You keep hearing that clean energy is winning. You also keep hearing that it's a subsidy mirage, that solar only works when the sun shines, and that nuclear is either the answer or a boondoggle. All of that can't be true. So which parts are real?
This is a plain-language translation of what the cost and deployment numbers actually say across three technologies: solar, batteries, and nuclear. No math jargon, no guarantees, and every figure labeled with its year and source so you can check it yourself.
The short version: solar and batteries have improved for a reason you can name and predict, that improvement is measured in real deployment and not just press releases, and the hard part now is less about the panels and more about the wires. Let's walk through it.
The one idea that explains solar: learning curves
If you understand one concept, understand this one. It's called a learning curve (sometimes an experience curve), and it explains most of the solar story.
The idea is simple: for many manufactured technologies, every time cumulative production doubles, the cost per unit drops by a roughly consistent percentage. Not because of a single breakthrough, but because factories get better at making the thing — better yields, cheaper inputs, refined processes, bigger scale. The more you build, the cheaper the next one gets, in a predictable band.
Solar panels have followed this pattern for decades. The learning rate for solar photovoltaics has been estimated at around 20% for every doubling of cumulative installed capacity (Our World in Data), which is unusually steep compared with most energy technologies — estimates vary from roughly 20% to the high 20s depending on the period and cost metric used. Because the world kept doubling installed solar, the price kept falling down that curve. The result is that solar module prices have fallen by roughly 90% since 2010 (Our World in Data, drawing on IRENA), one of the fastest cost declines of any energy technology on record.
Why this matters more than any single headline: a learning curve is a mechanism, not a miracle. It tells you why costs fell and gives a reasonable basis to expect further declines as deployment grows — while being honest that curves can flatten, that raw-material and interest-rate shocks can bend them, and that a global module price isn't the same as an installed cost in your country. Treat a learning curve as a reason for expectation, never a promise.
Keep that mechanism in your back pocket. It's the reason "solar is cheap now" is a durable fact and not a fad.
Batteries: why storage stopped being the dealbreaker
The classic objection to solar and wind is intermittency: the sun sets and the wind drops. For a long time that was a genuine dealbreaker because storing electricity was expensive. That objection is weaker than it used to be, and batteries are the reason.
Lithium-ion battery pack prices have followed their own learning curve. Average pack prices have fallen by roughly 90% since 2010 — reaching about $108 per kilowatt-hour in 2025 (BloombergNEF) — driven first by electric vehicles and increasingly by grid-scale storage. Cheaper cells plus scaled manufacturing mean that pairing solar with a few hours of battery storage has gone from exotic to routine in many markets.
What this changes in practice:
- Grid-scale batteries can now absorb midday solar and release it into the evening peak, smoothing the daily curve that used to be solar's weakness.
- Chemistry is diversifying. Lithium iron phosphate (LFP) has taken large market share for stationary storage because it's cheaper and more durable for daily cycling, even if it stores less energy per kilogram.
Here's the honest boundary. Batteries are excellent at shifting energy across hours. They are not, at today's costs, a practical way to store energy across weeks or seasons. A cheap four-hour battery does not solve a still, cloudy week in winter. That "long-duration" problem is a real, unsolved frontier with many contenders and no proven, deployed-at-scale winner yet. Storage stopped being the dealbreaker for daily balancing; it has not abolished the intermittency problem entirely.
Nuclear: what's actually new vs. what's still hype
Nuclear is where you have to be most careful separating what's deployed from what's announced, because the announcements are loud and the timelines are long.
What's genuinely shifting
- Attitude. After years of retirement, several countries and companies have moved to extend the life of existing reactors and, in a few cases, restart previously shut-down plants — decisions to extend the lifetimes of more than 60 reactors have been taken over the past five years (IEA, 2025). The political and corporate mood toward nuclear has warmed, partly driven by demand for firm, low-carbon power for data centers and electrification; in 2024 Constellation and Microsoft signed a deal to restart the Three Mile Island Unit 1 reactor to supply data centers (U.S. EIA).
- Existing plants are the workhorse. The most concrete near-term nuclear story isn't new designs — it's keeping proven reactors running longer. That's real, low-drama progress you can put in the "proven" bucket.
What's still mostly promise
Small modular reactors (SMRs) get the most excited coverage. The pitch is reactors built in factories from standardized modules, cheaper and faster than bespoke gigawatt plants, riding their own learning curve. The pitch is plausible and worth watching. But as of 2026, the number of SMRs actually generating power on commercial grids is very small, and high-profile projects have been delayed or cancelled over cost — NuScale's Carbon Free Power Project, once slated to be the first U.S. commercial SMR, was terminated in 2023 after its cost estimate jumped from about $5.3 billion to $9.3 billion (Utility Dive).
So the honest framing is: SMRs are an announced category with a handful of demonstrators, not a proven deployed fleet. Treat any SMR cost projection as a forecast, not a track record. We sort exactly this kind of distinction with the stage-of-the-future label on every directory entry — deployed and boring beats promised and shiny.
The grid is the real bottleneck now
If solar and batteries are cheap and nuclear is at least being taken seriously, why isn't the transition faster? Increasingly the answer is the grid, not the generators.
Three overlapping problems:
- Transmission. The best sun and wind are often far from where people live. Moving that power requires new high-voltage lines, and building lines is slow — permitting, land rights, and cross-jurisdiction coordination can take many years, often longer than building the wind or solar farm itself.
- Interconnection queues. In several markets, a large backlog of solar, wind, and storage projects is waiting for permission to connect to the grid. In the U.S. alone, nearly 2,300 gigawatts of capacity — with solar and battery storage each approaching a terawatt (roughly 960 GW of solar and 890 GW of storage) — was sitting in interconnection queues at the end of 2024, and a typical project reaching operation that year had spent about 4.5 years waiting (Lawrence Berkeley National Laboratory). The projects exist; the paperwork and studies to plug them in are the holdup.
- Siting. Local opposition and permitting can stall not just power plants but the transmission and storage that make them useful.
This is why grid infrastructure — hardware, software, and the startups working on interconnection and transmission — is one of the more interesting places to look right now. If you want to follow the companies attacking these constraints, we track them in startups worth watching in energy and grid.
The uplifting reframe: the bottleneck moved. A decade ago the hard problem was "clean energy is too expensive." Now it's "we have to build wires and clear queues." That's a genuinely better problem to have, and it's a solvable one — but it needs building, not just cheering.
Reading the numbers honestly: what the curves don't promise
Learning curves are powerful, and they invite over-claiming. Here's how to stay grounded.
- A curve is a trend, not a law. Past learning rates are the best guide we have, but they can flatten, and input-price or interest-rate shocks can bend costs upward for a while, as happened across parts of the energy supply chain in the early 2020s.
- Module cost is not system cost. A cheap panel doesn't mean cheap electricity at your meter. Installation, permitting, grid connection, and financing (the "soft costs") often dominate, and they don't fall as fast as hardware.
- Cheap-when-running isn't the same as always-available. Solar's low cost applies when the sun is up. Comparing it fairly means accounting for storage and backup, which is exactly where the daily-vs-seasonal storage distinction matters.
- "Proven" and "announced" are different words for a reason. Deployment figures describe what's on the grid. Project pipelines describe hopes. Keep them in separate columns, especially for nuclear.
None of these caveats mean the optimism is fake. They mean it should be specific. "Solar modules got dramatically cheaper along a measurable learning curve, batteries solved daily balancing, and the frontier is now grids and long-duration storage" is a hopeful sentence that also happens to be defensible. That's the kind we're after.
For the broader case that you can be hopeful about climate without waving away the hard parts, read climate optimism that isn't denial. And if you want the quieter wins that rarely make headlines, we round them up in progress you probably missed this year, climate edition.
In Plain English
What this is: a translation of the real solar, battery, and nuclear numbers into ordinary language, with each figure tied to its year and source and each caveat left visibly in frame.
Who it helps: anyone who senses clean energy is improving but can't separate the genuine trends from spin — and wants a mental model, not a slogan.
When to use it: the next time you read a confident headline in either direction. Ask: is this a learning-curve mechanism or a one-off claim? Is this deployed or announced? Does it account for the grid?
What to do next: hold one idea — learning curves explain the cost story; the grid is the current bottleneck — and treat SMRs as promising-but-unproven until the fleet is real.
This post is part of our climate cluster, anchored by the clean-energy and build-more-stuff projects making the future look bright.
Want to see the specific companies, tools, and projects behind these trends? Browse our Climate, Energy & Abundance hub or open The Optimism Directory and start with the climate finds. It's a curated map of who's building, honestly staged by how far along they are.