GLOBAL POWER SCOREBOARD · 全球电力计分板 · TABLEAU DE BORD MONDIAL — LAST FULL SEASON: 2024 · TIP-OFF: 2026
Fossil-based electricity plays for Team A. Renewables plus all carbon-free electricity — hydro, nuclear, geothermal, storage-firmed clean power — plays for Team B. One grid. One clock. Here is the score, counted in terawatt-hours, dollars and human habits, not in press releases.
Sources: Ember Global Electricity Review 2025 · IEA World Energy Investment 2025 · IRENA 2025 · Global Energy Monitor 2025 — see references
Going into 2026 the picture is not ambiguous, it is split. Money and new hardware have already switched jerseys: clean energy takes two dollars for every one that goes to fossil fuels, and nine out of ten new megawatts installed on Earth are renewable. But the electricity that actually flowed last year was still 59% fossil, because the installed fossil fleet is old, paid for, and permitted to run — and because demand grew about 4.3% in 2024, the fastest in years, pulled by heat, air-conditioning, electric mobility and data centres. Team B is scoring faster than ever and still only drawing level, because the game keeps adding minutes.
Read it as a basketball game at half-time: Team B has the better shooting percentage and all the young players; Team A still leads on the board because it started with a 150-year head start and refuses to sit down. Nothing here is decided by enthusiasm. It is decided by wires, hours and capital cost.
把它当作一场中场休息的篮球赛：B 队命中率更高、板凳上全是年轻人；A 队仍然领先，因为它领跑了一百五十年，而且不肯坐下。决定胜负的不是热情，而是电线、小时级的电量和资金成本。
Voyez-le comme un match à la mi-temps : l'équipe B a le meilleur pourcentage de réussite et tous les jeunes joueurs ; l'équipe A mène encore, forte de cent cinquante ans d'avance et refusant de s'asseoir. Rien ne se décide par l'enthousiasme, tout se décide par les réseaux, les heures et le coût du capital.
Your instinct is right: installation is the front end. Solar and wind are now the fast, cheap, distributed, almost boring part. What is not distributed, not fast and not cheap is everything behind the meter box: transformers, substations, high-voltage lines, market rules, and storage. That is where the mismatch lives, and it is measurable.
Team B is recruiting generators faster than it is recruiting wires, transformers, electricians and hours of storage. That gap, not public opinion, is the score line to watch.
| Symptom / 症状 | Evidence | Where it bites hardest | Status |
|---|---|---|---|
| Interconnection queues并网排队 | ~2,300 GW of generation + storage waiting in US queues; typical wait 4–5 years | USA (all ISOs), Brazil NE, India, Great Britain | jammed |
| Transformer & HV cable lead times变压器与高压电缆交期 | Large power transformers and HVDC cable now quoted in multi-year lead times; order books full | Everywhere; worst for small buyers and EMDEs | jammed |
| Curtailment & negative prices弃风弃光与负电价 | CAISO curtailed several TWh of solar in 2024; Germany saw ~450+ hours of negative day-ahead prices; NE Brazil curtailment escalated | California, Chile, NE Brazil, Iberia, N China, S Australia | warning |
| System strength & inertia系统强度与惯量 | The 28 April 2025 Iberian blackout made grid-forming inverters, synchronous condensers and voltage control a board-level topic | High-share, weakly-meshed grids | new frontier |
| Storage–renewable ratio储能配比 | Storage is scaling fast but from a tiny base; most systems still hold ≪4 hours of clean firm cover | India, Southeast Asia, Africa, Japan | behind pace |
| Distribution-level capacity配网容量 | Rooftop solar, heat pumps and EV chargers hit the low-voltage transformer long before they hit the transmission grid — and this data is rarely public | NL, BE, UK, AU, CA, urban China | invisible |
← swipe the table →
And here is the honest answer to "is the data clear?" — no. Generation and capacity data are excellent and near-real-time. Grid, substation and distribution data are patchy, national, often commercially confidential. We can tell you to the megawatt what was built; we frequently cannot tell you what it can be allowed to deliver.
关于"数据到底清晰吗"，诚实的回答是：不清晰。发电量和装机数据非常好，几乎实时；而电网、变电、配网的数据零散、以国别为界、常常属于商业机密。我们能精确到兆瓦地说出建了多少，却经常说不出它究竟被允许送出多少。
Et la réponse honnête à « les données sont-elles claires ? » est non. La production et la capacité sont bien suivies, presque en temps réel. Le réseau, les postes et la distribution restent fragmentaires, nationaux, souvent confidentiels. On sait au mégawatt ce qui a été construit ; rarement ce qu'il est autorisé à livrer.
Same sport, entirely different games. Below: clean share of generation, what was added, what is actually blocking, and a call-out where a market needs one. A high clean share does not mean a fast market — and a dirty grid is often where a single clean megawatt-hour does the most good.
| Market | Clean share | Recent build | Real blocker | Verdict 2025–2030 |
|---|---|---|---|---|
| China中国 | ~32% | ~278 GW solar + ~80 GW wind in 2024; ~212 GW solar in H1 2025 | Curtailment, market pricing reform, and 94.5 GW of new coal construction starts | on track (volume) at risk (lock-in) |
| European Union欧盟 | ~71% | ~65 GW solar in 2024, then the first flat/declining solar year in a decade in 2025; ~13 GW wind vs ~30 GW/yr needed | Wind permitting, grid capex, negative-price risk for merchant projects | on track (solar) behind (wind, grid) |
| United States美国 | ~41% | Record ~50 GW utility solar + fast storage growth | 2025 rollback of wind/solar tax credits, queues, transformer supply, gas turbine order books sold out to ~2030, data-centre load | strong hardware policy call-out |
| India印度 | ~22% of generation, but 50% of installed capacity is now non-fossil (reached 2025, ahead of target) | ~30 GW/yr renewables, first large storage tenders | Storage, inter-state transmission, DISCOM finances, ~30 GW coal under construction | on track (capacity) behind (energy share) |
| Brazil巴西 | ~88% | Big solar and wind pipeline, now curtailed | Transmission and curtailment without compensation | clean low additionality |
| Japan · Korea · Taiwan · Singapore日韩台新 | ~25–40% | Solar saturating good land; offshore wind repricing and cancellations in Japan in 2025 | Land and sea area, grid interconnection, cost of offshore | CFE-led: nuclear + offshore + imports |
| Southeast Asia (ID, VN, PH)东南亚 | ~15–30% | Indonesia's 2025 plan finally tilts new build heavily to renewables | PPA bankability, grid, captive coal for industry, slow JETP disbursement | decisive decade — behind |
| Middle East (SA, UAE)中东 | <10% (rising fast) | World-record-cheap gigascale solar; UAE's 24/7 solar-plus-storage tenders | Demand growth met partly by gas; grid and desal coupling | fast from low base |
| Africa非洲 | Varies hugely | Only ~4 GW of the world's ~585 GW of 2024 renewable additions | Cost of capital, currency risk, offtaker credit, ~600m people without access, diesel gensets filling the gap | capital-starved, not resource-poor |
| Australia澳大利亚 | ~40% | World-leading rooftop solar per capita; large storage build | Transmission lines, community consent, coal exit sequencing | on track delivery risk |
| United Kingdom英国 | ~60% | Coal-free since 30 Sept 2024 — the first G7 country out | Connection queue reform, offshore wind pricing | on track |
← swipe the table →
You named the misleading default precisely. For a decade the story was: if enough companies buy green power, we win. Volume of claims became the scoreboard. But a certificate bought in a grid that is already 95% hydro, matched annually, for a plant that would have been built anyway, moves almost no carbon. The same money spent on hourly-matched, additional clean firm supply in a coal-heavy grid can move an order of magnitude more. The lever is not how many. It is where, when, and instead of what.
| Tier | What it looks like | Typical markets today | Carbon moved |
|---|---|---|---|
| High impact高影响 | Hourly-matched, additional, new-build; paired with storage; sited in fossil-heavy or congested nodes; funds grid-enhancing tech; enables an early coal retirement | India, Indonesia, Vietnam, Poland, Japan, Korea, Taiwan, South Africa, US MISO / PJM / SPP | High per MWh — displaces coal or gas in the same hour |
| Middle impact中影响 | Long-term additional PPA, annual matching, liquid market, some curtailment exposure | Spain, Texas, Nordics, Chile, Australia, Germany | Moderate — real capital, blunt timing |
| Low impact低影响 | Unbundled certificates in already-clean or surplus grids; retroactive annual claims; projects with no additionality | Norway, Iceland, Brazil, Quebec, parts of the EU GO market | Near zero — accounting, not abatement |
Maturity is a distance, not a badge. A market is ready when it has: hourly settlement data, a functioning PPA contract standard, bankable offtake, storage in the tariff, and a transparent connection queue. Count how many of those five a country has — that is your real readiness score, and most markets score two or three.
成熟度是一段距离，不是一枚勋章。一个市场"就绪"的标准是：有小时级结算数据、可用的 PPA 合同范本、可融资的购电方、把储能纳入电价机制、以及透明的并网排队。数一数一个国家满足了这五项中的几项——那才是真实的就绪分数，而多数市场只有两三项。
La maturité est une distance, pas un label. Un marché est « prêt » s'il dispose de : données horaires, contrat-type PPA, acheteur bancable, stockage rémunéré dans le tarif, et file d'attente de raccordement transparente. Comptez combien de ces cinq critères un pays remplit — la plupart en obtiennent deux ou trois.
| Period | Main job | RE focus | CFE focus (where land is scarce) | Failure mode |
|---|---|---|---|---|
| 2026–2030建管道 | Build the pipes, not just the panels. Double grid investment; make storage standard; kill the queue. | Keep solar at 600–800 GW/yr; rescue onshore + offshore wind economics; distributed solar with local flexibility | Nuclear life extension — the cheapest clean firm MWh on Earth; first SMRs; geothermal; hydro refurbishment | Panels stranded behind a transformer that arrives in 2031 |
| 2030–2035补齐灵活性 | Make clean power firm and hourly. Clean share past 60–70% in leading grids; coal exits in the OECD; unabated gas capped. | Repowering; long-duration storage; interregional HVDC; demand response as a market player | New nuclear reaching cost discipline in Japan, Korea, UK, France, China, UAE, Central Europe; geothermal at scale; CCS only where genuinely abated | Adding clean energy without retiring anything — two fleets, one demand |
| 2035–2050收尾与替代 | Finish the last 20% and electrify everything else: heat, industry, transport, molecules. | Overbuild + curtail deliberately; clean hydrogen only where electrons can't go | Clean firm portfolio: nuclear, geothermal, long-duration storage, cross-border CFE trade | Cheap electricity, unreformed industry — the electrons get clean and the emissions move next door |
Your second question is the sharpest one: score every dollar. A dollar that builds fossil-based electricity with no clear compensating measure plays for A. A dollar that builds renewable or carbon-free electricity plays for B. On that scoreboard the trend is genuinely encouraging — and genuinely insufficient. 2:1 is progress. 1.5°C-aligned pathways need roughly 10:1 by 2030.
| Year | Team B — clean | Team A — fossil supply & unabated power | B : A ratio | Reading |
|---|---|---|---|---|
| 2010 | ≈ $0.3 tn | ≈ $1.0 tn | ≈ 0.3 : 1 | A dominant; B is a niche |
| 2015 | ≈ $1.0 tn | ≈ $1.2 tn | ≈ 0.8 : 1 | Paris year; roughly level |
| 2020 | ≈ $1.2 tn | ≈ $0.9 tn | ≈ 1.4 : 1 | B takes the lead in capital |
| 2025 | $2.2 tn | $1.1 tn | ≈ 2 : 1 | Solar PV alone is the single largest line item in global energy investment |
| 2030 | ≈ $2.6–3.0 tn (stated policies) · ≈ $4.5 tn (1.5°C) | ≈ $1.0 tn · ≈ $0.6 tn | ≈ 2.7 : 1 · ≈ 7–10 : 1 | The gap between "trend" and "aligned" is about $2.3 tn a year |
| 2035 | ≈ $3.2 tn · ≈ $4.5 tn | ≈ $0.9 tn · ≈ $0.4 tn | ≈ 3.5 : 1 · > 10 : 1 | Coal supply capex should be near-zero in aligned pathways |
| 2040 · 2045 · 2050 | ≈ $4–5 tn/yr sustained — IRENA's 1.5°C path averages ≈ $5.1 tn/yr ($150 tn cumulative to 2050) | Declining to maintenance-only; new unabated build ≈ 0 | > 15 : 1 | By then the question is no longer power. It is heat, industry and molecules |
← swipe the table →
Not institutions — behaviours. This is where your third question lands, and it is the most useful one, because behaviour is the only thing a company, a city or a person can actually substitute.
Budget: ≈ $1.1 tn/yr fossil supply & unabated power (2025) · plus ≈ $620 bn/yr of consumption subsidies · plus, on the IMF's broader definition including unpriced damage, ≈ $7 tn/yr
Budget: ≈ $2.2 tn/yr (2025) — of which grids ≈ $400 bn, storage ≈ $60–70 bn. Needed: ≈ $4.5 tn/yr by the early 2030s. Not enough — and mis-weighted toward generation.
Quarter scores, honestly: 2020 A 63 / B 37. 2025 A ~58 / B ~42. 2030 is winnable for B on generation share in the OECD, China and Brazil, and lost on the global average unless grids and storage catch up. 2035–2050 is decided by two things nobody applauds: retirement schedules and transmission permits.
各节比分，说实话：2020 年 A 63 / B 37；2025 年 A 约 58 / B 约 42。2030 年，在 OECD、中国和巴西，B 队有机会在发电占比上取胜；但若电网与储能跟不上，全球平均值仍会输。2035–2050 由两件没人鼓掌的事决定：退役时间表和输电线路许可。
Scores par quart-temps, honnêtement : 2020, A 63 / B 37. 2025, A ~58 / B ~42. 2030 est gagnable pour B dans l'OCDE, en Chine et au Brésil ; perdu en moyenne mondiale si réseaux et stockage ne suivent pas. 2035–2050 se décidera par deux choses que personne n'applaudit : les calendriers de fermeture et les permis de lignes.
Double grid investment to $600 bn+ a year, reform the queue so it clears on merit not on date-of-application, and treat transformer factories and electrician training as climate infrastructure. 80 million kilometres by 2040.
先建电网：把电网投资翻倍，改革并网排队规则，把变压器产能和电工培训当作气候基础设施。
Move from four-hour batteries to eight-plus, add long-duration and geothermal, extend every safe nuclear plant, and build new nuclear where land and sea are genuinely scarce — Japan, Korea, Taiwan, Singapore, the Netherlands, Belgium, Central Europe.
让清洁电源"可依靠"：延长储能时长，用地紧张的地区认真发展核电与地热。
Hourly, additional, location-aware. The Scope 2 revision now under way can turn a decade of low-impact certificate buying into high-impact procurement overnight — same budgets, far more carbon moved.
改会计规则：小时级匹配、额外性、看地点。同样的预算，减碳效果可以差一个量级。
Net coal additions to zero. No unabated gas turbine ordered without a retirement or abatement date attached. Adding clean capacity while adding fossil capacity is not a transition; it is two fleets sharing one demand curve.
停止新的锁定：煤电净增装机归零；新增燃机必须绑定退役或减排时间表。
Africa received roughly 4 GW of the world's ~585 GW of 2024 renewable additions while holding most of the world's unelectrified population. This is a cost-of-capital problem, not a sunshine problem. Guarantees, currency hedges, bankable offtake.
为增长所在的市场融资：非洲缺的不是阳光，是资金成本——需要担保、汇率对冲和可融资的购电协议。
Joining A is betting against physics and against cost curves. Staying purely in B and calling it won is how a decade of certificate-buying produced applause and thin abatement. The only trade that scores is A → B, with the old contract actually terminated: a plant retired, a turbine order cancelled, a gas boiler replaced — not a clean megawatt added next to a fossil one that keeps running.
So: merge into C, on three non-negotiable rules. One, every hour and every node is counted — no annual averages hiding a coal-fired evening. Two, no new unabated capacity without a dated exit. Three, capital follows carbon avoided, not claims made. A merger without those rules is not a team. It is a logo.
Keep attention where the impact is. Fewer, bigger, better-measured moves. We will keep deep-diving and updating this station as the 2025 full-year data lands.
合并成 C 队，但必须带着计分板：按小时按节点计量、新增不减排装机必须有退出日期、资金跟着"真实减碳"而不是"声明"。把注意力放在高影响的地方——更少、更大、更可衡量。2025 全年数据出来后，我们继续深挖更新。
Fusionner en équipe C — à condition de garder un tableau de bord : comptage horaire et par nœud, aucune capacité non abattue sans date de sortie, et des capitaux qui suivent le carbone évité, non les déclarations. Moins d'actions, plus grandes, mieux mesurées.
Electricity is the easy half. The next scoreboard has five columns and much worse data: industrial heat (the 200–1000 °C problem nobody photographs), molecules (hydrogen and ammonia only where electrons genuinely cannot go), flexible demand (AI compute that follows the sun instead of ordering a gas turbine), materials (copper, transformers, minerals, recycling), and land and consent (the permit is the new pipeline). Same two teams. Fewer referees.
Every figure above is traceable to one of these. Latest complete year of global generation data is 2024; 2025 values are estimates; anything after 2025 is scenario, not prophecy. Where I have rounded or approximated, I have said so.