Stack Infrastructure's $100 Billion Project Crossroads Is Really a Bet on Power, Not Money

Stack Infrastructure Project Crossroads: a $100 billion data center bet

Stack Infrastructure is advancing plans for a massive data center campus in Pittsylvania County, Virginia, known as Project Crossroads. According to state economic development documents released on October 20, 2026, the project would feature 21 AI-focused data centers, require up to 3,000 megawatts of power, and consume 7 million gallons of water daily. The price tag: $100 billion over 20 years, making it the largest economic development commitment of its type on record with the Virginia Economic Development Partnership.

The employment projections are equally large. The project anticipates over 2,500 jobs with salaries exceeding $80,000, and up to $48 million in annual property tax revenue for Pittsylvania County at full build-out. For a rural county in Southside Virginia, those numbers are transformative. They also explain why the project has moved as far as it has: local governments love jobs and tax revenue, and $100 billion is hard to say no to.

Why 3,000 megawatts is the number that matters

Forget the dollar figure for a moment. The number that determines whether Project Crossroads actually happens is 3,000 megawatts. That is roughly the output of three large nuclear reactors, dedicated to one campus. Power, not money, is the binding constraint on AI data centers in 2026, and every major project in the news this month runs into the same wall.

The pattern is now familiar. Microsoft walked away from Britain's biggest AI datacenter because the grid had no power to spare. NextDC crashed 44 percent in a day because power delays hit its Melbourne plans. Denmark pushed datacenters to the back of the grid queue as requests hit 60 gigawatts. Project Crossroads is bigger than all of them, which means its power problem is bigger too.

Water is the second constraint. Seven million gallons a day for cooling is a serious draw on local resources, and it will attract scrutiny from residents and environmental groups. Data center developers have learned to talk about jobs first and water second. Communities are learning to ask about water first.

What $100 billion actually buys

A $100 billion figure over 20 years needs unpacking. It is not a check written tomorrow. It is a phased investment, spread across 21 buildings, contingent on permits, power agreements, and customer demand. If AI demand cools, later phases slow or stop. If demand keeps growing, the campus fills faster than planned. The number is a ceiling, not a promise.

That phasing is also the project's strength. Building 21 data centers at once would be reckless. Building them in sequence, each one justified by actual customer contracts, is how the data center business has always worked. The question is whether the first few buildings get power and water approvals fast enough to keep the project credible.

The Virginia data center boom, and its backlash

Virginia is already the world's largest data center market, concentrated in Northern Virginia, and the boom has spread south. Pittsylvania County is far from the traditional hub, which is part of the pitch: land is cheaper, and local officials are welcoming. But the backlash that hit Northern Virginia, over power prices, water use, and landscape change, travels with the industry. Rural counties get the same arguments, just later.

The tax revenue figure, $48 million a year at full build-out, will be the project's best argument in local debates. The power and water figures will be its opponents' best arguments. Both sides are right about something, which is why these projects now take years of negotiation even after the announcement.

What to watch next

Three milestones matter. First, permit applications, which revealed the project's scope, will now face public scrutiny and potential challenges. Second, power agreements with the local utility will determine whether 3,000 megawatts is a real plan or an aspiration. Third, customer announcements: a data center campus without named anchor tenants is a bet, not a business.

Project Crossroads is best read as a sign of the times, not a finished fact. The AI buildout has reached a scale where single projects rival national infrastructure programs. Whether they all get built depends less on money, which is abundant, and more on power, water, and permission, which are not. Stack Infrastructure has announced the biggest bet yet. Now it has to prove the grid, the water, and the county can actually support it.