At 3 a.m., a data center with an annual wind contract can be physically served by gas-fired generation while its renewable certificates sit in a registry elsewhere. As data center electricity demand grows, that mismatch has pushed a group of large operators toward carbon free energy for data centers, an approach that matches each hour of consumption with carbon-free generation in the same period and, in stricter frameworks, the same grid region. The idea sounds straightforward, but the practical requirements are more demanding than annual certificate matching.
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 clean energy produced over a year matches the site’s annual consumption. The match does not need to occur during the hours the data center actually draws power.
This works for carbon accounting but less well for physical supply. A solar-heavy portfolio may produce most output during midday, while many data centers operate as continuous loads. At night or during calm periods, the facility buys from the wholesale market, where the marginal unit is often gas- or coal-fired. The annual contract stays intact because certificates are retired independently of physical delivery.
Under the GHG Protocol’s Scope 2 Guidance, renewable energy certificates can be retired against purchased electricity regardless of when that electricity was consumed. A data center can therefore claim carbon-free supply on an annual basis while drawing fossil-fired power for much of the year. Hourly matching arose partly in response to this limitation: it attempts to pair clean generation with consumption on a much finer time scale, aligning procurement with the hours in which load actually occurs.
Why Hourly Carbon Free Energy for Data Centers Is Harder Than Annual Procurement
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 shift surplus generation into hours when renewable output is short.
Because electricity is pooled on the grid, whether a certificate bought from a distant market with a different dispatch profile can satisfy an hourly claim depends on the framework. Programmes that require matching within a defined local market or grid region generally do not accept it. Some draw the boundary at the same balancing area or bidding zone; others allow broader geographic reach where transmission capacity supports imports. The stricter the geographic scope, the more the local resource mix and network constraints shape what is feasible.
The requirement is more demanding than buying enough energy over a year. A data center that operates around the clock 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 24/7 Carbon-Free Energy Compact notes that hourly matching can require far more clean generation and storage than annual matching, with the shortfall concentrated in specific hours rather than spread evenly across the year.
Operational requirements extend beyond procurement. A buyer must forecast its own hourly load, forecast clean generation from its contracted assets, and then match the two through a combination of storage dispatch, certificates and market purchases. Where the buyer is not a licensed market participant, it must rely on a supplier or aggregator to manage the imbalance. This layer of forecasting and settlement is absent from annual certificate matching and is a practical reason hourly programmes remain concentrated among large, technically capable operators.
That concentration is why siting decisions more often turn on the availability of firm clean generation, not only land cost and connectivity. Siting Data Centers: How Electricity Availability and Grid Constraints Shape Location Decisions explains that network access remains a separate constraint. Even where clean resources are available, the time required to study and approve a large load can exceed the lead time for assembling a clean portfolio.
The Marginal Carbon Problem and the Grid
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 to align each hour of consumption with qualifying generation and its attributes, a data center consuming electricity when the marginal unit is fossil-fired faces a mismatch unless enough clean generation or storage is matched to that hour within the relevant framework. Annual certificates do not address this short-run relationship.
In some northern markets, winter periods can combine low wind output with negligible solar for several 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. That is why some data center developers seek locations with existing nuclear or hydro capacity, or explore agreements tied to these assets.
The difficulty is clearest in regions with few firm clean options. In parts of Asia, where hydro, nuclear and large-scale storage are less available, hourly matching is a harder target than annual renewable procurement. The same logic is shaping demand in markets with strong hydro or nuclear fleets, where time-matched portfolios are easier to assemble. This regional variation is a recurring theme in Can Asia’s Power Grids Keep Up with the AI Data Center Boom?
Market Design and Verification Gaps
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. Connection timelines add further strain: Why Data Center Grid Connections Take So Long describes why the process of studying and approving large loads often moves more slowly than procurement teams expect.
What the Shift Rewards
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 practical difference 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. Some operators are also pairing procurement with behind-the-meter assets, a topic covered in How Behind-the-Meter Generation Is Changing the Data Center–Utility Relationship. Demand-side flexibility may 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 the implications for hourly clean-energy procurement by data centres.
- 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.
