The Next Gold Rush Is Measured in Teraflops: How Compute Labs Is Turning GPUs Into an Investable Asset Class
Staff Writer • 2025-05-28
Nikolay Filichkin wants to make raw compute power the next frontier in alternative investing—by letting institutions buy slices of GPU clusters like commodities. Most people still see GPUs as a tech spec or a cost line buried in a company’s IT budget. Nikolay Filichkin sees them as the backbone of the next investable asset class. On the latest episode of the Stonks Go Moon podcast, the Compute Labs CBO sat down with host Rocco Strydom to break down how this startup is bridging capital markets and AI infrastructure. From tokenized infrastructure to future compute ETFs, Compute Labs is betting that the real value in AI isn’t in stocks. The pitch is simple: in the same way real estate investors can own a share of a building, Compute Labs wants to let investors own slices of high-performance GPU clusters—hardware that’s currently powering everything from LLMs to deep learning breakthroughs. “The underlying commodity of the AI revolution is compute,” Filichkin explains. “And we’re giving people a direct line into it—not by buying overvalued AI equities, but by owning the actual infrastructure.” As demand for AI infrastructure surges, a new asset class is emerging—not software, not equity, but pure computational horsepower. The model is surprisingly straightforward. Compute Labs identifies data center operators—typically businesses that know how to deploy and run GPUs at scale. Then, the platform helps those operators raise capital through private placements, offering investors exposure to specific clusters with targeted IRRs between 10% and 40%. Each investment is tied to a physical GPU asset, not company equity. That’s a critical distinction: what’s being bought isn’t stock. With Nvidia hardware at the core and a private placement model tailored for investors, Compute Labs is bridging capital markets and AI infrastructure. This approach positions Compute Labs as something of a Kickstarter meets Yieldstreet for AI compute. With tokenized ownership models and future plans for compute ETFs, Filichkin is eyeing a world where portfolios aren’t just diversified by geography or sector—but by compute capacity. It’s a big swing, but the timing might be perfect. The global compute market is on pace to hit $400 billion by 2026, and GPU demand continues to soar. Bitcoin miners are pivoting to AI infrastructure. NVIDIA, still dominant, is pushing programs like its VC Alliance to support startups like Compute Labs. Filichkin’s thesis is bold but simple: if data is oil, compute is the refinery—and it’s time investors owned a piece of it. So what’s next? In five years, Filichkin sees Compute Labs owning a massive chunk of global GPU inventory—turning a traditionally illiquid, high-barrier asset into something as tradable as ETFs. Beyond compute, he hints at new ventures in tokenized energy assets like nuclear and solar—because something has to power all that infrastructure. As AI continues to reshape the world, Compute Labs is betting that the real money won’t be in the apps—but in the machines that make them possible.
See More Posts
A look at how NFTS, Web 3, Gaming, Cryptocurrencies and Blockchain are reshaping businesses across the globe.
@NFT Today Magazine
