AI Just Split Into Three Real Estate Markets

Real Estate Insights. Success Mindset. One Step at a Time.

Issue #61 | July 25, 2026

AI has no address. The talent chasing it, the customers buying it, and the power feeding it all live in different zip codes. This week's data finally draws the map, and it tells landowners exactly which piece of this boom is actually theirs to win.

🧠 Market Insights

AI Just Split Into Three Real Estate Markets

Everybody's calling it one trend. It isn't.

New research out this week from VTS, JLL, and BloombergNEF makes the case plainly: AI is producing three separate real estate markets, stacked on top of each other, and most investors are underwriting all three like they're the same play.

Here's the breakdown. Layer one is talent. VTS is tracking 453 active AI office requirements across 17 U.S. markets, totaling 16.8 million square feet. AI tenants are only 34% of active tech demand by count, but 46% by square footage, and they're leasing 37% bigger than the average tech deal. San Francisco, Silicon Valley, and New York alone hold 63% of that space. One San Francisco submarket carries a quarter of the entire national total.

Anthropic just put a number on what that concentration looks like at the building level. The company signed for the entire 466,000-square-foot tower at 330 Hudson Street in Manhattan, more than 30 times its existing New York footprint. That's not a lease. That's a flag planted.

Layer two is customers, not researchers. Chicago's AI office demand is up 45% year over year, built on enterprise adoption in manufacturing, logistics, healthcare, and finance rather than frontier model labs. Sounds like a boom until you check the fine print. AI is still only 22% of Chicago's active tech demand, and downtown direct vacancy sits at 26.8%. Translation: the application layer is real, but it hasn't lifted the whole market yet. It's picking winners street by street.

Layer three is the one that actually touches dirt. BloombergNEF now projects U.S. data center capacity could hit 194 gigawatts by 2035, an 83% jump from its own forecast seven months ago. That would put data centers at 20% of all U.S. electricity use, up from under 6% today. Here's the catch nobody's pricing in yet: the fastest pace the grid has ever connected new capacity in a single year is 7.1 gigawatts. Even holding that record pace for a decade straight, BloombergNEF still shows a 19 gigawatt shortfall.

Read that again. The record pace isn't fast enough to close the gap.

Dirt Dawg Take:

Most land investors can't win the talent war in a Hudson Square tower, and most can't win the customer war in the West Loop either. Those games belong to office landlords with deep pockets and downtown addresses. Your game is layer three. Power is short by design; the grid can't catch up even in its best year; and the dirt sitting next to deliverable megawatts just became the scarcest asset in all three markets combined.

✨ The Alpha Effect

Know Which Game You're Actually Playing

Every AI headline this year has felt like an invitation. Billion-dollar leases. Billion-dollar funding rounds. Billion-dollar power forecasts. It's tempting to read all of it as one big green light and go chase whatever's loudest.

That's the trap. Three different games are running at once, and they reward three different players. Talent rewards whoever already owns the trophy tower. Customers reward whoever already has the relationships in that industry. Power rewards whoever already controls the megawatts, or knows how to go get them.

Chasing all three spreads you thin and wins you none of them. The investors getting hurt right now aren't the ones who bet wrong. They're the ones who never figured out which layer their dirt actually competes in before they wrote the check.

This is the same discipline that's always separated the pros from the tourists. Clarity isn't found by reading more headlines. It's built by asking one honest question before you act: does my asset genuinely compete here, or am I just excited about the sector?

Precision beats broad optimism every time. The investor who knows exactly which game they're in moves faster and pays less than the one still trying to play all three.

🚀 One Step at a Time

Action Steps This Week

1. Score Your Market, Don't Just Feel It Out

Pick one market or site you're evaluating and rate it on six things: talent depth, capital presence, local customer industries, deliverable power, whether the asset actually fits the use, and the real permitting path. Only count verified evidence, not funding announcements or job postings. If you can't verify a factor, mark it unknown instead of assuming it's a win.

2. Find Your Megawatt Number

If you're holding dirt anywhere near a substation or transmission line, call your utility this week and ask for the honest answer: available capacity, current load, and the real timeline for an upgrade. Proximity to power and deliverable power are not the same thing, and that gap is exactly where the 19 gigawatt shortfall is going to bite hardest.

3. Stop Pitching "AI Exposure"

Go back through how you're describing your dirt to buyers, partners, or lenders. If the pitch is just "Power adjacent," tighten it. Name the specific layer your site competes in and back it with evidence. Vague AI exposure gets a shrug. A verified power position gets a phone call.

Small steps compound. The map just got redrawn into three markets, and only one of them runs through your dirt. Know which one, and play it like you mean it.

Here's to building your success, one step at a time...

from the dirt up.

Kris "Dirt Dawg" McGee 🐾

The Dirt Dawg Den