40 — What Is Genuinely Accelerating

40 — What Is Genuinely Accelerating

Cross-industry synthesis · 25 sectors · 500 scored trend records · research date 2026-09-15

0. What this document claims, and on what basis

"Hot" here is a defined query against the seed database, not an editorial impression. A record qualifies as accelerating if classification is current or emerging_signal and tags.direction is accelerating. That is 270 of 500 records. The ranked table in §2 further filters on composite_score, which is capped at 40 + 60E where E is the evidence factor — so a loud trend with a thin evidence base cannot enter the table at all. Twenty-five records are currently capped, meaning their raw score exceeded what their evidence supports; the cap binds hardest exactly where attention is highest (T-17-06 raw 83.5 → 64.0; T-13-05 raw 84.8 → 70.0; T-16-02 raw 91.8 → 82.0).

The first and larger part of this document is not a list. Six patterns recur independently across sectors that share no analyst, no source set and in most cases no companies. Their independence is what makes them worth more than any individual trend record. They are set out first; the ranked table follows.

Evidence quality is not uniform across sectors and this is stated rather than smoothed. Sectors 13–18 completed with 5–11 searches each instead of the ~22 budgeted, because the shared search pool was exhausted; their Tier-A verification is sound but their discovery is narrow. The arithmetic shows it: mean evidence factor is 0.70 for sectors 01–12, 0.62 for 13–18, and 0.67 for 19–25; triangulation rates are 77% / 61% / 64%. Every record in sector 22 carries an explicit data_gaps field. Where a pattern below leans on a 13–18 sector, a 01–12 corroborator is cited alongside it.

1. Six cross-sector patterns

1.1 Physical constraints are binding before capital constraints

The defining inversion of 2026. In sector after sector, money is abundant and a physical object, a permit or a person is rationed. This is the reverse of the 2010–2021 regime and it changes what an investment thesis has to prove.

The clearest statement of the inversion is in T-05-03: "capital is abundant and the physical asset is rationed." T-18-01 (90.2) is the same fact expressed as real estate — data-centre vacancy at 1.4% nationally and 0.24% in Northern Virginia while supply grew 33.7%.

Where the pattern does not hold, and this matters. Two high-scoring records point the other way. T-10-05 (79.7) finds that US vehicle volume is set by monthly-payment capacity in a 3.50–3.75% policy-rate environment — a pure credit constraint. T-01-01 (75.8) records the moment capital discipline reasserted itself in AI: Alphabet raised 2026 capex guidance to ~$205bn on 2026-07-22 and the stock fell ~5%, inverting a two-year pattern in which a capex raise read as a demand signal. The honest formulation is that physical constraints bind first in capital-abundant sectors, and that the capital constraint is beginning to re-emerge at the top of the AI stack.

1.2 Courts and sub-federal enforcement, not legislatures, are setting boundary conditions

In 2026 the binding changes to market structure came from judges and state officials. The legislated deadlines, by contrast, slipped.

• 09 Industrial — T-09-01 (80.0). The Supreme Court voided the IEEPA tariff architecture on 2026-02-20 (6–3). The level of protection barely moved, but it was rebuilt on Section 232/301 authorities, and CBP had certified roughly $107bn of refunds by 2026-08-21.
• 12 Retail — T-12-04 (79.5). The US Court of International Trade upheld the de minimis suspension on 2026-08-13 on a privilege-revocation theory that survived the IEEPA reversal. The EU imposed €3/item from 2026-07-01 across 5.9bn items, 91% of them from China.
• 16 Advertising — T-16-01 (57.6). On 2026-09-02 Judge Brinkema rejected DOJ's demand to divest AdX and DFP outright and refused to require open-sourcing of the auction logic, adopting behavioural remedies instead. Three years of sell-side strategy priced on a forced divestiture lost its catalyst in a day (T-16-17).
• 14 Media — T-14-04 (76.8). Federal clearance of the $110bn Paramount/WBD transaction was blocked by a twelve-state attorneys-general suit.
• 05 Energy — T-05-20 (52.2). The DC Circuit vacated DOE's Section 202(c) emergency order for the 1,420 MW J.H. Campbell plant on 2026-09-11, holding that "emergency" requires an immediate reliability crisis and that DOE's reading "invites frequent federal interventions that are unsupported by the statute."
• 07 Finance — T-07-17 (69.0) and T-07-07 (68.0). A Kentucky federal court enjoined the CFPB's Section 1033 open-banking rule; the Bureau shrank from ~1,700 to ~1,300 staff and requested ~$279.5m, and state AGs absorbed the function, several by hiring former CFPB enforcement leadership.
• 24 Publishing — T-24-01 (74.8). Bartz v. Anthropic received final approval on 2026-07-20: $1.5bn covering ~500,000 books at roughly $3,000 gross per work, with a 92.77% claims rate. No court has ruled on training-as-fair-use for a licensed corpus, so a settlement is the only observed clearing price in the entire AI-training-rights market.
• 15 Gaming / 17 Consumer apps — T-15-01 (82.7), T-17-07. The flat 30% app-store commission was dismantled across the US, EU, Japan, China, Korea, the UK and Brazil inside nine months, by a mix of court order, statute and regulator negotiation. Apple's EU schedule from 2026-10-01 is 26%/20%/15% plus a 5% Core Technology Commission; China went to 25%/12% on 2026-03-15.

The control case. T-01-16 (51.5): the most-marketed compliance deadline of the year — EU AI Act high-risk obligations for employment AI, due 2026-08-02 — was deferred to 2027-12-02, a ~16-month slip. The same pattern recurs in food (T-19-09: FSMA 204 pushed from 2026-01-20 to no-enforcement-before 2028-07-20) and in advertising (T-16-18: third-party cookie deprecation, a deadline that moved from 2022 to 2023 to 2024 to 2025 and was then abandoned). Where a court ruled, the change landed inside a quarter. Where a legislature scheduled, it slipped by a year or more. That asymmetry is the single most useful forecasting heuristic this research produced.

1.3 Aggregate figures are diverging from medians, in the same direction, everywhere

Every sector that publishes both a sum and a distribution is now showing the sum rising while the median falls or stalls.

The operating rule, established in sector 11 and confirmed in six others: never publish a sector total without the count and the median beside it.

1.4 The disclosure base is contracting

This is a trend in its own right, and it degrades every other trend the platform can measure.

• 15 — T-15-10 (63.6). Electronic Arts completed its take-private by PIF, Silver Lake and Affinity Partners on 2026-08-04, removing a top-five publisher's quarterly reporting from the public information set entirely.
• 14 — T-14-16 (44.2). Disney's Q3 FY2026 release did not headline Disney+ or Hulu subscriber counts; Netflix stopped reporting quarterly membership. The sector's governing KPI for a decade has been withdrawn.
• 02 — T-02-16 (64.8). Net revenue retention has vanished from seat-based SaaS disclosure. Among the vendors reviewed only Snowflake — consumption-priced, therefore not seat-dependent — published a figure (126%). The disappearance of the metric is the indicator.
• 04 — T-04-16 (44.3). Two years of the SEC cyber rule produced 29 mandatory Item 1.05 filings against 50 voluntary Item 8.01 filings, with a live petition (SEC File 4-856) to rescind Item 1.05. Separately, only 24% of H1 2026 US breach notices disclosed an attack vector, the lowest share ITRC has recorded.
• 20 — T-20-03 (61.7). The SEC voted on 2025-03-27 to stop defending its climate disclosure rules; EPA proposed removing Greenhouse Gas Reporting Program obligations for 46 source categories from a programme covering ~8,000 facilities; NOAA's billion-dollar disasters product has published nothing since 2025-06-12.
• 10 — T-10-19. Baidu's Q2 2026 release reported 28 cities and >240M driverless kilometres but no quarterly ride count, fleet size or unit economics, having previously disclosed ride counts. Disclosure became less specific as the claims became bigger.
• 15 — T-15-05. Microsoft stopped publishing Game Pass subscriber counts after the 34m figure of February 2024.
• 22, 24, 19. Instructure and PowerSchool are private-equity owned and 2U is gone, so the K-12 infrastructure layer no longer files; only Warner of the three major music groups publishes auditable quarterly economics; NASS discontinued selected agricultural data collection programmes in 2025.

The countervailing force, which is real and should be weighted. Public-market entry does the opposite. SpaceX's first quarterly report (T-08-04, 80.1) converted the sector's largest information asymmetry into quarterly fact: Space is $962m of $7.814bn revenue and the only loss-making core segment (−$542m); Starlink is 12.0m subscribers at an ARPU that fell from $85 to $66. The Unitree prospectus and the Agility Robotics S-4 (T-13-11) did the same for humanoids. The net direction is contraction, but the mechanism that reverses it — a listing — is identifiable and datable.

1.5 Revenue growth is price, not volume, across the consumer economy

Distinct from §1.3 and just as consequential for market sizing.

• 25 — T-25-04 (78.0) and T-25-20. Every headline "record travel revenue" figure in 2026 is a price statement. Global RPK +0.2% in July 2026; US passengers −1.4%; Amadeus air distribution bookings −3.7%; fares up ~7.5%.
• 24 — T-24-03 (79.5). Music revenue growth is coming from price, not from listeners.
• 14 — T-14-01 (80.1). Streaming profitability arrived "through price and advertising, not more viewing."
• 25 — T-25-05 (73.6) and T-25-13. Hotel RevPAR +3.4–3.9% into record pipelines; Royal Caribbean net yields +1.2% against 6% capacity growth.
• 14 — T-14-20. Theatrical, as above.

The inverse case is instructive: T-08-05 (85.2), where Starlink doubled subscribers to 12.0m by giving up 22% of ARPU. Volume bought with price is the minority behaviour in 2026 and it is happening in the one segment with a genuine cost-curve advantage.

1.6 Announcement-to-delivery is failing the same test in every sector

• 09 — T-09-02 (80.8). Private manufacturing construction put in place −21.7% year over year to $167.8bn SAAR while reshoring announcement flow did not slow.
• 08 — T-08-07 (84.1). Airbus delivered 475 of an 870-aircraft target through August, implying ~99/month for the rest of the year against 57 achieved.
• 06 — T-06-20. $480bn+ of pledged US pharmaceutical manufacturing investment with no public figure for how much is under construction.
• 18 — T-18-20. JLL reports >66 GW of North American data-centre capacity under construction; CBRE reports 7,481 MW. Roughly ninefold, and neither publishes its inclusion rule.
• 05 — T-05-18 (62.9). Exelon expects 22% of its 65 GW pipeline to materialise; ERCOT suspended its Batch Zero process against 474 GW of requests, ~90% data centres.
• 13 — T-13-19. Unitree shipped 5,632 humanoids cumulatively with <10% of revenue from industrial applications; Agility disclosed $1.8m of 2025 revenue against a $140m operating loss.

This pattern is the bridge to 42-overhyped.md, where it is the dominant failure mode.

2. The ranked table

Top 30 accelerating trends by composite_score. "Why it scores here" names the specific dimensions carrying the score and, where relevant, the dimension holding it down.

Sector distribution of the top 40 accelerating records: energy 5; healthcare, semiconductors, cybersecurity, food, aerospace and travel 3 each; real estate, automotive, VC, AI and retail 2 each. Energy and its physical supply chain is the most concentrated cluster of high-scoring accelerating change in the database, which is consistent with §1.1.

3. Hot and well-evidenced versus hot and thinly evidenced

This distinction is enforced arithmetically, not editorially. The four categories below are exhaustive for the accelerating set.

3.1 Hot and well-evidenced (E ≥ 0.9, triangulated, high confidence)

These can carry a decision on their own. Seventeen of the top 30 qualify: T-05-01, T-03-04, T-03-01, T-05-02, T-15-01, T-11-02, T-01-05, T-19-02, T-02-06 (all E = 1.0), and T-06-01, T-18-01, T-08-01, T-19-03, T-03-06, T-12-08, T-12-04, T-06-02 (E = 0.9). What they share is a primary issuer or regulator series: Census C30, NERC LTRA, FERC dockets, company 10-Qs, EIA STEO, USDA ERS. Where the primary series exists and is current, the evidence factor goes to 1.0 and the cap stops binding.

3.2 Hot and thinly evidenced (E ≤ 0.7, but still scoring high)

These are real and directionally supported but the evidence cap is doing visible work, and they should be cited with the cap stated.

How to read the two lists together. A trend in §3.2 is not less true than one in §3.1; it is less checkable. T-16-02 is almost certainly right — the three-platform share of US advertising is visible in the platforms' own revenue lines — but the specific number circulating comes from one modeller, and the cap encodes that. The correct use is to act on the direction and refuse to quote the figure without the modeller's name.

3.3 Accelerating but deliberately not ranked

Twelve accelerating records score below 60 because adoption or revenue is genuinely absent. T-18-11 (49.9) — interconnection queue position as a financeable real-estate asset — is the most interesting of these: a genuinely novel asset class with two named market examples (Dallas-Fort Worth queue-position financing; Hillsboro take-or-pay at 60–85% minimum utilisation floors) and no transaction series. It belongs in a watchlist, not in a model.

3.4 Sector-level confidence qualifiers

• Sectors 13–18 (robotics, media, gaming, advertising, social, real estate) completed with 5–11 searches each. Their contributions to §1.1–§1.6 above are each corroborated by at least one sector from 01–12: the disclosure pattern by 02 and 04, the physical-constraint pattern by 05 and 09, the aggregate/median pattern by 11 and 12, the courts pattern by 09 and 07. No pattern in this document rests on 13–18 alone.
• Sector 22 carries data_gaps on all 20 records — the only sector where the research contract's gap field is populated universally. Read its contributions as hypotheses.
• Sectors 19–25 show the highest Tier-A source share in the database (88% of cited sources) but the lowest mean composite score (59.5), which reflects a research pattern of fewer, better sources rather than weaker trends.

4. The single most decision-relevant conclusion

If a reader takes one thing from this document: in 2026 the scarce input is physical and the binding rule is judicial. A plan that assumes money buys speed, and that a legislated deadline will arrive on the date printed on it, is wrong on both counts — and the two errors compound, because the physical asset takes 128 weeks to arrive and the rule that was supposed to force the buyer to order it slipped sixteen months.

More research from the library · Explore trend records