Analysis examines the challenge of reliably forecasting available grid capacity for datacenter interconnection sites, given variable regional demand, transmission constraints, and multi-year utility planning horizons.

A data center site can pass every land, permitting and community screen and still stall on one critical number: how much grid capacity it can actually count on. The first post in this series examined why a powered site must prove it's buildable, permittable and connectable before it's ready for capital. This piece takes up where that left off—at the connectable stage.

Outside ERCOT, many large loads don't sit in a public interconnection queue, so developers can move deep into site planning without knowing who else is chasing the same local headroom. A nearby substation on a map doesn't answer the real question: how much firm load can that point of interconnection actually support, and by what date? Proximity to transmission is not the same as usable grid capacity.

A meaningful capacity assessment must start with the project's actual load profile: target size, energization date, ramp schedule, and how much of that load has to be firm versus how much tolerance exists for bridge power. These inputs form the foundation of any evaluation—the same parameters a development team would already maintain for any candidate site.

A single base case understates the risk. It doesn't account for other large loads competing for the same grid capacity or for stressed system conditions that trigger different network upgrades. A sound workflow tests multiple operating conditions side by side: a traditional case built on today's known queue, and a stressed case that assumes more projects ahead in line actually get built.

Consider the same candidate load modeled under both scenarios. In the traditional case: three network upgrades triggered, 209 MW of headroom, and roughly $73 million in upgrade cost. In the stressed case: five network upgrades triggered, 1 MW of headroom, and roughly $128 million in upgrade cost. Same load. Same node. Same target date. The only variable was which projects ahead in the queue were assumed to proceed. That's the gap a single-case estimate hides—the difference between a site with room to grow and one with almost none.

Once the range is established, the logic becomes concrete. How much firm load can the grid support? How large is the gap to the project's full requested load? What upgrades would close that gap, and when could they be in service? That gap defines the behind-the-meter generation requirement—sized for either an interim bridge while upgrades pending or a longer-term part of the power strategy. A real decision also emerges about what that generation does once the bridge period ends: remain isolated, continue supporting the campus, or eventually pursue its own interconnection. The cost spread matters as much as the headroom spread. Moving from traditional to stressed case not only shrinks available capacity but adds tens of millions in potential network upgrade costs. That range belongs in the site's economics from day one, not discovered later as an unwelcome surprise.

BYOG—bring your own generation—can reduce how much a project depends on immediate full grid service, but it rarely eliminates the interconnection question altogether. Grid timing still affects the power ramp. Network upgrade costs still move project economics. On-site generation may eventually need its own generation-side interconnection analysis once it's large enough to export or support neighboring load. For most BYOG configurations, the grid doesn't disappear from the analysis. It becomes a smaller, later part of it.

Before advancing a site, answer four foundational questions: How much firm load can this location support, and what's the credible range under stressed conditions? Which upgrades are triggered at the target load size, and when does each increment of power actually become available? How much bridge or permanent on-site generation does that gap require? Does the grid timeline still line up with the campus development plan?

A site can clear its land, permitting and grid questions and still face one more piece to confirm: whether enough gas can actually reach it, both physically and commercially.