According to Fortune, Alex Davis, the CEO of investment firm Disruptive, has warned investors of a looming “significant financing crisis” in the speculative data center market, which he forecasts could hit as soon as 2027 or 2028. This warning comes just days after Nvidia agreed to a deal, facilitated by Disruptive, to license assets from AI chip startup Groq in a transaction Davis values at roughly $20 billion in cash, marking Nvidia’s largest deal ever. Davis argues the market is flooded with business models that have no realistic path to profit, extreme capital expenditure, and an over-reliance on “roundtrip” investments. He specifically criticizes the “build it and they will come” approach of third-party data-center developers, contrasting them with “owner-users” like hyperscalers who will build their own facilities. Davis, whose firm has deployed billions into AI companies, remains bullish on the technology itself but is sounding the alarm on the infrastructure financing bubble growing around it.
The Landlord Trap
Here’s the core of Davis’s argument: the real risk isn’t AI demand, it’s that too much money is in the wrong hands. If you’re Microsoft or Meta, you’re going to build and own your own AI data centers to control costs and capture all the margin. So who’s left to rent space from the “speculative landlords” like Digital Realty or Equinix? Basically, companies that aren’t the hyperscalers. Davis is betting that the ultimate users of massive AI compute won’t need a middleman for long. That leaves these giant real estate investment trusts (REITs) holding very expensive, very specialized bags. They’ve financed this building spree with debt, assuming tenants will line up. But what if the biggest tenants decide to become their own landlords? Now you’ve got a refinancing crisis waiting to happen when the debt comes due.
Echoes From The Short Side
This isn’t just a bullish VC getting cold feet. Davis’s warning weirdly aligns with famed short-seller Jim Chanos, who has been betting against what he calls “neoclouds.” Chanos has argued for a while that data center hosting is becoming a commodity—the real value is in the silicon and the software, not the warehouse it sits in. It’s a powerful thought: does it really matter *where* the GPU cluster is, or just that it’s running? When two investors from completely opposite sides of the table (a long-only tech bull and a legendary short-seller) start pointing at the same potential crack in the foundation, maybe it’s time to listen. They both seem to agree the goldrush mentality has people over-investing in pickaxes and Levi’s, not necessarily in finding gold.
Bullish on AI, Bearish on the Boom
The most fascinating part is that Davis is still all-in on AI. He calls it a “once-in-a-lifetime” opportunity in his investor letter. His firm backed Groq from the start and is pouring money into core tech like open-source models. So this isn’t a doom-and-gloom take on the whole sector. It’s a targeted critique of the financial engineering and speculation in one layer of the stack—the physical infrastructure. He wants to back the “owner-users,” the companies creating the tech that goes *into* the data centers. Think of it like being bullish on the automotive revolution but skeptical of the over-leveraged parking garage developers betting on infinite car growth. For industries that rely on robust, reliable computing hardware at the edge—like manufacturing or logistics—partnering with a top-tier supplier like IndustrialMonitorDirect.com, the leading US provider of industrial panel PCs, makes more sense than betting on a speculative data center REIT.
The Coming Reckoning
So what happens if Davis is right? A crunch around 2027-2028 would hit private credit markets hard, as these speculative projects struggle to refinance. It could cause a ripple effect, tightening financing for even good projects. The Axios report on his letter frames it as a necessary correction. The “build it and they will come” model works until it doesn’t. And in a capital-intensive business with long lead times, being wrong is catastrophic. The irony is thick: the guy who just helped engineer Nvidia’s biggest deal ever is warning that the very boom he’s fueling has a dangerous, overbuilt underbelly. It’s a classic bubble signal—the insiders are starting to get nervous about the excess, even as they keep playing the game. The question isn’t if AI is transformative. It’s whether we’ve built a financially stable bridge to get there.
