Update: China R&D +8.1% in 2025 — Intensity 2.80%, Basic Research Crosses 7%, AI Patents ~75%

What changed since the Q3 OECD/AAAS restatement

Our Q3 measurement-science update closed on a converter story, not a new expenditure year. Versus the August NSF State of S&E print, OECD and AAAS restated 2024 GERD so China’s world share printed 29.4% against the United States at 29.1% — an overtake that survived by only 0.3 pp. Business BERD explained the knife-edge (+$100B China vs US in constant 2020 PPP); market exchange rates still put China near half US GERD; alternate PPP sensitivity put it at 90–95%. Non-OECD R&D was 92.3% China. That post answered: did finer shares and sector splits change the NSF headline? Barely on ranking, a lot on how fragile the ranking is.

This August 202608 vintage answers the rewrite condition Q3 itself listed first: what does the next flow year show once China’s domestic 2025 R&D ledger prints, and what do NSF Indicators Translation meters add on critical-and-emerging-technology (CET) patents and knowledge- and technology-intensive (KTI) output? Three information events matter. China’s National Bureau of Statistics / MOST briefings put 2025 domestic R&D at ¥39,262亿 (¥3.9262 trillion), up 8.1%, with intensity at 2.80% of GDP and basic research at ¥2,778亿 — 7.08% of total R&D, the first disclosed year above the 7% threshold. NSF/NSB State of U.S. Science and Engineering 2026 and the Translation thematic report show inventors in China receiving the most international priority patent families in AI, quantum information science, biotechnology, semiconductors, and advanced nuclear — with ~75% of AI priority patents alone. KTI value-added remains a near-duopoly: the United States 28% and China 25% of a $11.7 trillion world total, split so China leads manufacturing ($2.4T) while the US dominates services (43% of world KTI services).

The dashboard above is built as a vintage delta: intensity race (levels vs gap-to-US), China 2025 flow dumbbells, composition stacked bars, CET patent concentration, KTI share pie plus manufacturing/services split, and a volume-versus-impact scoreboard.

The headline table: Q3 2024 restatement → August 2025 flow

Read the table as two columns of different meters, not a single converter. The 0.3 pp PPP gap is still the live international ranking until OECD MSTI or NSF Discovery publish a 2025 GERD year. What August adds is the domestic flow that will eventually feed that international restatement — and the patent / KTI concentration panels Q3 treated only directionally.

The first post-2024 China flow year: +8.1%, intensity 2.80%

Q3’s honest limitation was that almost every “delta” restated 2024. China’s statistical communiqué for 2025 breaks that freeze. Domestic R&D expenditure rose 8.1% to ¥3.9262 trillion. Intensity — GERD as a share of GDP — printed 2.80%, a tenth of a point above the ~2.7% OECD-comparable China intensity that sat beside the US at 3.4% and the OECD-area plateau at 2.7% in the prior posts.

That intensity move is the cleanest answer to the theme’s concentration question on the funding effort axis. China’s domestic intensity is now above the OECD-area plateau even while it remains ~0.6 pp behind the United States’ 2024 international meter. The intensity-race panel in the dashboard shows the climb; the gap-to-US toggle stops at 2024 because a matched US 2025 international intensity is not in this vintage. The point is directional: the catch-up is continuing on the effort ratio, not only on PPP dollar levels that converter debates can erase.

Do not convert ¥3.9262T at market FX into a new “China passed the US again” headline. At roughly $550–570 billion EXR (official English briefings used figures near $569B), the domestic print is still the yuan ledger — the same EXR frame that kept China near half US GERD in Q3. PPP conversion of 2025 will arrive with the next OECD/NSF international vintage. Until then, treat +8.1% and 2.80% as disclosed domestic flow, not a restated world-share overtake.

Basic research crosses 7% — composition, not just scale

The composition story is sharper than the headline growth rate. Basic research expenditure reached ¥2,778亿, up 11.1% — faster than total R&D — and accounted for 7.08% of spending. Ministerial briefings framed that share as the first print above 7%. In a system long criticized (fairly or not) for overweighting experimental development, the basic-research share is the policy-relevant composition meter: is the ledger only scaling applied and experimental work, or is the foundational slice rising with it?

The stacked composition panel shows basic research as a thin but rising wedge. Applied and experimental shares in our module are estimated residuals around the disclosed basic share — labeled as such — because the communiqué’s headline cut emphasizes the basic threshold. Even with that caveat, the YoY gap between basic (+11.1%) and total R&D (+8.1%) is the disclosed signal: composition is shifting toward the research type that feeds long-horizon discovery, not only near-term productization.

China’s 2026–2030 planning language locks in an average annual R&D spending increase of at least 7%. That is not a 2025 outcome meter; it is a forward concentration commitment. If delivered, the domestic flow that just printed +8.1% becomes the floor, not a peak.

CET patents: volume concentrates even when citations do not

Q3’s AAAS scoreboard said China led PCT filings and ICT patenting. NSF’s Translation / State of S&E meters make the CET concentration explicit. Inventors in China were granted the most international priority patent families in AI, QIST, biotechnology, semiconductors, and advanced nuclear in 2024. For AI alone, China received about three-quarters of all international priority patents. That is volume concentration at a level the theme’s publication story already previewed (China 31%, US 12%, India 7% of world S&E articles — half the world from three countries).

The same NSF prose keeps the influence caveat that our theme has carried since the research print: patents granted to US inventors in CET areas remain among the most highly cited globally, and US researchers still produce disproportionate shares of highly cited articles. The USPTO awarded 326,000 utility patents in 2024, 47% to US applicants. Volume leadership and citation leadership are not the same column. The scoreboard scatter encodes that split: China volume / US impact is the modal stance across GERD-adjacent meters once you include patents and papers.

For the core question — where is progress funded and published, and is it concentrating? — August’s answer is harsher on volume than Q3’s converter debate implied. Funding effort (China intensity 2.80%) and CET patent families are concentrating; citation impact is not fully following. That is the same asymmetry the 2022 research ledger flagged for publications, now visible in critical-technology IP.

KTI output: near-duopoly with a manufacturing/services split

NSF’s KTI meters add the translation layer Q3 lacked. World knowledge- and technology-intensive value added totaled $11.7 trillion in 2024. The United States (28%) and China (25%) together clear 53%; the EU-27 follows at 18%. Inside that duopoly, roles diverge: China leads KTI manufacturing at $2.4 trillion VA (and remains the largest KTI goods exporter), while the United States holds 43% of world KTI services VA — larger than the EU-27 (19%) and China (11%) combined.

That split matters for readers who collapse “R&D lead” into a single industrial-policy conclusion. A China lead on PPP GERD and CET patent counts can coexist with a US lead on KTI services and highly cited CET patents. Concentration is real; the form of concentration differs by meter. The dashboard’s KTI panel is deliberately two charts with different units so that manufacturing dollars and services shares are not forced onto one misleading axis.

What would rewrite this update

1. OECD MSTI or NSF Discovery 2025 GERD that moves China/US PPP world shares by more than a rounding restatementthe true sequel to Q3’s 29.4 / 29.1 frame.
2. Matched US 2025 international intensity that shows whether the ~0.6 pp gap to China’s 2.80% domestic print is closing or widening on a common converter.
3. CET citation tables showing US highly cited patent shares recovering in AI / semiconductors after the 2010–2021 declines NSF flags.
4. Basic research share in China stalling below 7% on revision, or rising toward double digits with disclosed applied/experimental cuts.
5. KTI services share erosion for the US below ~35%, or China manufacturing VA growth stalling while services catch up.

Until those print, the live frame versus Q3 is: the PPP overtake debate is unchanged; China’s 2025 domestic R&D rose 8.1% to ¥3.93T with intensity at 2.80% and basic research at 7.08%; AI priority patents are ~75% China; world KTI VA is a 28/25 US–China near-duopoly split by manufacturing vs services.

Caveats and methodology

1. Domestic yuan ≠ OECD-comparable GERD shares. Do not splice ¥3.9262T into the 29.4 / 29.1 world-share frame without official PPP conversion.
2. China 2025 intensity (2.80%) vs US 2024 intensity (3.4%) are adjacent but not identical vintages; the gap is illustrative until a matched year prints.
3. Basic research 7.08% is disclosed; applied/experimental stacked shares in the viz are estimated residuals and labeled accordingly.
4. AI ~75% priority-patent share is disclosed in NSF prose; other CET China shares in the chart are estimated midpoints of “China leads volume” statements.
5. EXR dollar translations of China R&D (~$550–570B in English briefings) are presentation aids, not OECD PPP GERD.
6. Publication HCA meters are carried forward from prior theme posts; this vintage does not re-estimate Scopus-based HCA shares.
7. Planning floor (≥7% avg R&D growth) is a forward target, not a realized 2025–30 CAGR.

Primary sources: China NBS Statistical Communiqué 2025 (R&D section) and china.gov.cn / MOST briefings (5 Mar 2026); NSF/NSB State of U.S. Science and Engineering 2026 and Indicators Translation thematic report; prior theme posts measurement-science-update-2026q3 and measurement-science-update-2026.

The shareable takeaway

Versus the Q3 OECD/AAAS knife-edge (China 29.4% vs US 29.1% on 2024 PPP), August adds the first post-2024 China flow year: domestic R&D +8.1% to ¥3.93T, intensity 2.80%, and basic research crossing 7% at 7.08%. NSF CET meters show AI international priority patents ~75% China even while US inventors retain highly cited patent leadership. World KTI output is a 28% / 25% US–China near-duopoly — manufacturing for China, services for the United States. Progress is still concentrating; the newest vintage says the concentration is now visible in effort ratios, foundational composition, and critical-tech patent families, not only in the converter fight over who edged whom in 2024 PPP dollars.