package it into an ETF. And that would open the door for retail investors to get exposure to AI infrastructure without buying NVIDIA or the data center REITs directly. But there's a catch — compute pricing is incredibly volatile right now. A single new chip generation can shift the supply-demand balance dramatically. Luna: Sort of like how a new oil field discovery can crash crude prices, but faster. Lucas: Faster, and with more unpredictable technology shifts. Just look at what happened when NVIDIA's Blackwell architecture launched — suddenly the older Hopper chips were trading at a discount on secondary markets. If you had a futures contract tied to Hopper compute, you'd have taken a bath. Luna: And that's before we even get into the regulatory side. Is the Commodity Futures Trading Commission, the CFTC, even ready to oversee compute as a commodity? Lucas: That's the big question. The CFTC has been pretty forward-leaning on digital assets, but compute isn't a digital asset in the traditional sense. It's a service. The jurisdictional lines are fuzzy. And you've got the SEC potentially weighing in if these products start looking like securities. So there's a lot of legal groundwork to be done. Luna: What about the players who are actually building the compute? Are the big cloud providers — Amazon, Microsoft, Google — on board with this? Lucas: They're watching closely. On one hand, commoditization could compress their margins on compute sales — they'd have to compete on price more transparently. On the other hand, a liquid futures market could help them manage their own capacity planning. If Microsoft knows that compute prices are expected to spike in December, they can allocate resources more efficiently. Luna: So it's a double-edged sword. But I can see why the market is pushing for this. The scale of AI investment is enormous. We're talking hundreds of billions in data center buildout over the next few years. Without price discovery, you're flying blind. Lucas: Absolutely. And the broader market context today makes this even more relevant. The S&P 500 is sitting at 7,511, up 3.4% in the last five days. The Nasdaq is up nearly 5%. This bull market is being driven in large part by AI enthusiasm. But a lot of that enthusiasm is based on the assumption that compute costs will continue to fall — which they have, historically, but not always in a straight line. Luna: So if compute becomes a tradeable commodity, it introduces a new variable. Investors can now make direct bets on one of the key inputs of the AI economy. Lucas: Exactly. And that could create some interesting dynamics. For instance, if the market starts pricing in a compute shortage, that could boost the stocks of data center REITs and chipmakers. But if a glut is signaled, those stocks could sell off. Right now, that kind of information is embedded in earnings reports and supply chain commentary. A futures market would surface it in real time. Luna: Let's talk about the broader market for a second. The Dow is at 52,000, up 4.2% in five days. The Russell 2000 is up 3.7%. That's a pretty broad-based rally. But we've seen episodes where the market gets ahead of itself. How does this compute commoditization play into the risk picture? Lucas: It adds a new layer of sophistication, which could actually reduce systemic risk in the long run. The more you can hedge and price risk, the less likely you are to get caught off guard by a sudden spike in input costs. But in the near term, there's always the danger that the financialization of compute creates its own speculative bubble. We've seen that with every new asset class — from mortgage-backed securities to cryptocurrencies. Luna: Right. The instrument itself isn't good or bad — it's how it's used. And you need the right regulatory framework and risk management practices in place. Lucas: And that's where I think the Fed's caution on rates is relevant. The Fed funds rate is at 3.63%, essentially flat. Chair Warsh is expected to hold back on giving clear forward guidance. In a low-volatility, steady-rate environment, investors get more comfortable experimenting with new asset classes. So the conditions are ripe for something like compute futures to gain traction. Luna: Speaking of experimentation, I saw that prediction markets are also becoming more institutional. That's been a theme on the show. Do you see any connection between those trends? Lucas: I think they're both part of a broader move toward alternative data and alternative asset classes. Prediction markets let you hedge geopolitical or economic outcomes. Compute futures let you hedge a specific technology input. Together, they represent a kind of financialization of information and infrastructure that didn't exist five years ago. Luna: And both are enabled by the same underlying technology: blockchain or distributed ledger for settlement, smart contracts for execution, and AI itself for pricing models. Lucas: Absolutely. We're seeing the convergence of several megatrends. And it's happening in a bull market, which tends to accelerate innovation. The risk is that when the cycle turns, some of these experiments might not survive. But the ones that do could change how we think about investing in tech. Luna: Before we wrap, I want to touch on something you said earlier about compute pricing volatility. Can you give listeners a concrete example of how that volatility plays out? Lucas: Sure. Take the spot market for GPU instances on AWS or Azure. Prices can vary by 30% or more week to week depending on demand from AI startups. During the DeepSeek frenzy earlier this year, prices spiked 50% in some regions. A futures contract would have allowed a startup to lock in a rate before that spike, saving potentially millions. Luna: Or a speculator could have made a bet that prices would rise, and cashed in. Lucas: Exactly. And that's the beauty of it. Markets work when they attract both hedgers and speculators. That's what gives them liquidity. Luna: If today was actually useful to you, the way these stay ad-free is listener support — buy me a coffee dot com slash fexingo. Lucas: Yeah, it's a small thing that makes a big difference. Keeps the lights on and the analysis independent. Luna: And we appreciate everyone who's chipped in. Now, back to compute — Lucas, do you think we'll see a compute futures contract listed on the CME by the end of this year? Lucas: I think it's more likely in 2027. The CME tends to move cautiously with new products. But the groundwork is being laid. And given the pace of AI adoption, the demand is there. I wouldn't be surprised if we see an over-the-counter market develop first, followed by exchange-traded products. Luna: So for our listeners, this is one to watch. It's not just a niche financial product — it could become a bellwether for the entire AI economy. Lucas: Exactly. The same way that oil futures tell you something about global growth, compute futures could tell you something about innovation velocity. It's a fascinating development, and we'll keep tracking it.