BlackRock and ACS Launch Coravel: The CRE Read on a 1.7 GW Data Center Bet

On July 15, 2026, ACS Group and Global Infrastructure Partners (GIP), part of BlackRock, launched Coravel, the operating brand for the data center development platform the two first announced in November 2025. The company says Coravel is a vertically integrated platform that brings power strategy, site origination, design, construction, capital, and operations under one accountable partner for hyperscalers and technology companies. It launches with an initial development portfolio of about 1.7 gigawatts spanning the United States, Europe, and Australia, and the partners say a broader pipeline under review exceeds 11 gigawatts. Howard Boville, a former president of DXC and a veteran of IBM, Bank of America, and BT, was named chief executive.

The launch came with a first customer. Coravel says it signed a long-term hyperscale lease covering approximately 140 megawatts of critical IT load across three purpose-built facilities at its Dallas-Fort Worth Metroplex campus in Texas, with rights to roughly 100 megawatts of additional expansion across two more facilities. It did not name the customer. The financial frame traces to the November 2025 agreement, which GIP valued at about 2.0 billion euros on a 100 percent basis, with 1.0 billion euros in cash and up to 1.0 billion euros in earn-outs tied to commercial milestones. Those are the companies’ figures, not PSV findings.

The point is not the megawatts. It is who signed them. BlackRock, through GIP, is one of the largest infrastructure investors in the world, and ACS owns Turner Construction and Dragados, among the largest builders in the United States. When that combination stands up a dedicated data center developer with a hyperscaler already under lease, it confirms that the AI buildout is being underwritten as a real estate and infrastructure business, with the same long-duration, credit-tenant logic that governs any large single-tenant asset. The data center here is the shell, the power interconnect, and the multi-year contracted cash flow, not a software story.

For an operator, that raises the bar on the counterparty side of any data center thesis. Land sellers, power-adjacent landowners, and lenders now sit across the table from vertically integrated platforms that control site, design, construction, and capital in one entity, which changes negotiating leverage, delivery certainty, and who keeps the development margin. It also concentrates absorption: a single 140 megawatt lease at one Dallas campus is a large block of contracted demand that reshapes a submarket’s power and land picture. The read is to know whether you are competing with, supplying, or underwriting a platform of this scale before you price the deal.

Treat a launch like this as a trigger to build a counterparty and submarket exposure file from primary records: the ACS and GIP press releases, the November 2025 transaction terms, the campus’s local permits and utility filings as they appear, and the interconnection queue for the relevant grid. An AI assistant reads each record and returns a cited memo: what capacity is actually contracted versus announced, which facilities are permitted, the power and land already committed at the Dallas-Fort Worth campus, and what changed since the last check, with every figure traced to the document it came from.

The reviewer is the acquisitions or development lead working with a market analyst, and the approval gate is explicit: no land option, supply bid, or underwriting assumption moves on a model summary alone. The same review produces the questions a human sends: to the utility on interconnection and load at the specific site, to the county on permit and entitlement status, and to brokers on standing inventory near a campus that just added a large contracted block. The tool organizes the record. It does not decide whether to compete for the parcel.

The strategic calls stay human. Whether to sell into, partner with, or underwrite against a vertically integrated platform, how to price land near a campus with contracted hyperscale demand, and whether a submarket’s thesis should lean on one developer’s pipeline are principal decisions, not summary outputs. So is the concentration question that runs through every data center brief on this desk: a market that rises on one platform’s capital plan also carries that platform’s cycle and its single unnamed tenant’s demand.

The honest cautions. The 1.7 gigawatt portfolio, the 140 megawatt lease, and the 11 gigawatt pipeline are the companies’ figures, and a development portfolio is not built capacity: construction is phased over years, and an announced pipeline can move with power availability and demand. The customer is unnamed, so the credit behind the anchor lease is not public. The open questions worth tracking are practical and will land in filings and dockets, not the press cycle: which facilities actually break ground at Dallas-Fort Worth, what the interconnection and water agreements require, and how much of the 11 gigawatt pipeline converts from review into signed leases. Verify at the parcel, not the press release.