Data center M&A is no longer simply a trade in buildings, leases, and equipment. In a growing number of transactions, what the buyer is really paying for is the right to future megawatts: capacity that has been planned, permitted, or promised, but not yet delivered.

A stabilized facility with contracted revenue and an operating history can be underwritten with familiar tools, whereas a development platform, with its land holdings, pending entitlements, utility commitments and prospective customers, resists that kind of precision. The seller wants to be paid for scarcity and upside.

The buyer hesitates to pay full value today for execution risk it will bear alone tomorrow. Earnouts have emerged as one way to bridge that divide, allowing the parties to close a deal without pretending that the future is already certain.

Why data center deals are different

An earnout is a deferred purchase price mechanism. Instead of paying the entire purchase price at closing, the buyer agrees to pay additional consideration after closing if specified milestones are achieved. In most sectors, those milestones are tied to revenue, EBITDA, or customer retention. In data center transactions, however, value is often created before a dollar of operating revenue is recognized.

Power may be allocated before it is delivered. A hyperscaler may reserve capacity before a building is fully commissioned. A site may be strategically valuable because of grid access, water availability, fiber connectivity, permitting posture or adjacency to an existing campus. The most important earnout metrics in this sector are therefore often operational, commercial or development-based rather than purely financial.

Bridging the valuation gap

These conditions create exactly the kind of valuation gap that earnouts are designed to address. A seller may argue that a site should be valued as if the next 100MW will be delivered, leased and absorbed. A buyer may believe in the opportunity but refuse to pay full value at closing for capacity that still depends on utility work, interconnection, permitting, procurement, construction and customer commitments. Rather than force an all-or-nothing negotiation, the parties can allocate part of the purchase price to future outcomes. If the project reaches the agreed milestone, the seller shares in the upside. If it does not, the buyer has not overpaid for risk that never converted into value.

Milestones that mirror how value is created

The most effective data center earnouts are built around milestones that track how value is actually created in the sector:

• Power milestones may be triggered by execution of a utility service agreement, approval of interconnection, completion of a substation, delivery of a specified number of megawatts, or energization by a target date.
• Development milestones may include zoning approval, building permits, environmental approvals, completion of shell construction, commissioning or certificates of occupancy.
• Commercial milestones may focus on signed leases or service agreements, minimum contracted backlog, customer acceptance testing, preleasing thresholds or expansion commitments from investment-grade customers.
• Operating milestones may include uptime, service-level compliance, ramped utilization, recurring revenue or facility-level EBITDA.

Not every data center transaction calls for an earnout, of course. But when the parties do reach for one, its structure should follow the value being bought. A generic revenue earnout is poorly suited to a transaction where the asset's value turns on power delivery. An EBITDA earnout may arrive too late for a development platform whose most important risks are resolved before operations begin.

A lease-up earnout may work for one asset but not for another where the buyer's thesis is land banking, campus expansion or securing scarce powered shell capacity. Earnouts are most useful when they distinguish between capacity that is merely planned, capacity that is permitted, capacity that is powered, capacity that is contracted and capacity that is producing revenue.

Relevant public transactions

One caveat is worth stating plainly: much of the sector's M&A activity takes place between private parties, such as infrastructure funds, developers and operators whose purchase agreements never become public, so the visible record almost certainly understates how common these structures have become. The deals that do surface in securities filings, however, are telling.

In a 2025 Unifi/Enovum transaction, the parties amended a real estate purchase agreement to reduce the closing price and add a power-based earnout. The buyer agreed to use commercially reasonable efforts to obtain an electric service agreement for at least 99 megawatts from Duke Energy within two years, with an $8 million payment if that milestone was achieved within the initial period, a $5 million fallback payment if achieved in the following year, and an additional per-megawatt payment for capacity above 99MW, subject to a cap.

That structure is notable because it treats power deliverability as a purchase price event rather than a diligence condition. Other transactions show different ways to solve the same pricing problem. TeraWulf's 2025 acquisition of Beowulf E&D subsidiaries included earnout consideration tied to execution of a data center lease for one project, energization of the data hall for another project, and closing of project financing.

Where earnouts go wrong

Earnouts also present real risks. They can solve a valuation problem at signing and create a dispute after closing. That is especially true where the buyer controls the business during the earnout period, and the seller no longer controls the decisions that determine whether the milestone is achieved.

In data centers, post-closing disputes can be particularly complex because delays may be caused by multiple actors over an extended term: the buyer, the seller, a utility, a grid operator, local permitting authorities, equipment vendors, construction contractors, anchor customers, or market conditions generally. A milestone that sounds straightforward in the letter of intent can become hotly contested if the acquisition agreement does not define it with precision.

Drafting matters as much as economics

The agreement should state exactly what must happen, when it must happen, how it will be measured, and who determines whether the condition has been satisfied. It should address whether substitute customers count, whether partial megawatt delivery produces a partial payment, whether delays caused by utilities or government authorities extend the measurement period, and whether the buyer must operate the acquired business in any particular manner.

Sellers typically seek protections against buyers delaying, diverting, underfunding, or deprioritizing the acquired assets. Buyers typically resist covenants that limit their ability to integrate the asset, allocate capital, manage customer relationships or respond to market changes. The best agreements confront that tension and address it directly, rather than papering over it.

Information rights deserve equal attention. A seller waiting for a post-closing payment will want reporting, access to relevant project records, and a mechanism to verify the buyer's calculations. A buyer will want administrative simplicity, confidentiality protections, and clear procedures for resolving disagreements. Where the earnout turns on technical data center milestones, the dispute process may need an expert with industry knowledge rather than a general accountant or arbitrator. The parties should also consider whether the earnout is payable in cash, equity, or a combination, and how tax, accounting, and financing treatment may affect the structure.

The bottom line

A well-structured earnout works for both parties. The buyer avoids paying full value upfront for capacity that may never materialize, while the seller preserves the chance to be paid for upside that is real but not yet fully de-risked. Either way, it is a practical response to a market in which the most valuable asset may be the ability to deliver powered capacity on time.

As data center M&A moves deeper into the AI infrastructure era, purchase price negotiations will increasingly turn on future development, power availability, and customer demand. Earnouts are not a cure-all, and poorly drafted earnouts can become litigation magnets. But when they are tied to objective, sector-specific milestones, they can help buyers and sellers with price uncertainty without walking away from strategically important deals. When scarcity, speed, and execution drive value, that flexibility can matter as much as the headline valuation.