In most development joint ventures (JVs), the recurring fight is over money: who funds, who dilutes, and who gets paid first. In data center development JVs, the fight is increasingly over power. As hyperscalers, developers, and infrastructure funds pair up, the binding constraint is often how to procure, allocate, and increase megawatts. JV partners aligned on a budget can still deadlock over a single substation. In these deals, one of the core governance questions becomes not who controls the cash, but who controls the power?
Why power decisions lead to conflict
Three categories of decisions drive most disputes in fight for control over power. The first is power procurement: utility interconnection, power purchase agreement (PPA) terms, the mix of grid and behind the meter (BTM) generation, and allocation of change in law and curtailment risk. The second is capacity allocation: how scarce megawatts are distributed among tenants, phases, or the partners themselves. The third is expansion: whether to commit to the next phase, pursue another site, or accept a time-limited utility offer of incremental capacity.
These decisions are often high-stakes, time-sensitive, and zero-sum. A megawatt allocated to one partner’s workload is not available for the other partner’s use. An interconnection slot accepted today may foreclose a different configuration tomorrow. And unlike a typical budget dispute, the window to act is set by the utility and the grid, not the partner’s timetable.
Defining reserved matters
The first line of defense is a well drafted list of reserved matters that require supermajority or unanimous consent rather than a simple board or member vote. The drafting challenge is calibration. Set the threshold too low and a minority partner can hold the venture hostage over routine operations; set it too high and a partner with significant skin in the game loses protection over the decisions defining its investment.
For a power constrained asset like a data center development project, the reserved matters list should be tailored, not generic. In a typical development JV, reserved matters usually tie to decisions that involve significant capital or design discretion. In data center JVs, many of these decisions are time constrained by external parties (utilities, grid operators, equipment suppliers) rather than the partner’s own schedule, making governance delays particularly costly. Also, and particularly during the development phase, the pace and technical complexity of a data center differ significantly from a typical real estate development project.
Matters that deserve heightened consent thresholds in this space typically include execution or material amendment of long-term PPAs and interconnection agreements, commitments to new generation or storage, reallocation of contracted capacity among partners, approval of new anchor tenants, and decisions to draw on or forgo a utility’s incremental capacity offer. In contrast, day-to-day matters, both during the development phase and the operations phase, should remain with management or the operating member to avoid deadlocking ordinary development or operations. During the development phase, this should include routine interconnection queue management (including milestone payments), technical design adjustments within approved parameters, and time sensitive equipment procurement. During operations, it should include redundancy testing, minor power distribution adjustments within existing allocations, and standard utility coordination.
Capacity allocation
Capacity allocation warrants special mention because of conflict of interest issues. As is often the case in this space, when one JV partner is also an anchor tenant, hyperscale offtaker, or competitor in the same compute market, a vote on energized megawatts is not a neutral decision that benefits the JV as a whole, it is an adversarial and often zero-sum process among the partners.
Ideally, the JV agreement removes as much discretion as possible and embeds allocation rules at the outset. That can mean fixed allocation ratios, a pre-agreed priority or waterfall for new capacity, objective triggers for reserved capacity to revert to the pool, and clear pricing for capacity drawn above an allocation. Where discretion is unavoidable, conflicted partner recusal (if possible), independent committee review, or independent expert determination can preserve fairness and reduce the risk that an allocation dispute escalates into a value destroying deadlock or litigation.
Matching the deadlock remedy to the decision
If, notwithstanding the parties’ best efforts to address issues in a JV agreement upfront, the parties reach impasse, the remedy should fit the decision. A tiered escalation ladder that moves a dispute from the operating team to senior executives within defined time frames can resolve a dispute by elevating it, and suits decisions where negotiation is still possible.
For technical or valuation driven deadlocks, such as fair capacity allocation or the reasonableness of PPA pricing, independent expert determination is often faster and better suited than a casting vote because it substitutes neutral knowledge for self-interest.
Blunt instruments, such as buy-sell (shotgun) provisions and other exit triggers, are especially tricky in the data center space. A buy-sell assumes a partner can readily value and exit the venture. But these assets are long lived, capital intensive, and dependent on contracts and interconnection rights that may not survive a change in ownership. A forced exit can trigger PPA change of control provisions, jeopardize an interconnection queue position, or strand a half-built project. Therefore, a deadlock remedy that destroys the power rights the venture was built around is likely not a viable option.
To address the foregoing, JV partners should consider three drafting tips in the deadline space. First, define the trigger before the remedy. Specify the matters, thresholds, and time frames, so the dispute resolution mechanism engages when it should and stays dormant when it should not. Second, match the remedy to the decision type, that is, escalation for negotiable disputes, expert determination for technical and valuation questions, and exit mechanics only as a last resort. Third, protect the grid relationship in every remedy, ensuring that no deadlock breaking mechanism inadvertently forfeits a PPA, interconnection position, or contracted capacity that cannot be replaced on the same terms.
Power scarcity is not likely to go away as a major source of tension in data center JV agreements anytime soon. And in a sector where power is the scarcest input, partners who are proactive upfront in addressing the questions described above will be the ones still building when a dispute arrives.
