Time-to-Power

The duration required to connect a new facility, such as a data center, to an operational electrical power source.

Time-to-power is a strategic operational metric that measures the duration required to secure, permit, and connect a new facility to an operational electrical power source. While historically a standard project-management timeline for industrial sites, it has recently emerged as a critical competitive metric for energy-intensive industries, particularly artificial intelligence data centers and advanced manufacturing plants. To calculate or observe time-to-power, developers track the timeline from initial site acquisition or project design to the date the facility receives its full, continuous megawatt (MW) allocation. This includes the time spent negotiating utility interconnection agreements, constructing substations, and securing regulatory permits. Alternatively, companies may bypass traditional utility timelines by deploying on-site, behind-the-meter generation assets, which can significantly accelerate the schedule. For investors, time-to-power is a vital indicator of a company's speed-to-market and capital efficiency. In hyper-growth sectors, hardware like advanced GPUs depreciates rapidly, meaning a delay of even a few months in powering a facility can result in millions of dollars in lost revenue and stranded capital. Consequently, companies that can minimize this metric through creative energy sourcing or strategic site selection often command a premium valuation.

A data center developer is evaluating two expansion options. Option A relies on a traditional utility grid connection, which requires building a new substation and waiting for regulatory approvals, resulting in a time-to-power of 36 months. Option B utilizes on-site solid-oxide fuel cells and local solar arrays, allowing the facility to be fully energized in 12 months. By choosing Option B, the developer reduces its time-to-power by 24 months (36 - 12 = 24 months), allowing it to generate lease revenue two years earlier.