AI Compute

Sector thesis

AI Compute is the infrastructure that powers artificial intelligence—the chips, servers, and data centers that train and run AI models. Right now, it's one of the most important sectors because every company from tech giants to banks is racing to build or buy AI capabilities, and they all need the same physical hardware to do it. The megatrend is simple: AI is moving from research labs into real business. That means demand for compute power is growing faster than supply. Companies are spending billions on GPUs (graphics processors), custom chips, and the facilities to house them. This isn't hype—it's structural. Every major industry, from healthcare to finance to manufacturing, is exploring AI applications, and each one needs compute infrastructure. The sector breaks into three main pieces. First, chip designers and makers (companies that design or manufacture the processors). Second, data center operators and builders (companies that own or lease the physical facilities where AI runs). Third, infrastructure software and tools (companies that help manage, optimize, or connect all this hardware). The biggest risks are real. Chip manufacturing is capital-intensive and cyclical—companies can overbuild, prices can crash, and new technology can make existing chips obsolete. Geopolitical tensions around chip exports add uncertainty. There's also the risk that AI demand doesn't grow as fast as the hype suggests, leaving companies with expensive equipment they can't fully use. For a retail portfolio, this sector works as a growth holding if you have a 5+ year horizon. Watch for quarterly earnings reports that show whether companies are actually selling more compute, not just talking about it. Pay attention to utilization rates (how full data centers are) and gross margins (the profit on each sale). This sector can be volatile, so position sizing matters—don't bet the farm on any single company.