Sitewide Project Solutions LLC

Here’s the number that should worry every general contractor building in Texas right now: JLL’s latest North America report puts the state’s underconstruction data center capacity at 6.5 gigawatts, more than Virginia’s 4 GW currently under development and enough to put Texas on track to overtake Virginia as the world’s largest data center market by 2030. Nearly twothirds of all data center construction underway in North America right now is happening outside the traditional hubs and Texas is the single biggest beneficiary of that shift.

That is not a projection built on announcements. It is active construction, tracked against a pipeline that JLL says is almost entirely pre-committed by investment-grade tenants. The top five hyperscale’s alone have announced $710 billion in 2026 capital expenditure.

So the demand side of this story is not in question. What is in question is whether the market can actually build it.

The projects are real. The execution risk is where it gets interesting.

Take Vantage Data Centers’ Frontier campus in Shackelford County: $25 billion, 1.4 GW, ten buildings across 3.7 million square feet, tied directly into the Oracle and OpenAI Stargate buildout. Ground has broken, the first building is scheduled to go live in the second half of this year and the project is expected to employ more than 5,000 people across construction and operations. This is the kind of project where MEP execution, not permitting, not land, decides whether the schedule holds.

Then there is Fermi HyperGrid outside Amarillo: an 11 GW, 18-million-square-foot campus positioned as potentially the largest energy-and-AI complex in the country. The headline numbers are enormous, but for a GC or subcontractor deciding where to commit crews, the headline is not the whole picture. In the span of one week this past April, Fermi’s CEO and CFO both stepped down and the company is now facing a federal securities fraud class action tied to a stock drop following the loss of its first anchor tenant, a $150 million contract canceled in December. Fermi’s market capitalization has fallen from roughly $20 billion at IPO to around $3.4 billion.

I raise this not to criticize an ambitious project, but because it illustrates exactly what gets overlooked when everyone chases the gigawatt figure instead of asking who is actually capitalized to pay for the work. MEP subs and GCs need to underwrite a developer’s financial position with the same rigor they apply to power and water studies because a stalled tenant relationship or a leadership shakeup can freeze a schedule as fast as a failed commissioning test.

The real constraint is the trade base

The construction industry needs roughly 439,000 to 499,000 additional workers in 2026 to keep pace with current demand, according to ITIF and data centers are the single biggest driver of that gap. Electrical work alone accounts for 45% to 70% of total data center construction cost, per IBEW, meaning the electrician shortage is not one trade problem among many; it is the load-bearing constraint on the entire build.

Industry estimates put the number of new electricians needed to meet AI-driven data center demand north of 300,000, against roughly 20,000 electricians retiring every year and an apprenticeship pipeline of four to five years that cannot compress to match hyperscale timelines.

Everyone is covering the gigawatts. Almost nobody is covering the crews.

The downstream effect is already visible. Contractors on data center projects are carrying backlogs approaching eleven months on average, versus roughly eight months for other commercial construction. MEP engineer vacancies are taking over four months to fill. Oracle reportedly shifted completion dates on part of its Stargate buildout from 2027 to 2028, with labor availability cited as a contributing factor, a claim Oracle has publicly disputed, though the broader labor trend across the sector is not in dispute.

What this means for how GCs are choosing subs

A few patterns are becoming clear and they will shape who wins the next wave of Texas awards: Pre-bid relationships now beat the open bid. By the time an RFP goes out on a project this size, scope has usually already been shaped with subs the GC already trusts.

Simultaneous capacity is the real qualifier, not single-site delivery. Campuses coming online now need a partner who can run multiple crews across multiple sites, increasingly with prefabricated MEP modules, without quality slipping on any one of them.

Developer financial and leadership stability is now underwriting-level due diligence, not a courtesy check. The Fermi situation is the clearest recent example of why.

Safety record filters the bid pool before price enters the conversation. Mission-critical environments do not tolerate incidents and GCs are pre-qualifying on safety history first.

Where this leaves the market

Texas is not short on demand. It is short on the electrical and mechanical capacity to execute against that demand at the pace hyperscale’s are asking for. The projects that hit their commissioning dates will be the ones where the MEP partner was chosen for staffing depth and financial diligence, not just the lowest bid.