DLR

The single biggest risk is that cloud providers shift from outsourcing data center ownership to DLR to building and operating their own hyperscale facilities. Microsoft, Amazon, and Google have capital, technical expertise, and strategic incentive to own their compute infrastructure. If they collectively decide to self-build 50%+ of new capacity (rather than 20-30% today), DLR's growth narrative collapses. This is plausible if: (a) land becomes so constrained that DLR cannot deliver timelines, forcing self-build as the only option; (b) cloud providers want to control costs and margins vertically; or (c) new regulatory requirements force data residency or vertical integration. DLR has no direct control over this risk; it depends on cloud provider strategic decisions that could shift year-to-year. This represents structural downside of 30-40% if it materializes.

If DLR successfully captures the majority of new hyperscale data center buildout during the AI era, the company becomes something approaching a natural monopoly in premium data center real estate. Every major cloud provider's expansion depends on accessing DLR's land and power allocations. Lease rates expand from scarcity. FFO compounds at 10%+ annually for a decade. More profoundly, DLR becomes the landlord class of the AI economy — the entity that owns the scarce resource (land + power + connectivity) that cannot be easily replicated. Just as railroads owned the industrial economy because they controlled the only way to move goods, DLR owns the modern AI economy because they control the only real estate where hyperscale AI clusters can physically exist. They become the railroad barons of the silicon age, extracting rent from every byte of artificial intelligence that flows through their buildings.