N25 THE REALITY LAYER
Grid Queues Are Product Roadmaps
Interconnection timing determines which compute promises can become customer offers.
IN THIS NOTE · FEBRUARY 2026
A software roadmap can change in a sprint. A power roadmap may depend on utilities, substations, transformers, permits and construction schedules that operate on entirely different clocks.
The queue reaches the customer
Interconnection is often discussed as an infrastructure constraint far upstream from product. In reality it controls when a provider can energize racks, expand a site or make a credible delivery commitment.
A project in the queue is not the same as available power. Position, study status, network upgrades and equipment lead times all affect the path.
Roadmaps need physical dependencies
Product and sales teams should understand which capacity depends on which site milestone. This prevents one delayed component from silently contaminating several offers.
Alternatives such as behind-the-meter supply, phased modules or different geographies can reduce some constraints, but each introduces new reliability, regulatory and financing questions.
Sell the controlled schedule
The winning provider may not have the largest announced pipeline. It may have the clearest path through equipment, power, commissioning and acceptance for the exact workload a buyer needs.
Grid literacy is product literacy once compute growth is constrained by local deliverability.
A queue position is not a delivery date
Interconnection queues describe a process for studying and allocating network access; they do not represent a shelf of finished megawatts. Projects can change size, withdraw, face network-upgrade costs or depend on equipment and permits outside the queue study. For a data-centre buyer, the important object is the site-specific path from application through agreements, construction, energization and proven operation at the required load profile.
The roadmap therefore needs evidence from several clocks. Utility and transmission milestones sit beside land, planning, substations, transformers, switchgear, cooling and compute delivery. A date is credible only when the dependencies that control it are named and their slack is understood. Aggregated pipeline numbers are useful for strategy, but they should never be converted directly into customer-ready capacity without a probability and a critical path.
Plan with scenarios, not one heroic schedule
Power development contains uncertainty that cannot be managed away with a more confident forecast. The product plan should carry a base case, an accelerated case and a delay case, each linked to customer commitments, procurement timing and capital exposure. Options might include phased energization, a second geography, workload migration, behind-the-meter supply or flexible demand. Every option creates its own reliability, regulatory and economic questions, so contingency is a design exercise rather than a footnote.
Sales should expose the controlled portion of the schedule. A customer can evaluate a phased offer if the first tranche, expansion condition and fallback are explicit. The provider can reserve long-lead equipment without pretending that every upstream approval is final. Grid literacy becomes a commercial advantage because it replaces a brittle promised date with a schedule that explains how the service behaves when the physical world follows a different path.
- Attach each capacity offer to a named site and energy milestone.
- Track long-lead electrical equipment alongside GPUs.
- Distinguish queue position from energized capacity.
I would revise this if grid and electrical lead times stopped shaping material AI deployment schedules.
Primary and institutional sources used as the grounding layer. Interpretation and synthesis are Luca's.
01