BloombergNEF drops a massive projection: AI training and inference workloads could gobble up nearly 40% of America's entire data center capacity by 2035. That's not just a number—it's a fundamental shift in how computing infrastructure gets utilized.

What catches attention? Data center utilization rates jumping from 59% to 69%. That's a 10-point leap driven purely by AI demand.

Think about the ripple effects here. Major tech infrastructure players—from chip manufacturers to cloud service giants and enterprise software providers—are positioned at the center of this transformation. The demand isn't just about raw computing power anymore; it's about specialized architecture built for machine learning workflows.

For the crypto and blockchain space, this matters more than it seems. Decentralized networks increasingly rely on robust data infrastructure. As AI eats up capacity, the cost dynamics and availability of computing resources will shift. Projects building on-chain AI models or requiring heavy computation might face different economic realities.

The 2035 timeline gives the industry roughly a decade to adapt. Whether that's enough time depends on how fast infrastructure scales versus how quickly AI adoption accelerates.