A data center can report 100 percent renewable electricity while the electricity physically serving its load at 3 a.m. still comes partly from fossil generation. That gap is a consequence of annual renewable energy accounting, not a flaw in the procurement itself. A growing group of large buyers is pushing toward 24/7 carbon-free energy at their data centers. Delivering on that commitment is difficult, because the ability to procure such energy depends on market structure, procurement options, and grid conditions.
Unlike annual matching, which compares a facility’s total yearly consumption with an equivalent volume of renewable generation, 24/7 carbon-free energy asks whether each hour of consumption can be matched to carbon-free generation in the same period and, ideally, the same grid region. The distinction changes what a data center must buy, where it can be built, and how its claims can be verified.
The Limits of Annual Matching
Most corporate clean energy procurement still uses annual accounting. A data center signs a power purchase agreement with a wind or solar project, or buys certificates representing an equivalent amount of generation. The total clean energy produced over a year matches the site’s annual consumption. The match does not have to occur when the data center is actually drawing power.
This works for carbon accounting but less well for physical supply. A solar-heavy portfolio may produce most output during midday while the data center consumes continuously. At night or during calm periods, the facility buys from the wholesale market, where the marginal plant is often fossil-fired. The annual contract remains intact because certificates can be retired independently of physical delivery.
For some corporate buyers, that gap has become harder to ignore. Under existing Scope 2 accounting rules used in corporate greenhouse gas reporting, annual renewable energy certificates can be used to offset purchased electricity regardless of timing. The rise of 24/7 commitments is partly a response to that limitation. Hourly matching tries to pair clean generation with consumption on a much finer time scale.
What Hourly Matching Actually Requires
Hourly matching requires that for every hour a data center operates, an equal or greater amount of carbon-free electricity is generated or procured within a defined market or grid region. Definitions vary. Some programmes accept firm clean sources such as nuclear, hydro and geothermal. Others emphasize additional wind and solar. Some allow storage to bridge the gap between generation and consumption.
The requirement is more demanding than buying enough energy over a year. A continuously operating data center in a region with weak overnight renewable output needs either local clean generation that runs around the clock, storage to shift surplus renewable production, or a portfolio of assets with complementary output profiles. Where hydro or nuclear capacity is limited, assembling such a portfolio becomes difficult.
The clean portfolio required for hourly matching is often not a one-for-one replacement for an annual contract. The 24/7 Carbon-Free Energy Compact notes that hourly matching can require far more clean generation and storage than annual matching. The shortfall is concentrated in specific hours rather than spread evenly across the year. That is one reason siting decisions now turn on the availability of firm clean generation, not only land cost and connectivity.
Grid Physics and the Marginal Carbon Problem
Time and geography create the central difficulty. Carbon-free generation is not evenly distributed across hours or across the network. A market may have abundant solar on summer afternoons and little wind during winter evenings. At any given moment, the unit serving the next increment of demand may be a gas turbine or coal plant, even when the average grid mix over the year looks relatively clean.
The difference between average and marginal carbon intensity matters. If the objective is physical displacement rather than accounting offset, a data center consuming electricity when the marginal plant is fossil-fired would need either reduced fossil generation at that moment or enough added clean supply and storage to change the dispatch stack. Annual certificates do not address this short-run relationship under an accounting framework.
In some northern markets, winter periods can combine low wind output with negligible solar for days. The IEA’s Energy and AI analysis points to these stretches. They explain why hourly matching can require far more clean capacity and storage than annual matching would suggest. They are a direct consequence of variable resource meteorology and the limited duration of today’s storage fleet.
Storage, Firm Clean Power, and Regional Constraints
Storage changes the arithmetic but does not remove the constraint. Batteries can shift midday solar into evening hours. They struggle to bridge multiday periods of low renewable output. Pumped hydro, where geography permits, offers longer duration, but availability is uneven. Long-duration storage remains limited in commercial deployment, and the multiday duration problem continues to constrain most data center locations.
Firm carbon-free sources become especially valuable under hourly matching, though their weather independence differs by technology. Nuclear and geothermal generation run regardless of weather, while hydro’s availability depends on hydrology, reservoir management and seasonal inflows, so its firmness varies by project and region. This is why some data center developers seek locations with existing nuclear or hydro capacity, or explore agreements tied to these assets. The search for around-the-clock clean supply is beginning to influence contract design, not just renewable procurement volumes.
The difficulty is clearest in regions with few firm clean options. In parts of Asia, where hydro, nuclear and large-scale storage are sparse, hourly matching is a much harder target than annual renewable procurement. The same siting logic is shaping demand in markets with strong hydro or nuclear fleets, where time-matched portfolios are easier to assemble.
Market and Verification Barriers
Most electricity markets were not designed for hourly clean-energy accounting. Renewable energy certificates are typically issued for fixed energy volumes, often without the time or location granularity needed to link a generator to a consumer at hourly resolution. There is generally no central market mechanism to trace a specific clean megawatt hour from source to load.
Time-stamped or granular certificates exist in some pilot projects, but they are not yet standard. The absence of common definitions and tracking systems complicates procurement, auditing and reporting. Without a widely accepted method for verifying hourly matching, buyers may find it difficult to compare offerings or make credible public claims.
Regulators are still working through how hourly matching interacts with grid planning, emissions accounting and market participation. A data center that contracts for firm clean generation may still depend on the grid during outages or periods of low local output. The relationship between voluntary procurement and the physical generation mix remains unsettled among system operators and policy makers.
What the Shift Rewards
As data center electricity demand grows, procurement choices made by a handful of operators branch into grid-scale consequences. Annual matching rewards developers for the cheapest renewable projects wherever resources are strong, regardless of when consumption occurs. Hourly matching rewards portfolios that can deliver clean energy when demand is actually high, including storage, demand flexibility and firm clean sources.
The wider consequence is that procurement choices begin to influence which clean resources get built. By aligning procurement with actual consumption hours, large buyers create demand for clean capacity during periods when grids would otherwise rely on fossil generation. That can support investment in storage, long-duration resources and clean firm generation. It also raises cost and complexity, because buyers must cover low-renewable hours, not just an annual total.
Current experimentation points toward a combination of approaches rather than a single standard. Time-matched power purchase agreements, storage-backed wind and solar, direct contracts with existing nuclear or hydro, and on-site clean generation all appear in early programmes. Demand-side flexibility may also help, with some workloads shifted to periods of abundant clean supply, although core data center operations remain difficult to reschedule.
For data center operators, the choice between annual and hourly matching turns on simplicity and physical credibility. Annual agreements are simpler and less expensive. Hourly matching reflects when and where emissions occur more accurately. Neither approach solves grid decarbonization by itself. The more useful question is what each actually rewards in the electricity system.
References
- IEA — Energy and AI (2025) — analysis of low-renewable periods and data centre clean-energy procurement under hourly matching.
- United Nations — 24/7 Carbon-Free Energy Compact — definitions of hourly matching and analysis showing that hourly matching can require more clean generation and storage than annual matching.
- IRENA — Corporate Sourcing of Renewables: Market and Industry Trends — background on power purchase agreements and renewable energy certificate structures.
- GHG Protocol — Scope 2 Guidance — framework for corporate greenhouse gas accounting and the use of energy attribute certificates.
