At the American Public Power Association (APPA) National Conference 2026 in Boston, this topic was present in many conversations: load growth, data centers, affordability, resource adequacy, capital planning, and the need to make better decisions under uncertainty.

That is exactly why forecasting is becoming one of the most strategic capabilities for utilities.

The reason is simple: the future is arriving faster than many planning models were designed to handle.

According to Grid Strategies, the aggregate national load forecast has reached 166 GW projected by 2030, a six-fold increase over the “flat” growth forecast in 2022. In a related analysis on forecasting for large loads, the firm notes that data centers are expected to account for roughly 55% of projected demand growth, with other new large loads accounting for most of the remaining growth.

This makes forecasting much more than a demand estimate.

It becomes a strategic input into where to invest, when to invest, how to finance, how to manage risk, and how to explain decisions to boards, regulators, and communities.

APPA’s own agenda reflects this shift. In its National Conference program, APPA included sessions focused on the risks and opportunities AI poses to utilities, including infrastructure demands and long-term financial health. It also included sessions on public power experiences with data centers and how utilities are managing resource adequacy, system planning, community expectations, reliability, and affordability amid rapid load growth. APPA National Conference Agenda Public Power’s Experiences with Data Centers

That is the real issue.

The traditional forecasting challenge was accuracy. The new forecasting challenge is decision relevance. A utility does not only need to ask: “What is our most likely load forecast?”

It also needs to ask:

What happens if the forecast is wrong? What if large-load projects arrive earlier than expected? What if they are delayed or canceled? What if demand grows in one part of the service territory, but not another? What if infrastructure constraints prevent the utility from serving growth on time? What if capital costs or customer affordability change the investment path?

These questions cannot be answered by one forecast alone.

They require scenario-based forecasting.

As EPRI explains, load forecasting is key for many grid decisions across operational and planning timescales. Improving forecasts can support more efficient investment decisions and grid performance, but forecasting is becoming more complicated due to electrification, extreme weather, changing customer behaviors, and greater uncertainty.

That is the key point: forecasting is not becoming less important because uncertainty is rising. It is becoming more important.

But the purpose of forecasting must evolve.

A forecast should not be a static number that finance, operations, and engineering debate once a year. It should be a shared decision platform that connects assumptions, risks, scenarios, investments, and financial consequences.

This is especially relevant in the context of AI and data centers.

According to EPRI’s Powering Intelligence report, U.S. data centers could consume 9% to 17% of national electricity by 2030, driven by AI, streaming, and cryptocurrency. Even if the final number lands at the lower end of the range, the planning implications are significant.

The challenge is not only total demand. It is location, timing, concentration, interconnection, resilience, and cost allocation.

A Harvard Belfer Center policy brief describes the growth of AI data centers and the U.S. electric grid as a watershed moment, with consequences for grid expansion, reliability, electricity prices, local planning, and policy decisions.

That means forecasting must become more integrated.

Load forecasts must connect with capital plans. Capital plans must connect with financial models. Financial models must connect with rate and affordability scenarios. Operational constraints must connect with board and regulatory decisions.

This is where finance plays a central role.

The CFO does not need to own every technical forecasting model. But finance must help ensure that forecasts become usable for decision-making.

The future of utility forecasting is not one perfect forecast.

It is a governed forecasting capability that helps the organization understand multiple possible futures and the decisions each future would require.

For utilities, forecasting is now strategic infrastructure.

Because when demand, capital needs, affordability, and grid constraints are all moving at once, the quality of the forecast directly affects the quality of the decisions.

Sources