LCOS remains a useful cost metric, but it does not capture the full system value of storage in increasingly complex electricity markets.
On April 28, 2025, a blackout swept across Spain, Portugal, and parts of southern France, cutting power to roughly 55 million people for as long as 23 hours in some areas. In the months just before the collapse, Spain had about 32,043 megawatts of solar capacity installed — and in the minutes before the grid failed, solar was supplying just over half the country's electricity, with Red Eléctrica's own generation data putting it above 53 percent. That share reflects how far Spain's energy transition had progressed, not a cause in itself; plenty of grids run high shares of solar without incident. Its grid-scale battery storage fleet, by contrast, stood at 28 megawatts.
Do the division and the ratio comes out to roughly 1,144 to 1. Spain had, in effect, built more than a thousand times as much capacity to generate power as it had built capacity to buffer it. The comparison isn't a precise engineering measure of how much storage Spain needed — solar generation capacity and battery power capacity measure different things, and neither captures storage duration on its own. But as a measure of how lopsided the buildout had become between generating power and being able to do anything with it in the moment, the ratio is hard to argue with.
A Question LCOS Was Never Designed to Answer
That ratio is itself the puzzle at the center of this story, and it's worth sitting with for a moment before reaching for an explanation. Levelized cost of storage — LCOS, the metric most developers and investors have used for a decade to compare battery projects — was built to answer one specific question: how much does it cost to store and deliver a megawatt-hour of electricity? It has done that job well. Lazard's LCOS analysis, now in its tenth version and still the industry's most-cited benchmark, is explicit that storage economics vary by application, breaking its estimates into several distinct use cases rather than a single number. Even the report that popularized the metric doesn't claim one figure captures everything.
What LCOS was never built to answer is a different, harder question: what does it cost a system when storage isn't there at all? No single project's financial model carries that cost, because it doesn't live at the project level. It lives at the level of the whole system, and Spain's ratio is what that gap looks like once it's finally forced into view. LCOS is a cost metric. It was never built to be a resilience metric.
Cost itself hasn't been the obstacle. Stationary storage battery pack prices fell to $70 per kilowatt-hour in 2025, down 45 percent from the year before, according to BloombergNEF's annual survey — the sharpest drop of any battery segment the firm tracks, following what was already the steepest single-year decline in lithium-ion pack prices overall since 2017. Storage has gotten dramatically cheaper, fast. The question this raises isn't why more of it wasn't built on cost grounds; it's why a system that had gotten this good at building solar hadn't applied the same urgency to building the thing that buffers it.
Where the Value Actually Sits: Timing, Location, and Grid Services
Part of the answer is that storage value was never going to be captured by a single cost figure, because it depends on three things a levelized-cost calculation doesn't ask about: when the energy is delivered, where it sits on the grid, and what else the system can do besides store and release electricity.
Timing is the most familiar of the three, though it's worth being precise about what it does and doesn't explain here. California's "duck curve" describes an evening problem: solar floods the grid at midday and falls away by early evening, right as demand climbs, forcing other resources to ramp up fast. Spain's blackout happened at midday, not in the evening ramp — it struck at 12:33 p.m., during a period of very high solar output. The relevant timing lesson isn't the duck curve itself; it's the broader point both cases illustrate: the moments when a grid is most exposed shift as its generation mix changes, and a system has to be able to respond adequately at whichever moment that turns out to be, not just the moments planners anticipated.
Location matters just as much, and often gets less attention. Within Spain itself, storage deployment isn't uniform. Castilla-La Mancha, a solar-heavy region in central Spain where photovoltaics account for roughly half the local generation mix, has the country's highest battery storage capacity at 63 megawatts — still a small number in absolute terms, but notably ahead of other regions with comparable solar buildout. The national imbalance between generation and buffering capacity hides real variation underneath: some parts of the grid had at least begun to build out storage alongside their solar capacity, while most had barely started — though, as the blackout investigation would later make clear, storage hardware alone was never the whole picture.
Grid services is where the picture gets more technical, and more directly relevant to what a blackout actually is. A battery doesn't only store energy for later; it can respond to a grid disturbance in milliseconds, providing frequency support, and — in grid-forming configurations specifically, not as a default feature of every BESS — black-start capability to help restart parts of a network after a collapse. Which of those services a system is built to provide is itself a design decision, and it interacts with duration: batteries built for fast, short bursts of response look very different from batteries built to run for many hours. Independent market analyses generally place the crossover somewhere between six and eight hours of duration: below that range, lithium-ion tends to win on installed cost; above it, technologies like vanadium flow batteries increasingly compete or win, since their cost scales more with duration than with power capacity. That's a reminder that "which battery" is itself a question with no single right answer, only a right answer for a specific job.
That doesn't argue for building storage anywhere and everywhere — it argues for recognizing that a single cost-per-megawatt-hour figure was never going to tell a grid operator, or a country, how much of it, and what kind, a resilient system actually needs.
The Blackout: What LCOS-Style Thinking Misses
This is where Spain's ratio stops being an abstract curiosity and starts explaining something concrete. A grid built primarily around solar and wind is, by nature, exposed to two kinds of variability that a project-level cost metric doesn't price in. The first is straightforward: sunlight and wind aren't controllable, and their output can shift within minutes as weather changes. The second is more structural. Conventional power plants — coal, gas, nuclear, hydro — run on spinning turbines that carry physical momentum, and that momentum acts as a buffer, absorbing shocks to grid frequency for the seconds it takes protective systems to respond. Inverter-based resources like solar and battery-connected wind don't have that built-in inertia by default. As they displace spinning generation, something else has to help absorb that kind of shock — and storage, along with synchronous condensers, grid-forming inverters, and flexible generation, is one of several tools that can.
ENTSO-E's Expert Panel — a 49-member group drawn from transmission operators, regional coordination centers, and national regulators across Europe — published its final report on the blackout in March 2026, nearly a year after the event. Its conclusion doesn't center on a shortage of storage hardware. The report describes a combination of interacting factors: voltage oscillations, gaps in reactive power control, inconsistent voltage-regulation practices across different generators, and rapid, cascading disconnections that began in southern Spain. One specific finding stands out: a meaningful amount of reactive power capacity, sitting in equipment already installed on the grid, was available in the run-up to the blackout but wasn't activated. Damián Cortinas, chair of ENTSO-E's board, put it plainly at the report's release, saying the issue wasn't renewable energy but rather "voltage control, regardless of the type of generation." Wind was explicitly ruled out as a cause, and renewable generation in general received comparatively little attention in the report next to the operational and coordination failures.
That distinction matters for how this story gets told. The evidence points less to a hardware shortage than to a coordination failure: tools available to stabilize voltage, including some already installed, weren't activated the way a fast-changing, renewables-heavy grid now requires. LCOS still has nothing to say about that kind of readiness. It prices a battery project's electricity, not the operational procedures, market rules, and coordination systems that determine whether that battery, or any other flexibility resource, actually gets called on when a grid needs it most. Spain's own response reflects this: rather than treating the fix as simply "build more storage," the country updated Operational Procedure 7.4 in June 2025 to formally let solar and battery storage participate in voltage control, with full implementation completed in March 2026. That's a regulatory and operational change, not a purchasing decision — and it's exactly the kind of fix that a cost-per-megawatt-hour metric was never built to prompt.
Battery deployment has also accelerated sharply since the blackout, even if it wasn't the report's central recommendation: installed storage grew 589 percent in the year that followed, from 28 megawatts to 193 megawatts, according to Red Eléctrica. Spain's energy regulator, the CNMC, has separately opened 20 sanctioning proceedings against grid operator Red Eléctrica and several major utilities, including Endesa, Iberdrola, Naturgy, and Repsol, tied to the broader investigation. Even after that growth, Spain's absolute storage capacity still trails well behind Germany, Italy, and the UK. Whatever combination of hardware and operational fixes a grid like Spain's ultimately needs, closing a ratio like 1,144 to 1 was never going to happen in a single year.
The Question Worth Asking Instead
None of this means LCOS should be discarded. It remains the right tool for comparing the cost efficiency of two similar battery projects, and the steady decline in battery prices it tracks is real and important. But Spain's experience is a concrete illustration of what that number can't tell you: it can't tell a grid planner whether the system as a whole can actually call on its resources when a shock hits, because that's not a question about any single project's economics. It's a question about whether the system is built, operationally and not just physically, to use what it has.
That suggests a different question is worth adding to the standard one. Alongside "what does this storage project cost per megawatt-hour delivered," it's worth asking: when this system is under stress, can it actually reach the flexibility it has already paid for? Spain now has an unusually clear answer to how expensive it is to find out the hard way. Most grids would rather not have to learn it the same way.
