Aug Concentration: GERD Top-1 Still 29.4% China / Top-3 65.2% — CET AI Hits 75% While ¥ Flow Surges +8.1%

Our 2026 concentration lens and Q3 companion answered the distribution question on PPP GERD: Top-1 at 29.4% China, Top-3 at 65.2%, China–US gap collapsed to 0.3 pp, and converter frames that can flip who sits at Top-1. The August theme update then printed the first post-2024 flow year on China’s domestic ledger — ¥3.9262T R&D (+8.1%), intensity 2.80%, basic research share 7.08% — plus NSF Translation CET patent meters. This post asks the desk question that sits between those prints: did the August flow and patent vintages change concentration at the top — or only the tempo and invention sleeves around a sticky tip?

The interactive dashboard above is an August 202608 concentration lens. Toggle HHI / scoreboard, Ladder + Lorenz, Flow + intensity, and Patents + KTI. The punchline is deliberately two-clocked. On carried PPP world shares, nothing material moved: Top-1 still 29.4% China, Top-3 still 65.2%, US+China still 58.5%, gap still 0.3 pp. On flow and invention, the system looks more concentrated in tempo and volume: China domestic R&D grows faster than the plan floor (+8.1% vs ≥7%), CET AI priority families print ~75% Top-1, and USPTO utility applicants still US-led at ~47%. Progress can be knife-edge on dollars and cliff-like on patents in the same month.

The August Top-1 / Top-3 scoreboard

Read the table as a family of market shares plus one flow-year meter, not one slogan. GERD PPP answers who leads world expenditure under the OECD/AAAS converter. CET answers who leads critical-technology priority-family volume. USPTO answers who leads US-granted utility applicants. KTI answers who leads knowledge- and technology-intensive value-added. The China ¥ row answers whether the domestic flow is still accelerating — it is not a substitute for the PPP Top-1 cell.

Analysts who splice ¥3.93T into the 29.4% / 29.1% frame invent a new world share that neither OECD nor AAAS printed. Analysts who ignore CET while quoting the knife-edge GERD gap understate where concentration is most extreme. Analysts who ignore USPTO while quoting CET miss that patent ledgers disagree on who is Top-1.

Carried PPP tip: the knife-edge did not move

Open HHI / scoreboard. Analytical GERD HHI sits near 1,890 — the same mid-band Q3 scored after the NSF rounded → AAAS finer restatement. Top-1 / Top-3 bars put GERD and KTI in a contested duopoly band, publications milder on HHI because RoW is still large, and non-OECD plus CET AI in the extreme club.

The vintage slope panel is the first hinge of this August print. From research → 2026 concentration → Q3 → Aug, GERD Top-1 and the China–US gap freeze after Q3. Toggle the metric to CET AI Top-1 and the line stays pinned near 75%. August did not restate world PPP shares; it asked whether a new domestic flow year and patent meters change the distribution story. They change the sleeves, not the carried tip.

Pair this with the Q3 concentration companion: converter frames still matter for whether China is Top-1 at all under alt-PPP or EXR. This August lens keeps that caveat in the methodology and spends its interactive budget on flow tempo, intensity closing, and patent-ledger disagreement instead.

Flow + intensity: tempo concentrates even when shares freeze

Switch to Flow + intensity. The China 2025 domestic signals are the August-specific board:

• Domestic R&D: ¥3.632T → ¥3.9262T (+8.1%)clears the plan’s ≥7% annual growth floor.
• Intensity: 2.70% → 2.80% on the NBS domestic meter (OECD-comparable China intensity was ~2.7% in the 2024 international frame).
• Basic research share: first print above 7% at 7.08%, with basic spending up about +11.1%.
• Intensity gap to US: narrows toward ~0.6 pp against a carried US intensity near 3.4%.

That is the second hinge. A closing intensity gap is not diversification of world GERD shares. It is evidence that China’s domestic effort is still concentrating relative to its own GDP while the United States remains Top-1 on intensity. The dual-axis intensity path makes the geometry readable: China climbs; US plateaus; the gap line drifts down without crossing zero.

Treat the ¥ ledger as a tempo and composition meter. Experimental development still dominates the composition mix (~82%); basic research crossing 7% is a composition shift inside a still-development-heavy stack — not proof that world science funding democratised.

Ladder + Lorenz: six perimeters, one shape family

Toggle Ladder + Lorenz. Flip GERD, pubs, CET, USPTO, KTI, and non-OECD. The Lorenz-style curves all sit above the equal-share line; how far above depends on the perimeter:

• GERD: Top-1 29.4%, Top-2 58.5%, Top-3 65.2%duopoly floor.
• Publications: Top-1 31%, Top-3 50% (China–US–India).
• CET AI: Top-1 ~75%, Top-3 ~92%cliff, not knife-edge.
• USPTO: Top-1 ~47% (US), Top-3 ~72%oligopoly with a different Top-1 label.
• KTI VA: Top-1 28% (US), Top-3 71%.
• Non-OECD: Top-1 92.3%the extreme carried perimeter.

The donut beside the ladder makes Top-3 vs rest visual: for CET AI, “rest of world” after Top-3 is a thin residual; for GERD, RoW after Top-5 still clears about a quarter of the universe. That is why HHI and Top-1 can disagree in rank order across lenses even when every lens is “concentrated.”

Open the volume × impact scatter on the scoreboard view. China sits far right on publication share and only mid-pack on highly-cited-article rate (~1.3% vs US ~1.7%). Volume concentration and citation-intensity leadership still disagree — the same asymmetry the research ledger flagged.

Patents + KTI: where August concentration is loudest

Switch to Patents + KTI. CET field bars rank China volume share from AI (~75%) through QIST (~48%), semiconductors (~42%), advanced nuclear (~40%), and biotech (~38%). Every field in the data notes US citation leadership even where China leads volume — a volume/impact split that mirrors the publication scatter.

The patent-ledger panel is the third hinge. CET AI Top-1 is China; USPTO utility Top-1 is the United States. Same broad “patent” noun; opposite concentration labels. Desks that quote only international priority-family volume see a China cliff. Desks that quote only USPTO applicant share see a US-led utility franchise (~326k grants class). Both can be reading honest NSF Translation meters. Concentration answers are ledger-conditioned.

KTI splits close the board. On total knowledge- and technology-intensive value-added (~$11.7T), the US remains Top-1 at 28% with China at 25% and EU-27 at 18%. Flip to the manufacturing tip and China concentrates (~$2.4T mfg VA framing); flip to services and the US share jumps (~43% in this mix). Aggregate Top-1 is a services story; production-tip concentration is a China manufacturing story. Averaging them into “KTI is neck-and-neck” without the split is a category error.

Who is exposed — and what would change the story

Exposed: ministries that treat the August ¥3.93T print as a new world GERD Top-1 without PPP conversion; patent strategists who equate CET volume leadership with USPTO franchise leadership; industrial-policy desks that celebrate a 0.6 pp intensity-gap close while CET AI still prints ~75% Top-1; investors who price KTI “near-duopoly” without separating manufacturing tip from services share; multilaterals that model non-OECD R&D as a diversified club when one country is 92% of that ledger.

Relative winners under current meters: China-scale platforms that keep domestic R&D above the 7% growth floor and dominate CET priority-family volume; US services and citation franchises that still lead KTI aggregate VA, HCA rates, and USPTO applicants; labs that convert GERD into cited output rather than article counts alone; smaller high-intensity systems that punch above world-dollar share even when they are not Top-3.

What would change the story: a clean US reclaim (or sustained China lead beyond converter noise) on the next OECD/AAAS PPP restatement; CET AI China share compressing toward the semiconductor ~40% band; USPTO applicant Top-1 flipping away from the US; KTI total VA Top-1 flipping to China without a services offset; non-OECD Top-1 falling well below ~70% as India and others scale R&D dollars, not only papers; China domestic intensity closing the US gap to zero and showing up in OECD-comparable world shares. August shows tempo and patent volume concentrating while the PPP tip stays sticky.

Caveats and methodology

• GERD PPP shares are carried from OECD/AAAS 2024 (China 29.4% / US 29.1% / Top-3 65.2%). August does not invent a 2025 world-share census.
• China NBS 2025 domestic R&D (¥3.9262T, +8.1%, intensity 2.80%, basic 7.08%) is a yuan ledger. Do not splice into PPP world shares without conversion.
• Intensity gap (~0.6 pp) compares China domestic 2025 intensity to a carried US ~3.4% international meterdual-ledger, not a single NSF table.
• CET AI ~75% is an NSF Translation volume anchor; field residuals (QIST/semi/etc.) and USPTO Top-3 rungs are estimated closures.
• USPTO utility applicant share (~47%) and Top-3 (~72%) are estimated from NSF Translation framing around ~326k grantsnot a full WIPO world share.
• Publication and HCA figures follow NSF Indicators; HCA is a rate, not a world HCA market share.
• KTI manufacturing / services splits are analytical overlays on NSF State of S&E 2026 total VA shares; treat tip percentages as estimated.
• HHI values are analytical on disclosed-plus-residual buckets (0–10,000), not official NSF or OECD statistics.
• Non-OECD China share 92.3% is carried AAAS; India+ residual rungs are estimated.
• Cross-agency comparisons (NBS vs OECD vs NSF) are directional; scopes, years, and converters differ.

The shareable takeaway

Measurement and science remain concentrated at the top — and August mostly changed the flow and patent sleeves, not the carried PPP tip. On headline PPP GERD, Top-1 (China) is still 29.4%, Top-3 about 65%, and the China–US gap is still 0.3 pp, with US+China near 59%. On the China domestic ledger, R&D hits ¥3.93T (+8.1%) and intensity 2.80% while the gap to US intensity narrows toward 0.6 pp. On CET AI priority patents, Top-1 is about 75%; on USPTO utility applicants, Top-1 flips to the United States at ~47%. On KTI value-added, Top-1 stays US at 28% via services even as China concentrates manufacturing VA. Progress is still funded and published through a system that looks contested on PPP dollars and top-heavy once you rank patents, pubs, and non-OECD ledgers — with August proving that sticky Top-1 shares can coexist with accelerating domestic tempo.

Related reading: Q3 concentration lens · Concentration lens (2026) · August 202608 China flow + CET · 2022 research ledger.