2 — The Final 25 Industries: Ranking and Methodology
Phase 1 deliverable · research date 2026-09-15
2.1 What changed from the starting list
The brief supplied 25 starting categories and asked us to validate, merge, split or replace them where the evidence supported it. We made three splits and three merges, netting 25 → 25. Each change is defended by an observed divergence in the research, not by taxonomic preference.
Splits
"AI and software" → AI foundation models & compute infrastructure (01) + Enterprise software & SaaS (02). These two have opposite capital and pricing dynamics and the second is the disruption target of the first. Sector 01 is absorbing the majority of global venture capital and is financing itself increasingly with debt and leases; sector 02 is defending seat-based pricing against consumption metering. Sector 02's research found the cannibalisation everyone predicted for software is actually landing on IT services, not software — a finding structurally invisible if the two sit in one bucket.
"Energy, climate & sustainability" → Energy & power systems (05) + Climate tech, sustainability & carbon markets (20). The thesis was that these two capital flows are diverging. We tested it from both sides and the answer is more interesting than the hypothesis. Sector 05 returned a qualified verdict: supported on drivers and policy (interconnection queue composition inverted — gas +86%, solar −19%, storage −16%, wind −19%; clean-energy credits ended 2026-07-04), but contradicted on deployment volume (a record 86 GW of 2026 US additions, led by 43.4 GW of solar). Sector 20 returned a timing correction: policy and capital allocation have already separated (CCUS is 0.23% of clean-energy investment; the +55% climate-VC headline is carried by data centres at ~34% of total), while installation volume is a lagging indicator. The split is validated, and the disagreement between the two dossiers about when divergence becomes visible is itself a finding — one that a merged sector would have averaged away.
"Industrial, manufacturing, logistics, aerospace & defense" → Aerospace, defense & space (08) + Industrial manufacturing, supply chain & logistics (09). Five sectors in one bucket, on two incompatible driver sets: government procurement and geopolitics versus trade, tariffs and automation. Post-SpaceX-IPO, 08 is also a distinct public capital market with new Tier A disclosure that did not exist a year ago.
Merges
Quantum computing & frontier science → folded into Semiconductors (03). "Frontier science" was a residual bucket with no coherent market structure; it was dissolved (fusion → 05, bio-tools → 06, advanced materials → 09, space science → 08). Quantum was merged as a carved-out subindustry. The sector-03 analyst pushed back, arguing the merge is "justified on 2026 economics but analytically lossy," and specifically that routing post-quantum cryptography to sector 04 separates quantum hardware from its most monetisable near-term consequence. We are keeping the merge for Phase 1 and recording the objection.
Consumer technology & apps → distributed (devices → 03, consumer app business models → 17).
Music & audio + Publishing, books, comics & webtoons → Publishing, music & rights-based IP (24), merged on the rationale that both sectors' dominant 2026 trend is the same one: who owns and is paid for IP used in AI training. The sector-24 analyst tested this rationale and recommends reversing it. The shared legal substrate is real, but the halves sit at opposite ends of the AI event — book publishing has been paid $1.5bn, music has been paid nothing disclosed — and news publishing's dominant trend is inference-side traffic collapse, not training rights. Recommendation carried to Phase 2: split into music/rights, book & news publishing, and comics/webtoons, keeping "AI training-data rights" as a cross-cutting trend line rather than a sector. That is the correct resolution: it was a trend masquerading as a sector boundary.
Considered and excluded
Telecom & connectivity is the strongest omission and should be the first addition in
Phase 2; it was excluded only because the slots were full, with satellite routed to 08 and
network silicon to 03. Also excluded: insurance (→07), professional services and legal
(→02/22), mining and critical minerals (→09), water (→20). Full routing table in
02-industry-roster.md.
2.2 Ranking methodology
Each sector was scored 0–5 by its own analyst on the brief's 11 criteria, as part of the
dossier rather than as a separate exercise, so the scores are grounded in the research.
Analysts were explicitly instructed that "if everyone returns 5s the ranking is
worthless" — and the spread confirms they complied: capital_invested ranges from 2 to 5
and strategic_importance from 2 to 5.
The criteria are not equally informative for the question being asked, which is not "which industry is most important" but "which industry has the greatest need for continuous trend research." Those are different questions. Food and beverage is a larger share of GDP than crypto and changes far more slowly; the research need is lower. Weights therefore favour knowledge decay and demonstrated demand:
| Criterion | Weight | Rationale |
|---|---|---|
speed_of_change |
13% | How fast knowledge decays — the direct driver of continuous research need |
intelligence_demand |
13% | Demonstrated pull from decision-makers |
paid_research_opportunity |
11% | Commercial viability of covering it |
cross_industry_influence |
10% | Leverage: covering it improves coverage of other sectors |
capital_invested |
9% | Capital at risk creates willingness to pay for accuracy |
economic_importance |
9% | Baseline stakes |
strategic_importance |
8% | National-security and infrastructure criticality |
regulatory_impact |
7% | Rule-making creates dated, trackable events |
company_product_density |
7% | Trackable surface area |
consumer_impact |
7% | Breadth of audience |
data_availability |
6% | Feasibility — weighted low on purpose, because poor data availability is also the opportunity: sectors where good data is hard to get are where research adds most value |
Sensitivity check. Weighted and unweighted rankings were compared. 24 of 25 sectors move two places or fewer, and the maximum movement is three places. The ranking is therefore not an artefact of the weights — a reader who disagrees with our weighting will arrive at substantially the same list.
2.3 The ranking
Scores are 0–100. Δ shows movement versus an unweighted (equal-weight) ranking.
| # | Sector | ID | Score | Spd | Econ | Cap | Dens | Reg | Cons | Strat | Dem | Paid | Data | Xind | Δ |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Semiconductors, advanced hardware & quantum | 03 |
97.4 | 5 | 5 | 5 | 5 | 5 | 4 | 5 | 5 | 5 | 4 | 5 | — |
| 2 | Healthcare, biotech & life sciences | 06 |
94.2 | 4 | 5 | 5 | 5 | 5 | 5 | 5 | 5 | 5 | 4 | 4 | — |
| 3 | Energy & power systems | 05 |
93.8 | 4 | 5 | 5 | 4 | 5 | 5 | 5 | 5 | 4 | 5 | 5 | — |
| 4 | AI foundation models & compute infrastructure | 01 |
93.4 | 5 | 5 | 5 | 4 | 4 | 4 | 5 | 5 | 5 | 3 | 5 | — |
| 5 | Enterprise software & SaaS | 02 |
89.6 | 5 | 5 | 4 | 5 | 3 | 2 | 4 | 5 | 5 | 5 | 5 | +3 |
| 6 | Finance, fintech & payments | 07 |
88.8 | 3 | 5 | 4 | 5 | 5 | 4 | 4 | 5 | 5 | 4 | 5 | -1 |
| 7 | Industrial manufacturing, supply chain & logistics | 09 |
88.0 | 4 | 5 | 4 | 5 | 5 | 3 | 5 | 4 | 4 | 5 | 5 | -1 |
| 8 | Automotive, mobility & transportation | 10 |
87.6 | 4 | 5 | 4 | 5 | 5 | 5 | 5 | 4 | 4 | 4 | 4 | -1 |
| 9 | Startups, venture capital & private markets | 11 |
85.8 | 4 | 4 | 5 | 5 | 4 | 3 | 3 | 5 | 5 | 3 | 5 | — |
| 10 | Cybersecurity & digital trust | 04 |
82.6 | 4 | 3 | 4 | 5 | 5 | 3 | 5 | 5 | 4 | 3 | 4 | — |
| 11 | Aerospace, defense & space | 08 |
81.8 | 4 | 4 | 4 | 5 | 5 | 3 | 5 | 4 | 4 | 3 | 4 | — |
| 12 | Robotics & physical automation | 13 |
80.2 | 4 | 4 | 4 | 5 | 3 | 2 | 5 | 5 | 4 | 3 | 4 | +2 |
| 13 | Retail & e-commerce | 12 |
79.4 | 3 | 5 | 2 | 5 | 5 | 5 | 3 | 4 | 4 | 5 | 4 | -1 |
| 14 | Real estate, architecture & construction | 18 |
78.4 | 3 | 5 | 4 | 4 | 4 | 5 | 4 | 4 | 3 | 4 | 4 | -1 |
| 15 | Advertising, marketing & brand | 16 |
78.2 | 4 | 4 | 2 | 5 | 4 | 4 | 2 | 5 | 5 | 3 | 4 | +1 |
| 16 | Crypto & digital assets | 21 |
77.4 | 5 | 2 | 3 | 5 | 5 | 3 | 4 | 4 | 4 | 3 | 4 | +1 |
| 17 | Education, edtech & workforce development | 22 |
76.0 | 3 | 5 | 2 | 4 | 4 | 5 | 4 | 4 | 3 | 5 | 4 | -2 |
| 18 | Travel, tourism & hospitality | 25 |
74.8 | 3 | 4 | 3 | 4 | 5 | 5 | 3 | 4 | 4 | 4 | 3 | — |
| 19 | Social platforms, creator economy & consumer apps | 17 |
73.2 | 4 | 4 | 2 | 5 | 5 | 5 | 2 | 4 | 3 | 4 | 3 | +1 |
| 20 | Food, beverage & agriculture | 19 |
73.2 | 2 | 5 | 2 | 4 | 5 | 5 | 4 | 3 | 4 | 4 | 4 | -1 |
| 21 | Fashion, beauty & personal care | 23 |
72.8 | 4 | 4 | 2 | 5 | 4 | 5 | 2 | 4 | 4 | 3 | 3 | — |
| 22 | Media: film, television & video | 14 |
72.6 | 4 | 3 | 3 | 4 | 4 | 5 | 2 | 4 | 4 | 4 | 3 | — |
| 23 | Climate tech, sustainability & carbon markets | 20 |
72.2 | 4 | 3 | 3 | 4 | 5 | 2 | 3 | 4 | 4 | 3 | 4 | +1 |
| 24 | Gaming, virtual worlds & immersive media | 15 |
71.6 | 4 | 4 | 2 | 5 | 5 | 5 | 2 | 3 | 4 | 3 | 3 | -1 |
| 25 | Publishing, music & rights-based IP | 24 |
70.4 | 4 | 3 | 2 | 4 | 5 | 4 | 2 | 4 | 4 | 2 | 4 | — |
2.4 Why each sector belongs — by tier
Tier 1 — Cover daily, staff deeply (ranks 1–4)
These four score above 93 and share a property: they are physically and financially coupled to each other, which is the defining structural fact of 2026. Semiconductors constrain AI; AI demand constrains power; power constrains everything. A platform that covers one without the others will be wrong about all three.
1. Semiconductors, advanced hardware & quantum (97.4) — The only sector scoring 5 on seven of eleven criteria. Maximum speed of change, capital, density, regulation and strategic importance simultaneously. The research found the scarce asset has migrated away from logic: memory makers now out-earn the leading-edge foundry on margin, and advanced packaging gates accelerator supply. Great-power policy sets its boundary conditions in both directions.
2. Healthcare, biotech & life sciences (94.2) — Scores 5 on economic importance, capital, density, regulation, consumer impact, strategic importance, intelligence demand and paid research. Its economics are now set by three simultaneous government price-setting mechanisms, which makes regulatory monitoring existential rather than advisory. The largest established paid-research market of any sector here.
3. Energy & power systems (93.8) — Ranks this high because it became the binding physical constraint on the technology economy. The FOMC named the AI buildout as an inflation driver; sector 05 found the demand forecast underwriting ~$1.4T of US utility capex was revised 69% upward in one year and has already begun deflating in three jurisdictions. Perfect 5 on data availability — system operators publish extraordinary primary data — which makes it unusually tractable.
4. AI foundation models & compute infrastructure (93.4) — Maximum on speed, capital, strategic importance, intelligence demand, paid research and cross-industry influence, but held to 3 on data availability, deliberately. This is the sector's defining research problem: excellent Tier A on the periphery, near-zero at the core. No frontier lab discloses a gross margin. It ranks 4th rather than 1st because the thing everyone most wants to know about it is the thing that is least knowable from public sources — a genuine finding, and a warning to anyone pricing research on it.
Tier 2 — Cover weekly, staff seriously (ranks 5–11)
5. Enterprise software & SaaS (89.6) — Scores 5 on data availability, the only Tier 1–2 sector to do so besides energy: public filings make it unusually verifiable. Its central question — whether AI expands or cannibalises software revenue — is worth more than most sectors' entire coverage, and roughly $1T of market value has moved on an answer that disclosed customer metrics do not yet support.
6. Finance, fintech & payments (88.8) — Low on speed (3) and ranked on stakes, density and regulation instead. Credit risk that supposedly left the banking system is visibly re-entering it through bank lending to the funds that took it. Its intelligence demand and willingness to pay are both maximal.
7. Industrial manufacturing, supply chain & logistics (88.0) — 5 on data availability, strategic importance and cross-industry influence. Where tariff and chokepoint geopolitics become measurable in customs and price data rather than commentary.
8. Automotive, mobility & transportation (87.6) — 5 on economic importance, density, regulation, consumer impact and strategic importance. Growth has moved offshore from every Western home market. Also the sector with the single most hype-distorted claim in the programme (robotaxis), which makes disciplined coverage unusually valuable.
9. Startups, venture capital & private markets (85.8) — Not an industry; a cross-cutting capital layer, and flagged as such so its metrics are never summed with sector totals. It ranks 9th on intelligence demand (5), paid research (5), capital (5) and cross-industry influence (5) — it is the most monetised trend-research category in existence. It scores only 3 on strategic importance and data availability, and the research established why: no independent audited DPI benchmark exists anywhere.
10. Cybersecurity & digital trust (82.6) — 5 on density, regulation, strategic importance and intelligence demand, but 3 on economic importance and 3 on data availability. Held back by the sector's evidence problem: nearly all widely-cited statistics originate from vendors with a commercial interest in alarm, and public reporting institutions are contracting.
11. Aerospace, defense & space (81.8) — Record backlogs against record throughput shortfalls. Ranked here rather than higher by low consumer impact (3) and poor data availability (3) — appropriation-to-contract conversion is unpublished on both sides of the Atlantic. The SpaceX IPO created genuinely new Tier A disclosure.
Tier 3 — Cover weekly to monthly (ranks 12–18)
12. Robotics & physical automation (80.2) · 5 on density, strategic importance and intelligence demand. Flat 542,000-unit global market with all growth in China, underneath a humanoid narrative that acquired audited disclosure in 2026 and was falsified by it — the cleanest hype-versus-evidence case study in the programme.
13. Retail & e-commerce (79.4) · 5 on economic importance, density, regulation, consumer impact and data availability. Only 2 on capital invested, correctly: venture capital is functionally absent from retail in 2026.
14. Real estate, architecture & construction (78.4) · Where the AI buildout becomes physical. Data-centre real estate is the only structurally tight major property type in the world (1.4% vacancy), while everything else contracts, linked by shared crews rather than capital.
15. Advertising, marketing & brand (78.2) · 5 on density, intelligence demand and paid research. A large, well-funded research market; ranked here by low strategic importance (2) and weak data availability (3) — no public audited CPM index exists for any channel.
16. Crypto & digital assets (77.4) · 5 on speed of change — joint-fastest of all 25 — and 2 on economic importance. That combination is exactly what this tier is for. 2026 is the first year its price and its policy moved in opposite directions, which removes the sector's standard explanation for its own cycles.
17. Education, edtech & workforce development (76.0) · 5 on economic importance, consumer impact and data availability; 2 on capital. Enormous as employment, trivial as capital. It ranks here because the graduate labour market is the cleanest natural experiment on AI and jobs available anywhere.
18. Travel, tourism & hospitality (74.8) · 5 on regulation and consumer impact. Huge revenue, ~2.0% net margins, high consumer relevance, moderate change velocity.
Tier 4 — Cover monthly, high editorial leverage (ranks 19–25)
The compression here is the point: ranks 19–25 span 70.4 to 73.2 — under three points. These sectors are not ordered meaningfully relative to each other and should be treated as one band. Each scores 4–5 on consumer impact or regulation while scoring 2 on capital and strategic importance.
19. Social platforms, creator economy & consumer apps (73.2) — 5 on density, regulation and consumer impact. Regulation, not product, now determines architecture. 20. Food, beverage & agriculture (73.2) — 5 on economic importance, regulation and consumer impact; lowest speed-of-change score of all 25 (2), which is why the largest sector by consumer touch sits here. 21. Fashion, beauty & personal care (72.8) — Sustains a large existing paid-research market (5 on consumer impact, 4 on paid research) on modest strategic weight. 22. Media: film, television & video (72.6) — 5 on consumer impact, 2 on strategic importance. High audience interest, low decision stakes. 23. Climate tech, sustainability & carbon markets (72.2) — 5 on regulatory impact and 2 on consumer impact, the signature of a B2B-regulatory sector. Its ranking reflects 2026 conditions honestly: its regulatory scaffolding was dismantled faster than it was built. 24. Gaming, virtual worlds & immersive media (71.6) — 5 on density, regulation and consumer impact; 2 on capital and strategic importance. 25. Publishing, music & rights-based IP (70.4) — Lowest overall and lowest data availability of all 25 (2). Ranked last on its own economics, but it punches far above its rank on cross-industry influence (4), because the AI training-rights question decided here sets the legal boundary conditions for sector 01.
2.5 What the ranking does not say
The composite is a coverage-priority score, not an importance score. Three cautions:
- Rank 25 is not unimportant. Publishing/music ranks last on sector economics while hosting litigation that determines what is legal for the highest-ranked sectors.
- The Tier 4 band is not ordered. A 2.8-point spread across seven sectors is within the noise of 0–5 analyst scoring. Treat them as tied.
- Rankings decay. Sector 05 moved into Tier 1 on the strength of a demand shock that
was not visible eighteen months ago. Re-score annually, and treat a sector's
speed_of_changeas the indicator of how fast its own ranking will go stale.
- Source artifact
- 01-frameworks/11-industry-ranking.md
- Corpus date
- 15 September 2026
- Prepared for this site
- 16 September 2026
- Site publication
- 18 September 2026
- Verification
- Inherited; not fully rechecked