Aug Concentration: AI Power Top-1 Still 45% / Top-3 Still 85% — LBNL Moves the Queue Tip, Not the Stock Ladder

Our Q3 concentration print left a sticky scoreboard: the United States still holds 45% of global data-centre electricity, US+China+Europe still hold 85%, and Gartner’s 36% US cut of 565 TWh remains a different perimeter—not a rewrite of the IEA stock ladder. The August Mid-Year vintage then moved the path clocks—US demand to +1.8% in 2026 / +3% in 2027, services H1 +3%, and LBNL Queued Up 2026 restating US active interconnection at 2,061 GW (−10% y/y) with gas in queue +86% to 253 GW. This late-Aug concentration lens asks the next desk question: did Mid-Year and Queued Up rewrite the top-of-distribution shares, or only re-meter the queues and demand tip around a still-sticky ladder?

The interactive dashboard above is built as a vintage concentration lens. Toggle Scoreboard, Perimeters, Queues & tech, and Pace & local. Use the Lens control on the cumulative curve to flip stock versus growth, the queue Metric control to swap GW stock for YoY %, and the cluster Y-axis control to swap global capacity share for pipeline intensity. The punchline is sticky on purpose. IEA Top-1 remains 45% and Top-3 85%. What did restate is the US interconnection tip—solar still Top-1 of the active book at ~37.5%, but gas is the only major technology rising—and the Mid-Year evidence that services (including data centres) concentrate growth inside a soft US total.

The headline ladder: Q3 vs late-Aug restatement

Read the table as a stability-plus-queue scoreboard. Path and interconnection meters moved. Stock concentration architecture did not. That is the Aug contribution relative to the Q3 concentration companion: Mid-Year and Queued Up re-meter delay and demand tips; the top-of-distribution country shares stay put—once you refuse to confuse a gen+storage queue restatement with a regional electricity census.

Scoreboard: five perimeters, one sticky architecture

Open Scoreboard. The vintage-delta bars show flat Top-1 / Top-3 stock shares beside rising US gas-in-queue, a longer median interconnection clock, and a Mid-Year US demand print that softens slightly versus the prior soft narrative. The scoreboard table separates those objects on purpose: IEA stock, Gartner composition, growth, local intensity, and US queue-tech mix are not interchangeable numerators.

The lens scatter puts stock (45% / 85%), growth (~45% / ~78% US+China as Top-2), US queue tech (~37.5% / ~86% solar / solar+storage+gas), and local intensity (Virginia 25% / Ireland 20%) on one plane. Stock and growth sit in the high Top-3 corner. The queue-tech lens is the Aug-specific addition: concentration inside the interconnection book, not a rewrite of who owns global DC watts. Global averages still hide the politics; Aug path meters did not invent that fact—they sharpened the firm-power bid inside the queue.

Perimeters: why 45% and 36% can still both be true

Switch to Perimeters. The US-share compare panel remains the analytical heart of the theme. IEA’s 2024 DC electricity stock still puts the United States at 45%. Gartner’s June 2026 newsroom path still puts the US at 36% of 565 TWh—about 204 TWh. Those are not competing claims about the same pie. Mixing them into a “US share fell nine points” headline remains a category error. Pair this with the August update for the Mid-Year path narrative and with the US data-center power vs grid capacity post when you need LBNL demand-versus-miles rather than world shares.

The growth donut still shows US ~45% and China ~33% of incremental TWh to 2030—together ~80%. Europe remains a single-digit growth slice. The dual-ledger line keeps IEA central (~950 TWh in 2030) and Gartner (>1,200 TWh) on separate tracks. Do not average them. The gap (~250 TWh) is still a forecast-dispersion meter. It does not move Top-1 / Top-3 shares by itself—but Mid-Year’s US demand path (+1.8% then +3%) raises the stakes of where the next gigawatts land if the higher Gartner path proves closer to right while queues stay multi-year.

The regional stock bars are carried for a reason. Until a superseding IEA period census prints, US 45% / China 25% / Europe 15% is the live Top-3 architecture. Aug did not democratise that ladder.

Queues & tech: the concentration that did restate

Toggle Queues & tech. This is the Aug desk’s distinctive panel set. LBNL Queued Up 2026 Edition restates US active generation plus storage at 2,061 GW—down 10% year-over-year—while the IA backlog sits near 549 GW. Flip the queue Metric control from GW stock to YoY % and the story flips: solar (773 GW, ~37.5% of active) and storage (749 GW, ~36%) still dominate the stock tip, but both are shrinking; gas (253 GW, ~12%) is the only major technology rising, at +86% y/y.

That is the first Aug hinge. Concentration of firm-power bids into the interconnection book can intensify even while the total queue shrinks. A −10% headline on active GW is not deconcentration of AI power risk—it is a composition shift toward the fuels and storage that hyperscale campuses need when intermittents alone cannot clear deliverability in five-hub corridors.

The Mid-Year demand path and H1 sector split sit beside the queue pie for a reason. World electricity demand accelerates (+3.6% in 2026 / +3.8% in 2027). The US softens to +1.8% in 2026 after a mild H1 (+1.0% total), then re-accelerates to +3% in 2027. Inside that soft H1, services (including data centres) print +3.0% while residential falls −1.7%. Growth concentration inside the US demand story is still a data-centre story even when the national aggregate looks calm. Wholesale price geography sharpens a related asymmetry: EU and Japan Q2 wholesale prices jump ~30% on Hormuz LNG pass-through while the US prints flat—another reminder that “AI power stress” is not one global meter.

Pace and local intensity: clocks still outrun campuses

Open Pace & local. Demand clocks (campus ~2.5 years, GPU refresh ~1.5) still outrun grid clocks. What Aug restates is the US median interconnection clock: 5.5 years IR→COD in Queued Up 2026, a half-year longer than the prior ~5-year print carried in earlier theme posts. Transmission in advanced economies still often takes 4–8 years. Completion rates remain brutal historically (~13% of requests from 2000–2020 reaching COD; ~75% withdrawn).

Local intensity bars are unchanged for a reason. Virginia ~25% of state electricity and Ireland ~20% of metered supply are still an order of magnitude above the ~1.5% world average. Six US states remain above the 10% threshold. Northern Virginia still anchors the cluster scatter at roughly 4.9 GW operating IT load and about 13% of reported global operational capacity. Nearly half of US data-centre capacity still sits in five regional clusters. The second Aug hinge: a longer median queue clock plus a gas surge into the same corridors is a worse planning problem, not a calmer one. Firm generation announcements do not clear deliverability by themselves when half the US pipeline still aims at five hubs.

Who is exposed — and what would change the story

Exposed: desks that treat a −10% US queue headline as proof AI interconnection risk eased; soft-landing narratives that read Mid-Year’s +1.8% US demand print without the services +3% tip; interconnection reform debates that quote national generation surplus while Virginia and Ireland intensity stay extreme; capital plans that ignore gas’s +86% queue bid as the firm-power answer to AI load in congested hubs.

Relative winners under current rules: corridors that can site outside saturated US five-cluster footprints without losing latency or fibre depth; utilities and ISOs that clear flexible interconnection and storage packages while gas projects race studies; China growth share that still captures ~one-third of incremental TWh; equipment and EPC suppliers whose lead times bind wherever the same two countries bid ~80% of sector growth.

What would change the story: a superseding IEA period census that drops US stock share well below ~35% and Top-3 below ~70%; multi-year evidence that US pipeline share in existing large clusters falls well below 50%; Virginia and Ireland intensity peaking and then declining; US median IR→COD compressing back toward ~3 years with completion rates well above the historical teens; gas-in-queue growth reversing without simply relocating congestion to Dublin, Singapore, or the Pearl River Delta. None of those clear this late-Aug window. The architecture is sticky; the queue composition and Mid-Year path meters are not.

Caveats and methodology

1. IEA regional stock shares (Top-1 45% / Top-3 85%) and US+China growth share (~80%) are carried from the prior concentration prints and Energy and AI regional cuts. No new IEA period census in this window.
2. Gartner’s 36% US share of 565 TWh is a different perimeter from IEA’s Energy-and-AI stock cut—do not average or chain them into a single “US share path.”
3. LBNL Queued Up 2026 (2,061 GW active) is gen+storage and is not comparable one-for-one to earlier gen-only queue proxies or to IEA’s worldwide stalled connection stock (>2,500 GW, carried).
4. Gas queue +86% is a disclosed LBNL year-over-year print for end-2025 data; it is a firm-power bid into queues, not COD. Completion rates historically remain low.
5. Mid-Year US +1.8% / services H1 +3% are demand-path meters. They do not rewrite regional DC electricity stock shares by themselves.
6. Cluster IT-load and pipeline shares for non-NoVa markets remain estimated public tallies; treat rank order as directional.
7. Dual-ledger 2030 gap (~250 TWh) is forecast dispersion between IEA central and Gartner—not a concentration share.

Primary sources: IEA Electricity Mid-Year Update 2026; LBNL Queued Up 2026 Edition; Gartner newsroom 10 Jun 2026 (carried path); IEA Energy and AI regional shares (carried); theme posts Q3 concentration, prior concentration, August Mid-Year update, Q3 update, research, and US data-center power vs grid capacity.

Bottom line: Late-Aug 202608 does not rewrite the ladder of country shares—it re-meters the tip inside the queues and the US demand path. Top-1 stays 45%; Top-3 stays 85%; US active interconnection restates at 2,061 GW with gas +86% to 253 GW while solar remains the stock Top-1 of the book. Path clocks moved. Stock architecture did not.