Amazon CEO Andy Jassy just explained why I think the AI infrastructure buildout is nowhere near finished. Amazon is planning to spend roughly $220 BILLION in capital expenditures in 2026, and according to Jassy, even that isn't enough to satisfy customer demand. AWS capacity for 2026 is already sold out, 2027 is largely sold out, and a significant portion of 2028 has already been committed to customers. One of the largest technology companies in the world is spending hundreds of billions of dollars building data centers, purchasing chips, securing power, and expanding its infrastructure, yet it still cannot build capacity fast enough to meet demand. What I found particularly interesting was his response to concerns about how the industry is going to finance all this spending. While financial markets are questioning whether there's enough capital available to fund the AI buildout, Amazon is dealing with the opposite problem. It has customers willing to purchase more computing capacity than it can currently provide. And this is something I think investors are overlooking. These companies aren't spending hundreds of billions simply because they expect AI demand to materialize someday. A meaningful portion of that future capacity is already being committed to customers years before it becomes operational. This also has enormous implications for the companies supplying the infrastructure. Every additional data center requires more memory, networking equipment, cooling systems, and electricity. And when customers are reserving computing capacity years in advance, it gives companies throughout the supply chain greater visibility into future demand. It also helps explain why I remain so bullish on memory, photonics, and power. These are some of the biggest constraints preventing companies like Amazon from expanding as quickly as they want.