24/7 Renewables? A Reality Check on IRENA’s “Firm” LCOE

IRENA just released 24/7 Renewables: The Economics of Firm Solar and Wind, promoting a new “Firm Levelized Cost of Electricity” (F-LCOE) metric. The pitch: overbuilt solar and wind plus batteries can deliver smooth, round-the-clock power competitive with fossil fuels — perfect for power-hungry data centers and the AI boom. The report opens with an appealing schematic showing variable solar output transformed into a flat, reliable line through overbuilding and storage.

UN Secretary-General António Guterres chimes in about escaping fossil fuel dependence. IRENA says solar, wind, and batteries can now deliver “24/7” power for AI data centers cheaper than fossil fuels.

There’s just one problem: Their definition of reliability is not the one data centers actually use.

The report correctly acknowledges longstanding flaws with traditional LCOE — it ignores system integration costs, balancing needs, and the variability of wind and solar. It rightly stresses that delivering “firm” power requires extra investments in storage, overbuilding, and flexibility, and that resource quality (sunny deserts or strong wind corridors) is decisive. Without strong sites, no amount of batteries or overbuild fully compensates. But then it redefines key terms. Reliability, defined on page 52, is treated at the asset level as the ratio of annual energy delivered (by renewables + storage) to total annual demand. In plain English: they’re averaging over the whole year. Cloudy weeks, calm wind lulls, or snow-covered panels get smoothed into the annual total — along with any overproduction above demand. That’s not what most people — or hyperscale data center operators — mean by “24/7.”

“Firm” gets similarly loose treatment. IRENA models scenarios targeting 80–95% annual energy coverage (their definition of reliability). In plain terms: the system can fall short for individual hours — or even days during poor weather — so long as the yearly total hits the target. This is an energy-coverage metric, not a guarantee of continuous availability. The report explicitly notes that this “differs from standard concepts of power system reliability,” which focus on adequacy (meeting peak demand at all times) and security (resilience to disturbances).

IRENA’s 80–95% annual energy coverage falls far short of the 99.671–99.995% uptime required by Tier 1 and Tier 4 data centers. A 95% figure can mask dozens or hundreds of hours of shortfall over a year—completely incompatible with the near-zero-downtime expectations of hyperscale AI facilities.

---

The Key Admission

After the hype, the report concedes: “At moderate levels — 80% to 90%…. Beyond this threshold, costs rise non-linearly: each additional percentage point of reliability requires disproportionately more storage or overbuild.” Pushing from 90% to 95% already drives costs up sharply. Reaching the 99.99%+ availability data centers demand would require even more extreme measures — or, as IRENA itself states, “dedicated back-up strategies.”

Crucially: “Users with the most demanding supply requirements — including data centres, hospitals and precision manufacturing facilities — typically require higher availability standards and will need dedicated back-up strategies…”Translation: For true data-center-grade reliability, you’ll still need gas, diesel, or other dispatchable backup. The “firm renewables” solution isn’t actually firm enough on its own.

---

Cost Claims in Context

IRENA presents F-LCOE figures showing solar + storage in prime locations becoming competitive with new coal or gas (e.g., $54–82/MWh in high-resource areas by 2025 using 4-hour lithium-ion batteries). These reflect real progress in cost declines — no one disputes falling battery and renewable prices. But the numbers are project-level estimates using a simplified energy-based reliability metric, apply best to high-quality resource sites, and still require backups for the most critical loads. Real-world factors — grid integration, transmission, multi-day weather events, and full system costs — add further layers. The New York Times ran with the optimistic headline, amplifying the narrative that renewables have solved intermittency.

---

Bottom Line

IRENA’s mandate is to promote renewable energy transitions. The report offers useful modeling on hybrid systems in ideal conditions and highlights genuine cost improvements. But the promotional framing — claiming “24/7 renewables” are ready for data centers without major caveats — overreaches. Wind and solar plus massive storage and overbuild can deliver high capacity factors in the best locations. They are not yet a drop-in replacement for firm, dispatchable power at the reliability levels demanded by modern industry without additional backup. As AI and electrification drive explosive electricity demand, we need honest accounting of full costs and true reliability — not redefinitions that obscure the gaps. Policymakers and investors should look beyond project-level F-LCOE to system-level realities. The energy transition will be shaped by physics and economics, not wishful rebranding.