In partnership with
⚡ THIS WEEK'S SIGNAL
Coravel launched with a customer already signed — roughly 140 MW of critical IT load across three buildings at a Dallas-Fort Worth campus, plus rights to another 100 MW across two more. The pitch is that one company controlling power, construction, and operations beats coordinating five. That may be true. It's worth noting the sponsor making that argument, BlackRock's GIP, also owns a stake in CyrusOne and is buying Aligned. The integrated model is being validated partly against itself.
━━━━━━━━━━━━━━━━━━━━━━━
📰 MAIN STORY
Eight months from joint venture to signed anchor tenant. ACS Group and Global Infrastructure Partners announced on July 15 that their 50-50 data center JV, first disclosed in November 2025, had formally launched as Coravel with its first major hyperscaler agreement in hand — approximately 140 MW of critical IT load across three purpose-built facilities in Dallas-Fort Worth, with rights to roughly 100 MW more. The platform opens with a 1.7 GW development portfolio representing 1.2 GW of IT load across Europe, the US, and Australia, of which about 150 MW is currently under construction. ACS subsidiary Turner Construction will deliver the campus. Howard Boville, previously president of DXC, is CEO. Delivery timelines were not disclosed, and neither was the customer.
Each party wants something different. ACS is a Spanish construction conglomerate converting an EPC backlog into recurring, contracted infrastructure revenue — the same margin migration every large contractor is attempting right now. GIP wants an asset class with utility-like cash flows, and it wants several of them: GIP's other digital holdings include CyrusOne and Vantage Towers, both with KKR, and it is part of the consortium with AIP and MGX acquiring Aligned Data Centers from Macquarie at roughly $40 billion. The hyperscaler wants one throat to choke and a delivery date it can put in a capacity plan. Notably, it wanted that badly enough to sign with a platform that had no operating history under its own name.
Here's the arithmetic that isn't in the press release. Under the PUCT's proposed rule for ERCOT loads of 75 MW or greater, a large load customer would pay a non-refundable interconnection fee of $50,000 per MW plus 100% of direct interconnection costs. On 140 MW that's about $7 million; on the full 240 MW, roughly $12 million — real money, and a rounding error against a campus of this scale. On the other side of the meter, Oncor serves more than 13 million customers, and TDU charges on an average Texas residential bill rose 44% between 2016 and 2026. Oncor's $47.5 billion capital plan adds roughly $2.00 a month to delivery charges on its own. One side buys certainty for eight figures. The other side gets it spread across a decade of bills.
The numbers behind the thesis: large power transformer lead times now run near 128 weeks on average, three to five years for the biggest units, against 24 to 30 months before the boom; switchgear is effectively spoken for through 2028 at some suppliers, and generator step-up units sit around 144 weeks. Of roughly 16 GW of US capacity announced for 2026, about 5 GW was actually under construction, with Sightline Climate estimating 30–50% of the pipeline gets delayed or cancelled. Meanwhile ERCOT is tracking about 438,000 MW of large load requests, roughly 89% from data centers. That's the moat argument in one line: capital is abundant, land is available, and the copper-and-steel is not.
One caution on the thesis. ABB told investors that its capacity expansions are coming online and that it doesn't consider itself the limiting factor, pointing instead to gas turbines and large power transformers as the tighter constraints. If that's right, part of the moat is narrower than the pitch deck.
Scenario A: ERCOT notifies Batch Zero applicants of their project classifications by August 2026. If the DFW campus lands in an early classification with a credible energization path, the integrated model gets its proof — and Aligned, CyrusOne, and QTS all have sponsors capable of replicating it within a year. Premium pricing for delivery certainty holds.
Scenario B: If the campus falls outside Batch Zero, the wait gets long. Batch 1 applications are expected to open in Summer 2027, with a final transmission plan for the batch expected by Fall 2027. In that case, integration didn't beat the queue — it just made the queue easier to manage — and behind-the-meter gas stays the actual delivery mechanism, with the platform as the wrapper around it.
Watch this: The August 2026 Batch Zero classification notifications. Whether Coravel's DFW campus is on that list is the difference between a validated model and a well-capitalized queue position.
━━━━━━━━━━━━━━━━━━━━━━━
Stop making AI decisions in the dark.
Leadership is asking: are we getting value from AI? Which tools are worth the spend? Where are we exposed? Right now, most teams have no idea.
Harmonic Security Usage Explorer changes that.
You get a complete picture of how your organization uses AI, automatically categorized into custom tasks and use cases.
You’ll see the projects being worked on, who’s using what tools, where AI investments are driving value, and where employees are engaging in risky behavior.
CIOs can rationalize spending and cut wasted licenses. CISOs can pinpoint where risk exists and neutralize it. AI committees can show exactly how their efforts are paying off.
━━━━━━━━━━━━━━━━━━━━━━━
⚡ QUICK HITS
New York Pauses Hyperscale Permits Statewide: Governor Hochul signed an executive order on July 14 directing the Department of Environmental Conservation to pause discretionary state permits for construction or expansion of data centers at 50 MW or more, unless applications were already deemed complete; the pause runs until a Generic Environmental Impact Statement is finished or one year, whichever comes first. Local permits are unaffected. If you are underwriting Northeast sites, the entitlement risk just moved from local to state level. New York Governor
Texas Approves Batch Zero — and the Queue Keeps Growing: The PUCT approved ERCOT's Batch Zero process on June 18, moving large-load review from individual studies to grouped evaluation, with about 438,000 MW in the queue and roughly 89% from data centers. Grouped study means your project's timeline is now partly a function of who else filed alongside you. Texas
The Equipment Constraint Is Not Uniform: Transformer lead times near 128 weeks and switchgear spoken for through 2028 remain the headline, but ABB has publicly said it isn't the limiting factor and pointed to turbines and large transformers instead. Before you pay a premium for a developer's "supply chain moat," ask which specific equipment class they've actually locked. Tech InsiderThe Next Web
The Largest Platform Deal Ever Is Still the Backdrop: Macquarie agreed to sell Aligned Data Centers to a group comprising AIP, MGX, and BlackRock's GIP at roughly $40 billion, the largest data center acquisition on record, with closing expected in the first half of 2026. The same sponsor now anchors two large-scale delivery platforms, which matters for anyone negotiating against either. Data Center Frontier
FERC Turns to Computational Load Standards: Per this week's briefing, FERC directed NERC to develop new reliability standards for computational load integration with filings due by December 31, 2026.
━━━━━━━━━━━━━━━━━━━━━━━
🔧 TOOL / RESOURCE OF THE WEEK
ERCOT Large Load Integration Hub: ERCOT's public page for large-load interconnection, hosting the Batch Zero process documents, submission milestone timelines, FAQ, and the net metering arrangement forms tied to PURA §39.169. With Batch Zero classifications landing in August and Batch 1 not opening until Summer 2027, this is where you find out what your actual timeline is instead of what a developer's deck says it is. → https://www.ercot.com/services/rq/large-load-integration
━━━━━━━━━━━━━━━━━━━━━━━
💬 CLOSING THOUGHT
The integrated platform pitch is good. Fewer interfaces, fewer handoffs, fewer places for a schedule to quietly slip. I've watched enough projects die in the seams between contractors to believe that's real value. What I can't tell yet is whether integration actually shortens delivery, or whether it just puts one accountable name on the same wait.
So — on your projects, has consolidating scope under one delivery partner actually pulled your energization date in? Or did it mostly move the risk around?
