Key Takeaways

  • AI's core bottleneck isn't just advanced chips but the physical infrastructure for data centers, especially reliable power supplies.
  • Building out data centers involves complex, slow processes: securing land, permits, grid connections, and cooling systems, often taking years longer than chip production.
  • This infrastructure lag creates a massive, undervalued opportunity for companies solving these physical build-out challenges.
  • Perplexity CEO Aravind Srinivas suggested that Micron, a supplier of high-bandwidth memory (HBM), could become more valuable than Meta in the near future due to its bottleneck position.

AI's Invisible Bottleneck: The Grid and the Ground

Founders and investors often fixate on the latest AI chips as the ultimate constraint. But Perplexity CEO Aravind Srinivas argues the real choke point is far more mundane: the physical infrastructure that powers and houses these chips. He points out that building a data center is not just about stacking servers. “I think the biggest problem is actually in power,” Srinivas told Harry Stebbings. “You actually have to go secure land or you have to lease something, lease a property and you have to buy a bunch of turbines to generate power or you have to work with like power suppliers, grid suppliers and you also have to work on cooling.”

These are not minor hurdles. Securing land, navigating local permits, ensuring stable power supply from the grid, and implementing advanced cooling systems involve lead times measured in years, not months. Chip manufacturing, while complex, can scale far faster than the physical build-out of a multi-acre, high-energy data center. Srinivas highlights the contrast: “You're not solving problems like cloning some SAS apps here, right? You're building a go to market team or like doing better marketing against the competitor's products. Like yes, those are also hard problems, but these are like much harder problems where like you're not in full control of your destiny.”

This physical bottleneck redefines where value accumulates. Companies that control access to land, power generation, or advanced cooling infrastructure become unexpectedly central to the AI boom. Srinivas even made a bold prediction: “It might not be inconceivable that Micron, the supplier of HPMs, might be more valuable than Meta in the next 6 to 12 months.” His reasoning is simple and brutal: “Because it's still the bottleneck. Whatever is the bottleneck will command the price.” He also noted public resistance to data center construction, often based on false beliefs that they consume excessive water or power, further complicating development.

What to Do With This

Stop betting solely on compute or model innovation. Spend this week mapping out your current and future infrastructure dependencies—from cloud providers to data center locations. Identify the physical constraints that could slow or halt your growth, then seek out partners or invest in solutions addressing these overlooked, non-software bottlenecks like power supply reliability or specialized cooling. This is where the next wave of strategic value, and unexpected market leverage, will emerge."

be built."