Update: Renewables Overtake Coal in Power (33.8% vs 33.0%) as Clean Capex Hits $2.2T

What changed since the EI Statistical Review update

Our August energy-systems update closed on the Energy Institute Statistical Review of World Energy 2026 (calendar 2025): world total energy supply (TES) past 600 EJ (+1.7%), renewables (ex-hydro) the largest TES growth source outside a recession (+3.3 EJ), fossils still 86.2% of primary energy, solar overtaking wind in the power mix (8.7% vs 8.4%), US LNG exports +27%, and batteries at 302 GW (+66%). That post answered the primary-energy flow question after the theme research ledger.

This Q3 refresh answers the next vintage question the official stack actually trades on: does the newest electricity census and capital-allocation print agree that the mix is greening at the margin while trade and security still pull fossil money? Two documents force the rewrite. Ember’s Global Electricity Review 2026 shows all renewables at 33.8% of world generation overtaking coal at 33.0% for the first time in the modern power system, with clean sources meeting all 2025 demand growth and fossil generation falling 0.2%. IEA’s World Energy Investment 2026 puts total energy investment at $3.4 trillion in 2026e (+5%), with ~$2.2T to clean (renewables, nuclear, grids, storage, low-emissions fuels, efficiency, electrification) versus ~$1.2T to oil, gas, and coal supply — almost a 2:1 clean-to-fossil capital split even as LNG investment more than doubles and gas-fired plant orders hit a 25-year high.

The dashboard above is built as a Q3 vintage delta: renewables-vs-coal power shares, demand-growth attribution, the primary-vs-electricity dual ledger, the IEA capex stack with bucket filters, trade/LNG/manufacturing meters, and a capex path with companion bubbles. It is not a second encyclopedia of the EI Review.

The headline table: EI update vs Ember + IEA WEI

Read the first two rows carefully. The primary-energy stock from EI did not get a newer Statistical Review in this window — fossils remain ~86% of TES. What moved is the electricity census (Ember’s all-renewables vs coal crossover) and the capital ledger (IEA’s clean-vs-fossil investment split). Averaging those ledgers into one “energy mix” percentage is how readers get lied to. The theme question — how countries source, mix, and trade energy — now needs three companion meters: primary stock, power flow, and where the money goes.

Renewables overtake coal on the electricity ledger

Ember’s 2025 census is the cleanest electricity-side vintage confirmation of the story the prior EI post only hinted at with solar-vs-wind shares. Renewables (solar, wind, hydro, and other RE) contributed 33.8% (10,730 TWh) of world generation; coal contributed 33.0% (10,476 TWh). Coal generation fell 63 TWh (−0.6%) — the first decline since the Covid year — and dropped below one-third of global generation for the first time in the modern system.

That crossover matters more than another solar percentage point. The prior update correctly celebrated solar overtaking wind (8.7% vs 8.4%) and nearly matching nuclear (8.8%). Ember keeps those peer shares in the same neighborhood and adds the system-level punchline: the renewable bundle is now larger than coal. Solar alone rose a record 636 TWh to 2,778 TWh (+30%) — enough, Ember notes, to displace gas-fired electricity equivalent to all LNG exports through the Strait of Hormuz last year (~550 TWh of gas generation). Ember also flags that both solar and wind are expected to overtake nuclear in 2026, which would retire the last “firm low-carbon peer” framing the August post used.

Toggle the dashboard’s RE vs coal panel. The story is not that coal vanished. Absolute coal TWh remain enormous. The story is that the electricity mix finally printed a structural share flip that primary TES still refuses to show.

Clean power met all demand growth — fossils fell 0.2%

Open Demand growth. Ember’s attribution is the Q3 flow sentence markets should remember: low-carbon generation rose 887 TWh while electricity demand rose 849 TWh (+2.8%). Solar alone met 75% of net demand growth; wind plus solar met 99%. Fossil generation recorded a small fall of 38 TWh (−0.2%) — the first decline since 2020 and only the fifth year this century without a fossil-power rise.

The geography of that halt is as important as the global total. China and India — historically the largest contributors to rising fossil power — both recorded fossil-generation declines in 2025 (China −56 TWh / −0.9%; India −52 TWh / −3.3%) as clean additions outpaced demand. That is a different sentence from the EI primary-energy path, where absolute fossil joules still rose even as the fossil share of TES eased to 86.2%. Power can flatten fossils while primary TES keeps adding oil and gas for transport and industry. The theme’s three-ledger warning still holds; Ember just sharpened the electricity ledger.

Pair this with the global electricity generation mix post when you need country-level stacks rather than the world crossover.

Capital follows two stories: clean almost 2× fossils, LNG doubles

IEA World Energy Investment 2026 is the capital vintage the August EI post lacked. Total energy investment rises to $3.4 trillion in 2026e, a 5% increase from 2025, even under Middle East security shock. About $2.2T goes to the clean stack; about $1.2T goes to oil, gas, and coal supply. Clean is nearly twice fossil supply investment and about 65% of total energy spend. Roughly three-quarters of anticipated 2026 investment was already locked in before the conflict — so this year’s print is mostly pre-committed capital, not a crisis-time rewiring.

Inside the fuel stack the split is sharper still. Oil supply investment falls for a third consecutive year to less than $500 billion (−3%) despite higher prices — capital discipline, depleted exploration portfolios, and Middle East project delays dominate. Natural gas supply investment rises to $330 billion, the highest level in a decade (>~10% growth), powered by US demand and LNG export facilities. Coal supply investment prints its highest level since 2012. On the power side, renewable power projects still take about $665 billion (solar $365B — roughly $1 billion per day — wind $200B, hydro $75B) and about 70% of power-generation investment, even after YoY soft patches from cost deflation and China policy. Nuclear investment exceeds $80 billion annually with close to 80 GW under construction across 15 countries. Gas-fired power investment approaches $120 billion after a 25-year high of 130 GW of plant orders in 2025 — with US data-centre demand a major driver.

The LNG overlay is the trade-map companion. 2025 was a record FID year for LNG export capacity (>~100 bcm sanctioned, requiring ~$80B of multi-year capex), with nearly 90% of sanctioned projects in the United States. IEA expects LNG investment globally to more than double in 2026 from 2025 as over 230 bcm of projects outside the Persian Gulf advance toward peak construction — even as Qatar’s North Field East first train slips and incremental 2026 LNG supply is cut from ~45 bcm toward ~40 bcm. The prior post’s US LNG exports +27% flow print now has a forward capital sentence: the US export lead is being cemented in steel, not only in 2025 cargoes.

Toggle Capex stack and filter by clean / fossil / oil & gas / power. The dual message is intentional: clean dominates the investment numerator, while gas and LNG still win the security-diversification race that oil can no longer claim on a rising-capex path.

Primary vs electricity: do not average the ledgers

The Dual ledger panel exists to stop a bad viral chart. EI’s primary TES still prints fossils near 86%, oil near 33%, coal near 28%, renewables ex-hydro near 6%, nuclear near 5%. Ember’s electricity ledger prints all-renewables 33.8% and coal 33.0%, with solar/wind/nuclear each in the high single digits. Those are not contradictions. Oil is transport-heavy and tiny in power. Coal weighs more in electricity than in primary. “All renewables” on Ember includes hydro and other RE that EI’s ex-hydro primary bucket deliberately excludes.

If you need the import-dependence and LNG-concentration stock map from Eurostat-style meters, keep the energy systems research post beside this update. India still imports about 86% of the oil it consumes; Europe about 75%; China about 73%. Europe and India each import roughly half their gas. Those EI trade exposures did not get a newer Statistical Review rewrite in Q3 — the IEA investment print simply raises the price of remaining exposed while capital tries to buy domestic renewables, nuclear, efficiency, and diversified LNG.

Who is exposed under the new vintage

More exposed to the dual-ledger trap: analysts and policymakers who quote Ember’s renewables-over-coal headline as if primary TES had flipped; importers still on 70%+ oil-import shares when Hormuz-class shocks reprice security; utilities and hyperscalers ordering gas capacity into a world where power fossils just flatlined globally but local resource adequacy still needs firm MW; and coal-plant owners in systems where Ember’s share crossover arrives before retirement schedules do.

Relatively better positioned: solar-plus-storage developers in markets where Ember’s “anytime solar” battery note is already cutting daytime prices; countries that can FID domestic renewables and nuclear faster than they can re-open upstream oil; US LNG project sponsors sitting on the FID wave; and nuclear programs that can convert IEA’s >$80B / ~80 GW UC resurgence into completed GW before Ember’s 2026 solar/wind-over-nuclear expectation lands.

What would rewrite the next update: a new EI Statistical Review that moves primary fossil share by more than rounding; Ember’s 2026 monthly track showing fossil generation rising again; IEA investment revising clean below 1.5× fossils; LNG FID collapse outside the US; or gas-fired order books reversing from the 130 GW spike.

Caveats and methodology

• Primary ≠ electricity. EI TES shares and Ember generation shares use different denominatorsnever average them into one mix.
• Investment is 2026e MER dollars from IEA WEI 2026; ~75% of the year’s spend was locked in before the Middle East conflict. Treat the print as mostly pre-committed capital, not a full crisis-response map.
• Clean investment bundles renewables, nuclear, grids, storage, low-emissions fuels, efficiency, and electrificationwider than “renewable power projects” alone ($665B).
• Oil <$500B is an upper-bound framing from IEA (“less than USD 500 billion”); the −3% YoY is the directional signal.
• LNG FID / UC bcm figures are capacity, not delivered cargoes; Qatar delays cut incremental 2026 supply toward ~40 bcm.
• Gas-fired orders (130 GW) are approvals/orders, not completed plants; the $120B 2026e gas-power investment is the spend companion.
• Battery 302 GW and energy CO₂ +1.1% remain prior EI companionsEmber’s storage note (shifting ~14% of new solar) is qualitative context, not a restated GW census.
• This post is a vintage delta. For the full systems ledger, use the research post; for the EI 2025-year flow print, use the August update.

The shareable takeaway

Versus our EI Statistical Review update, the newest official stack says the electricity ledger and the capital ledger moved faster than the primary-energy stock. Renewables overtook coal in world electricity (33.8% vs 33.0%), clean power met all 2025 demand growth, and fossil generation fell 0.2%. At the same time, IEA prints $3.4T of energy investment (+5%), with clean at $2.2T — almost 2× the $1.2T still flowing to oil, gas, and coal supply — while LNG investment doubles and gas-fired orders hit a 25-year high. Fossils remain ~86% of primary TES. The theme answer is sharper: countries are greening the power mix and the investment numerator while still trading and financing a fossil stock that has not flipped.

Related reading: Energy systems EI update (Aug 2026), Energy systems research, and Global electricity generation mix.