The dependency is unambiguous: solar’s viability for continuous data center supply is entirely contingent on storage scale-out, and that scale-out is not complete
The System Pressure
The current clean energy procurement model for data centers — built on RECs and legacy PPAs — is approaching its limits. A transition toward location-based and time-matched accounting is anticipated across the industry, and when it arrives, intermittent renewable contracts without firm dispatchable backing will no longer satisfy the accounting standard. That transition carries no confirmed regulatory deadline, but the directional pressure is visible in how hyperscalers are repositioning their generation portfolios ahead of it.
The deeper constraint is continuous baseload. Data centers run 24/7. Solar and wind, without firm backing, cannot match that profile. The question for global heads of data center energy is no longer whether to diversify beyond RECs — it is which combination of technologies can deliver firm, low-carbon power at the scale and timing that portfolio expansion demands, before the accounting transition forces the issue.
The Drivers, Dependencies, and Constraints
Solar paired with battery storage has become the first-response architecture. Grid-scale storage continues to grow as more renewable capacity comes online, and behind-the-meter solar-plus-BESS configurations are increasingly competitive with onsite gas turbines in select geographies. Lithium-ion dominates near-term deployments, but the evidence base for longer-duration alternatives — vanadium redox flow, sodium-ion, second-life EV batteries — is still developing at commercial scale. The dependency is unambiguous: solar’s viability for continuous data center supply is entirely contingent on storage scale-out, and that scale-out is not complete.
Nuclear is the second major arc. Hyperscalers have moved from interest to financial commitment. Google has structured a deal with Kairos Power, an MSR developer; Amazon with X-Energy, an HTGR developer; and Meta with both Oklo and TerraPower, both LMFR developers. The spread across reactor types — molten salt, high-temperature gas, liquid metal fast — signals that no single Gen IV design has established commercial certainty. Each commitment represents an option on a technology pathway, not a confirmed supply contract with binding delivery dates. Commercial SMR timelines remain measured in years to decades, not procurement cycles.
A third pathway is moving faster than its profile suggests: fuel cells and carbon capture attached to gas infrastructure. In January 2026, AEP announced a $2.65 billion agreement to deploy up to 1GW of Bloom Energy solid oxide fuel cells co-located with AI data centers. Solid oxide fuel cells emit less CO₂ than direct combustion and are designed to eventually transition to hydrogen feedstock — though a commercially defined transition timeline does not yet exist. In October 2025, Google announced its first CCS project through Broadwing Energy in Decatur, Illinois, designed to capture and permanently store approximately 90% of CO₂ from gas generation. Large-scale CCUS projects currently require over five years to develop, and CO₂ storage and transport infrastructure remains thin outside a handful of credible geological basins.
At the edge of the technology map sits space-based solar. In April 2026, Meta signed an agreement with Overview Energy to secure capacity from a system that would collect solar energy in geosynchronous orbit and transmit it to existing ground installations via near-infrared light. The architecture bypasses land acquisition, interconnection queue delays, and nighttime intermittency by design. Meta secured access to up to 1GW of future capacity, with orbital demonstration targeted for 2028 and commercial power delivery anticipated for 2030 — though these timelines are projections, not independently verified commitments, and the technology remains pre-commercial. High launch costs and thermal management constraints at full scale are real; the viable case depends on whether launch economics continue their current downward trajectory and whether regulatory clearance across jurisdictions can be resolved.
Open Dependencies
Several dependencies carry material uncertainty. The economics of solar-plus-storage scale nonlinearly with power density: configurations competitive at 50MW may not close financially at 500MW campuses. SMR commercial timelines remain genuinely unlocked — hyperscaler Gen IV investments are strategic hedges, not operational supply contracts. Green hydrogen, which underpins the fuel cell transition case, is supply-constrained and cost-uncompetitive across most markets today. And CCUS infrastructure for data center applications is geographically narrow; most major data center clusters sit outside practical transport distance from credible storage formations.
Space-based solar adds its own dependency cluster: orbital slot coordination, receiver infrastructure integration with existing ground solar assets, and multi-jurisdictional regulatory clearance. None of these have been resolved to confirmed commercial status.
The Operating Exposure for Global Heads of Data Center Energy
The immediate exposure is procurement architecture. If time-matched accounting becomes the standard, energy teams holding primarily REC portfolios will face reclassification risk on their Scope 2 reporting. That transition would force a re-underwriting of what qualifies as 24/7 carbon-free energy — and most existing portfolios were not structured to survive that test.
The medium-term exposure is technology sequencing. Each pathway described above carries a different lead time, capital commitment structure, and risk profile. SMRs are decade-scale; CCUS is five-plus years; space solar is pre-2030 at the earliest for any material commercial volume. That leaves solar-plus-BESS and fuel cell co-location as the only options with near-term commercial maturity. Energy heads making site-level decisions today need a bridge technology stack that holds while longer-duration solutions mature — without locking capital into assets that become stranded if SMR or space solar timelines slip again.
The scale implication sharpens the urgency. Source analysis projects global installed data center power capacity could exceed 300GW by 2036 — roughly triple current levels. At that magnitude, a diversified generation portfolio is not a sustainability aspiration; it is the only architecture that avoids dangerous concentration risk across a single technology type or single interconnection pathway.
Signals the System Is Shifting
Watch for three early confirmations. First, regulatory movement toward time-matched or location-based Scope 2 accounting in any major jurisdiction — this is the trigger that converts optionality into obligation on a defined timeline. Second, the first SMR developer reaching a binding commercial power delivery agreement with a hyperscaler, rather than an investment or strategic offtake letter of intent; that contract-type shift would move nuclear into the active procurement window rather than the hedge column. Third, if space-based solar advances from orbital demonstration to a grid integration agreement before 2030, it changes the interconnection calculus for sites where land and queue access are the binding constraint — not just the technology calculus.
Sources
- Cyprusshippingnews — Meeting Data Center Power Demand Solar, Nuclear, and Beyond (Link)
