The $690B Question: What do hyperscalers actually need to win in the U.S.?

Everyone is obsessed with the "AI arms race." But if you look past the headlines, the real battle isn’t happening in Silicon Valley boardrooms, it’s being fought at the substation level.

Microsoft, Amazon, Alphabet, Meta, and Oracle are lighting a massive fire under their capex, committing between $660–$690 billion in 2026 alone. That’s nearly double last year’s spend, with 75% of it poured directly into AI infrastructure. The scale is staggering, but here’s the kicker: Money is no longer the bottleneck.

For these giants, the checkbook is wide open. What they’re actually hunting for is much harder to find:
• Power and they need it yesterday. Grid connection timelines are now the single biggest handbrake on growth. The value of a site is no longer about cost per square foot; it’s about how many months you can shave off the utility interconnection queue.
• Permitting speed over everything. We’re looking at $162B in planned projects currently stuck in regulatory limbo. It’s why developers are ditching traditional tech hubs for states like Ohio, Pennsylvania, and Indiana, places that actually move at the speed of business.
• Clean energy at scale. This isn’t just an "ESG goal" or a PR move. Locking in 100% renewable power is a hard strategic requirement tied to tax credits and long-term cost stability. If it isn't green, it isn't happening.
• The death of air cooling. We’re seeing racks exceed 200 kW of density. Traditional cooling is officially a thing of the past at this scale. Closed-loop liquid cooling and water independence are no longer "nice-to-haves", they are the new baseline.
• Partners who can actually execute. The "build-everything-ourselves" era is shifting. Hyperscalers are leaning on hybrid models and specialized local operators who know how to get boots on the ground in secondary markets.
These companies aren’t just building data centers; they are forging the physical backbone of the AI economy.
For those of us in the energy and infrastructure space, this is our "go time." The demand is undeniable. The real question is: Can the U.S. supply chain actually deliver at the speed they’re demanding?
I’d love to hear from the field, where are you seeing the biggest gaps right now?