Core themes
23 active investment themes, each scored 0–100 for how widely known the story is and how fully priced it already is. The gap between the two is the signal. Categories: Emerging 7 · Energy 6 · Frontier intelligence 1 · Geographies 3 · Sectors 6. Scores are the median of named AI-model reviews (Claude Fable 5, GPT-5.6, Grok 4.5, Gemini, Meta); missing scores are never invented. 3 active plays sit above as live desk positions. 39 disqualified themes are retained below so rejected ideas are not recreated. Full records, adversarial reviews, falsifiers and sources are in the interactive ledger and in themes.json.
Sectors · Dec 2026–Jun 2027 · Active research · 2 reviewers · known 82 / priced 50 / gap +32 · sourced by GPT-5.6 · reviewed by Claude Fable 5
Mandatory central clearing starts for eligible cash Treasury trades on December 31, 2026 and repos on June 30, 2027. The first order is safer settlement. The second order is a new price on balance-sheet and collateral velocity: clients need clearing access, margin, custody, default-fund capacity and intraday liquidity for the deepest market in the world.
Verdict: The rule is universal knowledge inside fixed income and barely a stock-market theme outside it. Clearing houses and custodians get a volume and workflow pool, futures/electronic venues can gain hedging flow, and balance-sheet-intensive basis trades face a higher explicit carrying cost. But this is not free revenue: netting can reduce dealer balance-sheet use, competition can crush fees, and exemptions can move the pool.
Sectors · Jan 2027–2030 · Active research · 2 reviewers · known 77 / priced 48 / gap +29 · sourced by GPT-5.6 · reviewed by Claude Fable 5
Federal work requirements and more frequent eligibility checks begin January 1, 2027. CBO/KFF estimate millions lose coverage. The second order is not simply lower Medicaid spending: eligibility churn migrates revenue from managed-care premiums and hospital reimbursement into state IT, eligibility operations, patient collections, bad debt and revenue-cycle automation.
Verdict: The policy is known and hospital exposure is partly priced; the workflow reallocation is not. Medicaid-focused MCOs lose members and scale rather than cleanly winning on lower utilization, hospitals in expansion states absorb more self-pay and charity care, and WAY-style payment software can sell into the mess. The best trade is likely dispersion by Medicaid mix and state implementation—not shorting healthcare wholesale.
Sectors · Jul 2027–2030 · Active research · 2 reviewers · known 52 / priced 23 / gap +29 · sourced by GPT-5.6 · reviewed by Claude Fable 5
The FCC has adopted a plan to auction 160 MHz of upper C-band by July 2027. Carriers must pay for spectrum before they can deploy it, creating a two-step equity path: auction spending pressures telecom balance sheets first; radios, antennas, tower amendments and site work follow only after licenses and clearing schedules are real.
Verdict: The auction is known in telecom policy and barely visible in battered tower valuations. AMT/CCI/SBAC gain a real 2028–30 amendment cycle if VZ/T/TMUS buy aggressively, while ERIC/NOK compete for radio spend. The catch is sequencing: a rich auction can delay capex and worsen carrier leverage before it helps towers, so this is a calendar spread—not a buy-everything telecom theme.
Emerging · 2026–2032 · Meta frontier · 3 reviewers · known 62 / priced 34 / gap +28 · sourced by Meta AI · reviewed by GPT-5.6, Claude Fable 5
Humanoid economics are approaching a task-specific crossover in warehouses, factories, and ground operations where labor is scarce, repetitive, and physically constrained. China’s deployment push and live airport trials move the debate from demos toward utilization, while automotive and automation supply chains provide components and manufacturing capacity.
Verdict: Consensus is 50 known / 27 priced: a +23 gap. The wage-shock framing is earlier than most labor models, but the supplied $2/hour figure is a future running-cost estimate—not today’s fully loaded cost. Public exposure is indirect: NVDA and TSLA are narrative-saturated, TER/HON/ROK sell enabling systems, and SYM is warehouse automation rather than a humanoid pure play. Track paid productive hours, not robots shipped.
Emerging · 2026–2032 · Theme hunt · 2 reviewers · known 79 / priced 51 / gap +28 · sourced by Grok 4.5 · reviewed by GPT-5.6
The source thesis says humanoid actuator density increases demand for high-performance NdFeB magnets while China still dominates finished-magnet manufacturing. The scarce step is qualified non-Chinese magnet production and process know-how, not merely rare-earth ore.
Verdict: Supply sovereignty is fully visible after the Department of Defense's 10-year support package for MP Materials; humanoids are optional upside, not required for the core policy case. The investable edge is qualification and profitable finished-magnet output, but a state-backed floor can protect a project without guaranteeing attractive shareholder returns, especially across pre-profit peers.
Sectors · 2026–2028 · Theme hunt · 2 reviewers · known 76 / priced 49 / gap +27 · sourced by Claude Fable 5 · reviewed by GPT-5.6
China's export suspension holds 50–80% of its fertilizer export volume off the world market through at least August 2026, retail DAP/MAP sits at $800–900/st, and the World Bank warns fertilizer prices could rise >30% in 2026 — yet MOS trades ~15% above its June low and −39% from its high, because tripled sulfur and +15% ammonia costs crushed H1 margins and the market read a cost-cyclical miss. The structural read nobody models: the scarcity is policy-durable, the input spike is refinery-byproduct economics that mean-revert — and the Aug 1 Iran framework accelerates exactly that reversion (Gulf sulfur/ammonia supply normalizing) while touching the China leg not at all.
Verdict: Known 70, priced 40 — the widest gap of the August hunt, with two dated catalysts inside 30 days: MOS's Q2 call (Aug 5) and the China export-window expiry this month. The de-escalation framework is asymmetric help: it lifts the cost mask (sulfur/ammonia) faster than it adds supply (Gulf phosphate flows were never the binding constraint — China's policy is). NTR is deliberately NOT the lead expression here: its nitrogen windfall is unwinding (see the nitrogen-unwind theme) and blends against the phosphate story.
Emerging · 2026–2035 · Meta frontier · 3 reviewers · known 45 / priced 20 / gap +25 · sourced by Meta AI · reviewed by GPT-5.6, Claude Fable 5
NIST standards and national-security transition schedules turn post-quantum cryptography from research into a migration program. The scarce work is not selecting an algorithm; it is finding cryptography across software, hardware, firmware, PKI, TLS, VPNs, code signing, identity, and long-lived data—then making those systems crypto-agile without breaking them.
Verdict: Consensus is 31 known / 12 priced: a +19 gap. The supplied 18/5 score understates awareness inside security and government, but the migration workload is still largely absent from public-company segment reporting. The near-term spend is crypto discovery, PKI, certificates, HSMs, identity, protocol testing, and hybrid deployment—not buying “quantum stocks.” IBM and Cloudflare have visible PQC work; PANW, CRWD, OKTA, and HPE are adjacent, diversified beneficiaries.
Geographies · 2026–2030 · Frontier catalyst · 1 reviewer · known 60 / priced 35 / gap +25 · sourced by GPT-5.6 · reviewed by none yet
France’s September 1, 2026 mandate and parallel European rollouts move tax compliance from periodic filing into the live invoice path. Structured documents, taxpayer identity, validation, delivery status and reporting become always-on infrastructure.
Source verdict — review pending: The transition creates real transaction density, but the easy “buy tax software” story ignores price competition and local platforms. The residual edge is recurring cross-border orchestration and error handling—not one-time format conversion.
Energy · 2026–2032 · Frontier signal · 2 reviewers · known 75 / priced 50 / gap +25 · sourced by Grok 4.5 · reviewed by GPT-5.6
AI data-center load and advanced-reactor construction hit the nuclear fuel cycle before reactor deployment. The source thesis identifies HALEU enrichment, conversion, centrifuge manufacturing, qualified components, and specialized labor as the scarce chain.
Verdict: HALEU scarcity is now policy consensus, not hidden: DOE has awarded multiple suppliers and Centrus has a $900 million contract to commercialize added output. The bottleneck is real, but LEU and CCJ already trade as strategic assets and value depends on reactor schedules, appropriations, execution, and financing; underwrite signed task orders and deliveries, not demand forecasts.
Energy · Q4 2026–2029 · Meta frontier · 3 reviewers · known 80 / priced 56 / gap +24 · sourced by Meta AI · reviewed by GPT-5.6, Claude Fable 5
AI infrastructure has moved from a chip-allocation problem toward a power-delivery problem. PJM capacity scarcity, load-growth backstops, transmission lead times, and customer rate design can force both a consumer rate shock and a faster pivot to onsite generation. The second order is who pays for reliability and who can energize capacity before the grid catches up.
Verdict: Consensus is 78 known / 48 priced: a +30 gap after a hard overlap haircut. This is not a new macro thesis; it is the rate-shock and onsite-power branch of AI electricity scarcity. Grid equipment is already crowded, merchant power already carries scarcity expectations, and regulated utilities face customer-allocation politics. Residual edge sits in signed load contracts, minimum-take terms, tariff treatment, interconnection dates, fuel supply, and who funds redundant capacity.
Energy · 2026–2031 · Active research · known 92 / priced 69 / gap +23 · sourced by Grok 4.5 · reviewed by Claude Fable, Gemini 3.1 Pro, GPT-5.6
AI and hyperscale data-center construction will create a durable U.S. electricity-load shock. The scarce assets are not generic ‘energy exposure’; they are firm megawatts, energized interconnections, turbines, transformers, skilled electrical labor, cooling, and fuel delivery in the regions where projects can actually connect.
Verdict: The demand thesis survives. The easy trade does not. Knowledge saturation is 92/100 and price saturation is 69/100 overall, with huge dispersion by layer. The residual edge is contract-level underwriting—signed minimum-take load, cancellation terms, rate treatment, existing capacity, and time-to-power—not knowing that AI uses electricity.
Emerging · 2026–2030 · Theme hunt · 2 reviewers · known 60 / priced 37 / gap +23 · sourced by Claude Fable 5 · reviewed by GPT-5.6
California's insurer of last resort now holds ~684,000 policies and ~$750B of exposure — up 234% since 2022 — and 5.6% of Q1 2026 owner-occupied mortgage originations used it. The backstop of last resort is inside the mortgage machine. Rates jump 29% October 15. The private absorption valve is E&S: surplus-lines homeowners policies in California passed 300,000 — 'without precedent' — with transactions up 119% year over year, now described as the only game in town for ordinary homeowners.
Verdict: The insurance crisis is known; the housing-finance transmission is not priced. Home equity in exposed zips is impaired before any price print shows it — a direct underwriting input for HELOC books (this desk's FIGR position carries it as an unwritten threat). MBS credit models do not distinguish FAIR/E&S collateral from admitted-market collateral — zip-level insurance status is a free credit signal until someone prices it. Muni credit in exposed counties compounds the same loop. E&S specialists inherit a captive market until the next major fire converts growth story to litigation story.
Energy · 2026–2030 · Theme hunt · 2 reviewers · known 79 / priced 56 / gap +23 · sourced by Claude Fable 5 · reviewed by GPT-5.6
The famous AI-power trade is IPPs, gas and nuclear; the unfashionable derivative is that the same load growth keeps coal plants alive. 2025 had the fewest US coal retirements in 15 years (2.6 GW retired vs 8.0 GW planned), DOE has issued 43+ Section 202(c) emergency orders since May 2025 (latest Jul 14, 2026 directing PJM dispatch), five delayed plants are ~a year past planned closure, and PJM's capacity auction cleared near the cap a third straight time. ARLP's Q2 call put numbers on it: 21.2M tons of fresh commitments, data-center demand keeping plants running 'potentially until 2034.' The market prices these as melting ice cubes on a fixed clock; every deferred retirement extends the annuity.
Verdict: Known 75, priced 48. The reliability leg (delays, 202(c) orders, capacity prices) is strengthening while the equities sit 9–47% below their highs after BTU's idiosyncratic met-mine washout. The honest tension is pre-registered: Henry Hub at ~$2.75 is the falsifier actively flashing — this is a duration-of-cash-flows trade, not a volume-growth trade, and it is cheap precisely because the volume year looks bad. ARLP (pure Illinois-Basin thermal, 9%+ yield, contracted out to 2027) is the cleanest expression; CNR the largest US thermal; BTU only for leverage with Centurion noise priced.
Energy · 2026–2031 · Theme hunt · 2 reviewers · known 78 / priced 55 / gap +23 · sourced by Gemini · reviewed by GPT-5.6
New high-voltage transmission can take a decade or more, so the source thesis treats advanced reconductoring and dynamic line rating as the pre-2030 bridge for extracting more capacity from existing rights-of-way. FERC Order 1920 is the regulatory forcing function.
Verdict: FERC Order 1920 requires planners to consider and explain grid-enhancing alternatives; it does not mandate deployment. Reconductoring and dynamic line ratings can accelerate capacity on suitable corridors, but PWR and HUBB already carry grid and AI premiums while ITRI is not a pure GET play; edge requires funded regional projects and disclosed revenue.
Geographies · 2026–2030 · Active research · 2 reviewers · known 81 / priced 58 / gap +23 · sourced by Gemini · reviewed by GPT-5.6
The source thesis says U.S. restrictions on Chinese biomanufacturing suppliers force pharma companies to begin multi-year tech transfers and re-validation before formal deadlines, reallocating work toward geopolitically secure Western preclinical and CDMO capacity.
Verdict: The BIOSECURE framework enacted in 2025 is real, but it did not automatically name WuXi, so 'WuXi exodus' overstates certainty. Pharma tech transfer is slow and western CDMOs can gain, yet broad biotech recovery and already-rerated proxies muddy the signal; only named-entity designations plus disclosed transfer wins validate incremental earnings.
Sectors · 2026–2028 · Active research · 2 reviewers · known 84 / priced 63 / gap +21 · sourced by Claude Fable 5 · reviewed by GPT-5.6
The capability gap collapsed: sub-10B-parameter models now beat last year's 30B+ flagships, and production agent stacks route most calls to small models, escalating only the hard queries — NVIDIA's own June 2025 paper ('SLMs are the future of agentic AI') is now operating doctrine. Inference crossed half of all AI compute in H1 2026, heading to two-thirds-plus. The investable version is not 'SLMs win'; it is the workload-mix repricing: tokens migrate to cheap capacity-bound serving (LPDDR, SRAM, ASICs), software AI COGS deflate, and value leaks out of anything priced for durable frontier-API economics.
Verdict: The technology story is ~16 months into consensus — Gartner's 3x-by-2027 call is dated April 2025, and CES/MWC 2026 keynotes were saturated with it — and the clean proxies (ARM, AMBA, CBRS) already re-rated in H1 and started breaking on earnings. Three layers are still mispriced: SaaS COGS relief (the market spent 2026 pricing AI as a revenue threat to software and gave zero credit for the cost side deflating — DUOL already printed the proof), SLM-shaped datacenter silicon at value multiples (QCOM's LPDDR-capacity AI200 racks at ~17.5x), and model-maker franchises nobody capitalizes (IBM Granite, Microsoft MAI/Phi). The live counter-trade is Jevons: cheap small inference grows total compute — a point the sell side is only one Jefferies note into writing.
Geographies · 2027–2028 · Frontier catalyst · 1 reviewer · known 45 / priced 25 / gap +20 · sourced by GPT-5.6 · reviewed by none yet
Europe’s revised wastewater directive requires pharmaceutical and cosmetics producers to fund at least 80% of qualifying micropollutant-removal costs by December 31, 2028. The forced pool can accelerate quaternary treatment, analytics and process-control spending.
Source verdict — review pending: An obscure transfer from pharma and cosmetics into water infrastructure is more interesting than another generic water-scarcity basket. VLTO and XYL are credible picks-and-shovels, but the catalyst needs national fee schedules and utility tenders before it earns an aggressive score.
Emerging · 2026–2028 · Theme hunt · 2 reviewers · known 60 / priced 40 / gap +20 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Software's famous 25–30% FCF margins were built on renting infrastructure cheaply. AI is ending that: Zscaler cut its FCF-margin guide from ~27% to ~23% in May citing a datacenter capex ramp (and added a capacity-procurement risk factor), ServiceNow's GAAP gross margin fell 81% → 77.9% on self-hosted AI buildout, Akamai runs capex at 19–24% of revenue for GPU deployment, and multiple SaaS 10-Qs began disclosing inference costs at 4–9% of revenue this year. The market treats each confession as idiosyncratic — ZS fell 27.5% alone — while the un-confessed names (NET, DDOG) sit at 52-week highs into this week's prints. The repricing is systemic and it arrives one earnings call at a time.
Verdict: Known 45, priced 22 — and the pricing is maximally bifurcated: the confessed (ZS −54% from high, NOW −41%) already carry the tax in their multiples while the unconfessed carry none of it at all-time highs. The expression is dispersion, not direction: the confessed-and-derated names are the long leg (the market extrapolated one guide cut into permanence), the at-high infrastructure-cost-deniers are the short/avoid leg into their own future confessions. This week is a live experiment: DDOG (Aug 6 am) and NET (Aug 6 pm) print from record highs.
Emerging · 2026–2030 · Theme hunt · 2 reviewers · known 37 / priced 20 / gap +17 · sourced by Claude Fable 5 · reviewed by GPT-5.6
GPUs are 'the largest unhedged corporate treasury asset in existence' (OneChronos): five companies now spend more on compute annually than the entire global oil and gas industry spends on exploration and production, with no reference price, no forward curve, and no way to hedge. That is ending on a dated calendar: CME and ICE have each announced cash-settled chip-price-index futures targeting launch later this year (Silicon Data supplying CME's indices, Ornn supplying ICE's), ElectronX won its CFTC exchange and clearinghouse licenses for direct-access power derivatives in August 2025, and crypto/prediction venues (Hyperliquid, Polymarket, Kalshi) are already listing H100 spot and GPU-availability markets as proto-price-discovery. The end-state instrument — a power-adjusted compute forward locking the 'compute spread,' AI's version of the spark spread — would attach exchange economics to the largest capex stream on earth.
Verdict: Known 25, priced 10 — the widest gap on the current ledger. The AI-infrastructure trade has repriced everything that MAKES compute; nothing in CME or ICE multiples reflects compute as a new listed asset class, and both trade ~18-20% below their highs on rate-cycle volume worries. Distinct from the Treasury-clearing theme despite sharing venues: that thesis is a regulatory mandate taxing collateral velocity; this is a brand-new commodity class arriving on the same rails. The deepest prize may be the index/data layer (Silicon Data, Ornn — private; the Platts/Argus position of the token era), which makes venue equities the accessible expression and M&A the watch item.
Frontier intelligence · 2026–2035 · Initial synthesis · known 85 / priced 69 / gap +16 · sourced by GPT-5.6 · reviewed by Claude Fable
Another jump in model intelligence matters most where outcomes are still limited by scientific search, experiment design, simulation, verification, and control of physical systems. As hypotheses become cheap, scarce laboratory capacity, trusted measurement, proprietary data, and regulated validation loops should capture disproportionate value.
Verdict: The bottleneck framing survives; the stock map does not. The strongest public-market economics still sit one layer down in laboratory automation, instruments, simulation, metrology, test, and provenance—but the accessible expressions of that layer re-rated hard between April and July 2026, before this theme was written. Consensus is now the median of two independent reviews at 85/100 knowledge and 69/100 price. The unresolved question is not whether physical validation becomes scarce; it is whether any public vendor is large enough to capture it.
Energy · 2026–2030 · Frontier catalyst · 2 reviewers · known 72 / priced 57 / gap +15 · sourced by Claude Fable 5 · reviewed by GPT-5.6
In 1962 the Apollo program paid $43.50 a chip and bought Fairchild down the learning curve until chips cost $1 — guaranteed demand did what no subsidy could. AI datacenters are running the same play on frontier power: Bloom Energy's backlog has gone $3.2B (2019) to ~$20B (2025, ~36% CAGR) on multi-gigawatt deals with Oracle, Equinix/AEP sites, and a $5B Brookfield partnership — because fuel cells deploy in months while grid interconnection takes years, and 'time-to-power' now carries a structural premium hyperscalers gladly pay. The same solid-oxide platform produces electrolyzers, so the demand pull is simultaneously funding the green-hydrogen learning curve. What two decades of climate policy couldn't finance, eighteen months of hyperscaler capex is financing at commercial rates.
Verdict: Known 60, priced 40. The Bloom story got famous in 2025 and then washed out — BE now sits 35% below its high, which is what makes it currently ownable rather than chased. The honest structural weakness is ticker breadth, pre-registered here: BE is the only pure expression (GNRC is the backup-power adjacency at −26%, CAT the diversified genset reference), while PLUG and FCEL are explicitly flagged as broken vehicles this theme refuses to include — the precision-fermentation precedent. If a second clean expression never emerges, this theme should expect the graveyard's thin-ticker rule to come for it; until then, the demand-pull evidence is the strongest in frontier energy.
Emerging · 2027–2033 · Theme hunt · 2 reviewers · known 55 / priced 40 / gap +15 · sourced by Grok 4.5 · reviewed by GPT-5.6
AI and computational design can make biological strains and molecules cheaper to design, while large bioreactors and downstream processing remain slow, capital-intensive physical bottlenecks. The source frames industrial biomanufacturing capacity as strategic infrastructure.
Verdict: U.S. capacity gaps and federally backed demonstration facilities validate industrial biomanufacturing as strategic infrastructure. The public basket is weak: TMO, DHR, and RGEN derive most economics from biopharma, while industrial fermentation remains small and often uneconomic, so the theme can be right while the listed proxies never feel it.
Sectors · 2026–2030 · Active research · 2 reviewers · known 84 / priced 70 / gap +14 · sourced by Grok 4.5 · reviewed by GPT-5.6
The source thesis says the next AI-power bottleneck is physical transformers, switchgear, and skilled electrical labor. Multi-year lead times determine which data-center campuses energize on schedule even when generation and interconnection plans exist.
Verdict: Multi-year transformer lead times are independently documented and utilities, hyperscalers, and investors already know the bottleneck. The easy trade is crowded across ETN, PWR, HUBB, and GEV, so the remaining edge is supplier-specific backlog quality, cancellation protection, and capacity-ramp execution; falling lead times can compress both margins and multiples.
Disqualified themes
Retained so rejected ideas do not get recreated.
Emerging · 2026–2030 · Frontier catalyst · 2 reviewers · known 83 / priced 30 / gap +53 · sourced by Gemini · reviewed by GPT-5.6
Disqualified: No clean eligible public expression: Hologic became private on April 7, 2026, broad insurer-mandated DEXA testing is unproven, and BRBR is a diffuse single-ticker protein proxy. The clinical question survives; the listed-equity map does not. Flagged tickers: HOLX · BRBR
The source thesis shifts the GLP-1 trade from food-volume shorts to the quality of weight loss. If lean-mass loss drives monitoring and preservation protocols, DEXA utilization and high-yield protein demand could rise as asymmetric long expressions.
Verdict: Lean-mass loss during GLP-1 weight reduction is clinically real and widely discussed, but there is no evidence of broad insurer-mandated DEXA renewal testing. Hologic has been private since April 2026 and BRBR is a diffuse consumer-protein proxy, so the stated public-equity expression is broken even though muscle-preservation protocols may grow.
Geographies · H2 2026–2027 · Active research · 3 reviewers · known 82 / priced 37 / gap +45 · sourced by Claude Fable 5 · reviewed by GPT-5.6, Grok 4.5
Disqualified: Event passed: the 8/1/26 Iran de-escalation framework broke the oil leg — Brent left $100 in late July and the war premium is unwinding, which this theme's own status note pre-registered as the falsifier within reach. The policy-triangle and positioning legs survive as context, but the registered forcing mechanism ($100 oil through a 40-year-low yen) is no longer live. Resurrect as a new registration if the deal collapses or oil re-spikes. Flagged tickers: MUFG · SMFG · MFG · FXY · DXJ · EWJ
Japan is running an incoherent policy triangle: a dovish government stacking the BOJ board, a finance ministry that has spent ~$72.5B defending a currency its own policy weakens, and last week's oil spike (Brent above $100 before easing to the high $80s on de-escalation) flowing through a 40-year-low yen into import costs. Consensus expects one polite 25bp hike by December. The pipeline says imported inflation accelerates into autumn while leveraged yen shorts sit at nine-year highs. When the political cost of ¥170 gasoline exceeds the political cost of higher rates, the BOJ moves faster than consensus — into record short positioning. August 2024, but from a wider spread and a bigger stock of positions.
Verdict: The widest known/priced gap of the July geographic hunt. Everyone knows the carry trade exists — August 2024 made it a household word — but the market prices the polite path: 86% of surveyed economists see just 1.25% by December, while June core CPI at 1.6% is still below target, giving the doves cover exactly as petroleum import values run +59% and the pass-through arrives in autumn prints. Status note (Jul 28): Brent's break from $100 to the high $80s on Middle East de-escalation puts the oil falsifier within reach — the near-term fuse is wetter than at publication, while the positioning and policy-triangle legs are unchanged. The asymmetry is the point: the consensus path pays carry holders roughly the 250–275bp spread; the forced path repriced the world in three days in 2024. Own the paid leg where it is real (megabanks — at 52-week highs), price the tail where it is cheap (the yen itself, −11% from its high), and treat USDJPY 5-day appreciation over 1% as the tripwire it already is in this desk's risk rubric.
Geographies · 2027–2031 · Theme hunt · 3 reviewers · known 57 / priced 17 / gap +40 · sourced by Claude Fable 5 · reviewed by GPT-5.6, Grok 4.5
Disqualified: No liquid tradeable expression — the verdict itself concedes there is no public Gulf proxy, and the only listed ticker (CEG) is the US name the theme argues to haircut. Retained as an adversarial input on the power themes' terminal values. Flagged tickers: CEG
At $0.05–0.06/kWh with sovereign wealth funding 5GW-scale campuses, marginal AI training migrates to where power is cheap and permitted. $100 oil widens the arbitrage — it funds the Gulf and raises US gas-fired cost simultaneously. The second-order effect this ledger must price: Gulf capacity is a relief valve that caps the duration of the US electricity-scarcity theme scored above at 92/69.
Verdict: Known as spectacle, unpriced as consequence. The announcements are everywhere; the implication — that US power-scarcity trades quietly assume a terminal value the Gulf is busy arbitraging away — appears in no one's model. There is no liquid Gulf proxy, which is itself information: the tradeable expression is an adversarial haircut on US scarcity names, not a purchase.
Emerging · 2027–2035+ · Frontier signal · 3 reviewers · known 60 / priced 20 / gap +40 · sourced by Grok 4.5 · reviewed by GPT-5.6, Claude Fable 5
Disqualified: Same failure as orbital deflation: the actors are private and the listed basket is already-known, beaten-down space names that are adjacent rather than pure orbital-compute plays. Retained as the second duration-cap input on the power themes. Flagged tickers: SpaceX (private) · Starcloud (private) · RKLB · RDW · LUNR · PL
Orbital compute attacks terrestrial constraints simultaneously: solar availability, land, local permitting, and operational water. Filings and pathfinder programs make the direction real enough to monitor, even though gigawatt constellations remain economically and physically extreme. If launch and on-orbit compute curves keep falling, orbital capacity can cap the distant terminal value assigned to terrestrial AI infrastructure.
Verdict: Consensus is 52 known / 14 priced: a +38 gap, the widest of the new frontier set—and also the least immediately investable. The important correction is that vacuum cooling still needs radiator area and mass. The thesis belongs in terminal-value stress tests for terrestrial power and cooling, not in 2027 capacity forecasts. RKLB and RDW sell enabling infrastructure; LUNR, PL, and IRDM are adjacent, not orbital-compute pure plays.
Emerging · 2026–2032 · Frontier signal · 3 reviewers · known 58 / priced 22 / gap +36 · sourced by Grok 4.5 · reviewed by GPT-5.6, Claude Fable 5
Disqualified: The mapped incumbents are already priced — the theme's own layer table scores measurement and analytical instruments 78 known / 65 priced, and every name on the map trades 20–76% above its 2025 low rather than undiscovered. The only near-pure-play, Ginkgo, is a junk vehicle under the quantum-sensing precedent (contract conversion, gross margin, cash runway) and was already flagged as too suspicious in the disqualified precision-fermentation theme. Bruker's 8/4/26 Q2 — a 1–2% FY organic guide and −21.8% on the day — fired this theme's own third falsifier: verification demand absorbed by existing instrument capacity without incremental orders. The verification stack survives as an industry map, not a trade. Flagged tickers: TMO · DHR · A · BRKR · MTD · WAT · DNA
As self-driving laboratories make hypothesis generation and experiment scheduling abundant, the scarce layer moves downstream: calibration, metrology, reproducibility, phase and novelty validation, provenance, and human review. Public funding can build autonomous capacity faster than science can build trust in autonomous claims.
Verdict: Consensus is 58 known / 28 priced: a +30 gap. The headline—AI accelerates science—is saturated; the verification stack is not. The cleanest public exposure remains boring instrumentation and metrology rather than agent software. TMO, DHR, A, BRKR, MTD, and WAT can benefit, but autonomous labs must become material to installed-base growth before this is more than a good industry map.
Sectors · 2026–2030 · Theme hunt · 3 reviewers · known 75 / priced 42 / gap +33 · sourced by Claude Fable 5 · reviewed by GPT-5.6, Grok 4.5
Disqualified: The claimed winners — energetic-materials and specialty-chemical suppliers — have no clean US-listed expression; the listed primes are the names the verdict says not to buy at premium multiples. A thesis whose beneficiaries cannot be bought. Flagged tickers: LMT · LHX · NOC
The government has formally confirmed the bottleneck: a June 11 Defense Production Act invocation for munitions supply chains, citing solid rocket motors, igniters and guidance. Capacity is ramping — toward ~6x for large motors, 3x tactical, $700M+ invested by the primes, Anduril entering as the third US SRM supplier targeting 6,000 motors a year — and GM is exploring missile-parts production with Lockheed: auto capacity converting to arsenal. The real chokepoint sits below the motors: energetics chemistry — propellants and precursors with near-single-source supply.
Verdict: Defense spending is consensus; the layer question is not. Primes at premium multiples carry backlog that new entrants (Anduril's SRM line) are starting to contest — backlog is less of a moat than priced. The unpriced winners are energetic-materials and specialty-chemical suppliers, plus the strange second order of idle auto capacity finding a defense bid exactly as Chinese EV pressure hollows it.
Emerging · 2026–2030 · Theme hunt · 2 reviewers · known 65 / priced 32 / gap +33 · sourced by Grok 4.5 · reviewed by GPT-5.6
Disqualified: Owner call (8/6/26): one ticker, and not a clean one. COIN is the only named expression, and it is far too multi-faceted to isolate machine-to-machine settlement — exchange-cycle beta, custody, stablecoin economics, and regulatory risk dominate any agentic-payments contribution. Same two precedents that killed prior-auth APIs (one-ticker theme) and GLP-1 life-vs-annuity (diversified carriers can't target the exposure). The mechanism stays interesting; it returns only if a purer public expression lists. Flagged tickers: COIN
The source thesis separates agentic payments from ordinary RWA credit rails: autonomous software agents need high-frequency settlement, identity, compliance, and machine-readable payment protocols, with early activity concentrated on stablecoin-native infrastructure.
Verdict: x402 makes machine-readable USDC settlement real, but protocol activity is not yet material COIN economics. Open standards, low fees, bot spam, and identity or compliance constraints can push value to users or stablecoin issuers rather than the exchange; treat this as optionality inside COIN, not a standalone earnings thesis.
Geographies · H2 2026–2028 · Active research · 2 reviewers · known 63 / priced 31 / gap +32 · sourced by GPT-5.6 · reviewed by Claude Fable 5
Disqualified: Owner call (8/4/26): not differentiated enough — a diffuse basket of EU-compliance beneficiaries with no sharp second order; the sharpest sub-mechanism (e-invoicing as a transaction rail) already stands as its own entry. Flagged tickers: SAP · TRMB · DASTY · AMCR · BALL
Two hard dates turn product data into a border credential: the EU Packaging and Packaging Waste Regulation applies from August 12, 2026, and deforestation due-diligence obligations follow at year-end. The immediate spend is not recycled resin; it is proof—supplier identity, geolocation, composition, chain of custody, conformity files and audit trails that travel with every SKU.
Verdict: The regulations are known to compliance teams, not yet modeled as a durable software and verification revenue pool. The second order is a digital tollbooth on EU market access: SAP/DASTY-class systems of record, geospatial and logistics data, and testing firms gain recurring workflow while exporters and packaging users absorb a permanent SKU-level data tax. Packaging-material winners are less clean because substitution mandates phase in and customers can redesign around price.
Sectors · 2026–2031 · Theme hunt · 2 reviewers · known 92 / priced 60 / gap +32 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Disqualified under the desk mandate: the clean public-market expression is primarily short XLP, PEP, and GIS—not an asymmetric long opportunity. Flagged tickers: XLP · PEP · GIS
Lilly's Foundayo (orforglipron) — approved April 1, launched April 6 — is the S-curve steepener: a daily pill, no cold chain, no food restrictions, $149/month self-pay and $50 Medicare from July 1. Demand-side damage is already measured: 20% of US households include a GLP-1 user, snack and confectionery categories are down 12.4%, estimated CPG impact runs $32B growing to $44B in 2026, and three dozen non-healthcare companies now discuss GLP-1 on earnings calls. General Mills is raising capex 23% to reformulate — spending more to defend flat volume.
Verdict: The megatrend is the most known thing on this ledger; the act-two specifics are not priced. The oral inflection at $149 removes every scale constraint injections had; Medicare's $50 price opens the senior wave. Staples face a margin trap — reformulation capex arms race against structurally flat volumes — while XLP trades flat, the market not yet choosing. The adversarial note this desk must log: mass adoption slowing Type 2 progression is a slow question over Libre's TAM — an input for the ABT hypothesis's threat list.
Sectors · 2026–2029 · Active research · 2 reviewers · known 75 / priced 43 / gap +32 · sourced by Gemini · reviewed by GPT-5.6
Disqualified: Owner call after inspecting both tickers (8/6/26): FIX is the right business at the wrong price — $1,737, up ~2.65x off its 52-week low and within 16% of its high, so the mechanical-contractor retrofit cycle is already paid for. HDSN is the wrong economics — refrigerant price deflation has been the dominant force, not the phasedown: operating income fell 86% from the FY2022 peak ($132M to $19M) even as the regulatory calendar advanced, and cheaper cooling-cost dynamics keep working against the reclaimer's spread. A two-ticker theme where one is expensive and the other's P&L contradicts the thesis is not investable. The 2029 step-down mechanism stays real; revisit only on a FIX drawdown or hard evidence that reclaimed-refrigerant pricing has inflected. Flagged tickers: FIX · HDSN
The source thesis says the AIM Act HFC phasedown and the shift to mildly flammable A2L refrigerants create a regulatory capex cycle because legacy commercial systems cannot accept A2L gases without equipment, sensing, ventilation, and mechanical changes.
Verdict: The 2029 HFC step-down is statutory, but 'mandatory retrofit cycle' is too strong: technology-transition rules constrain new equipment while existing systems can often be serviced with reclaimed refrigerant. HDSN is the purer scarcity and reclamation exposure; FIX is a broad contractor already carrying data-center expectations, so only measurable retrofit and reclamation bookings turn this into an equity edge.
Sectors · 2026–2028 · Frontier catalyst · 1 reviewer · known 80 / priced 50 / gap +30 · sourced by GPT-5.6 · reviewed by none yet
Disqualified: Owner call (8/6/26): the AI-rights regime is moving too fast to underwrite, and the best expression is already expensive. NYT is the cleanest pure play on licensable content but trades at a premium multiple (-24.6% off its high at $65.48 is a valuation, not a discount). Two pieces of supporting evidence: (1) the legal foundation is openly unsettled — Thomson Reuters' own FY2025 filing warns that courts 'grappling with the application of copyright law to generative AI... could result in an inability to rely on traditional copyright to adequately protect our content'; (2) the theme's own tickers have diverged violently with no market consensus on who wins — TRI -45.9% off its high versus MCO -11.1%, RELX -24.3%, SPGI -24.5%. When the constituents disagree that much and the case law is being written in real time, there is no stable read to trade. Revisit when copyright precedent settles or NYT de-rates. Flagged tickers: RELX · TRI · SPGI · MCO · NYT
Copyright risk, enterprise indemnification and provenance requirements turn authoritative, licensed corpora into a scarce AI input. The economic prize is not generic content volume; it is data whose ownership, update cadence and permitted uses can survive procurement and litigation.
Source verdict — review pending: The headline is crowded but the revenue mechanism is not. RELX- and TRI-class owners can become rights-cleared toll roads if contracts produce recurring, usage-linked economics; treating every publisher as an AI winner is lazy.
Emerging · 2026–2032 · Meta frontier · 3 reviewers · known 40 / priced 12 / gap +28 · sourced by Meta AI · reviewed by GPT-5.6, Claude Fable 5
Disqualified: The reprogramming economics are owned by private companies; the public basket is tools adjacency, not ownership — the precision-fermentation pattern. A science watch-item, not a theme. Flagged tickers: Altos (private) · NewLimit (private) · TMO · DHR · ILMN · TWST · CRSP
ER-100 moves partial epigenetic reprogramming from animal data into a controlled human trial for optic neuropathies. If localized, inducible OSK expression proves tolerable and produces credible functional signals, it can validate a new therapeutic modality rather than merely another longevity biomarker story.
Verdict: Consensus is 33 known / 13 priced: a +20 gap. The science milestone is real; the “GLP-1 moment for aging” analogy is premature. ER-100 Phase 1 tests safety and tolerability in the eye, not systemic rejuvenation. Public exposure is mostly tools—TMO/DHR manufacturing, ILMN/TWST measurement, and CRSP/NTLA as gene-therapy comparables—not ownership of partial reprogramming economics.
Emerging · Apr 2027–2031 · Active research · 2 reviewers · known 68 / priced 41 / gap +27 · sourced by GPT-5.6 · reviewed by Claude Fable 5
Disqualified: Merged, not rejected (8/5/26): consolidated into the qualified water-geography record — both theses express through XYL, which the desk holds and is building slowly. The testing-first sequencing, tool-vendor map, and EPA cadence are carried there as the regulatory annex. Flagged tickers: XYL · TMO · DHR · ECL · DD
Public water systems must complete initial PFAS monitoring by 2027 and begin reporting results; treatment deadlines may move from 2029 toward 2031. That separates the value chain into two waves: analytical instruments, labs and measurement sell into a fixed near-term clock, while filtration and construction depend on contamination maps, municipal funding and the final compliance date.
Verdict: PFAS liability is known; the sequencing is mispriced. Investors jump to giant remediation totals, but the first investable revenue is measurement and recurring compliance data. XYL is the closest integrated public proxy; TMO/DHR sell broad analytical tools; DD has treatment technology but also legacy exposure; MMM is liability, not a picks-and-shovels winner. The extension risk shifts value toward testing rather than killing the theme.
Geographies · 2026–2027 · Theme hunt · 2 reviewers · known 75 / priced 50 / gap +25 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Event passed: the sorting premise was persistent $100 oil; the 8/1/26 de-escalation framework removes it, and the India leg (merged here from the maximum-pain theme) partially resolved in the relief direction. Revisit only on an oil re-spike. Flagged tickers: EWZ · INDA · THD · VNM
Oil above $100 re-rates countries by balance sheet, not by story. The first-order trade (energy equities) is saturated — retail 'should you buy oil stocks' coverage is the tell. The second order moves slower: importers with subsidy budgets (India, Thailand, Vietnam) absorb fiscal and currency damage monthly, exporters with clean grids (Brazil) compound the windfall, and the freight echo — Bab el-Mandeb transits −34%, longer routes — re-tightens shipping the way 2021 did.
Verdict: Known 70, priced 40. The oil price is front-page; the country-level sorting is not. EWZ at +0.22 desk z has barely begun moving for a net exporter with the strongest EM currency; INDA at −10% YTD has priced maybe half its squeeze; the Suez/freight echo is priced nowhere.
Sectors · 2026–2030 · Theme hunt · 3 reviewers · known 67 / priced 42 / gap +25 · sourced by Claude Fable 5 · reviewed by GPT-5.6, Grok 4.5
Disqualified: Two beneficial tickers, both near all-time highs — the deferred borderline fails both new lenses. The observation that V/MA multiples embed no agent-commerce optionality survives as a note, but there is no entry that respects the at-highs precedent. Flagged tickers: V · MA · ZD
The click economy is measurably dying — Ahrefs finds AI Overviews cut top-result CTR 58% (nearly double a year earlier), a randomized experiment shows a causal −39.8%, and publishers forecast −43% search traffic in three years. What replaces it is being built by the payment networks: Visa's Intelligent Commerce has run hundreds of live agent-initiated transactions with 100+ partners and an OpenAI integration; Mastercard's Agent Pay went live in Hong Kong in March. The consensus says crypto disrupts cards; the 2026 evidence says the networks are becoming the authentication and authorization layer of agent commerce — a new toll on a new road.
Verdict: The destruction side is priced — publisher and adtech equities already trade like it. The construction side is not: V and MA are valued as mature payment processors, not as the identity-and-authorization toll of agentic commerce. The desk's own XLC at deeply negative z confirms the loser leg is in the price; nothing in V/MA multiples reflects the winner leg.
Sectors · 2026–2028 · Frontier catalyst · 1 reviewer · known 65 / priced 40 / gap +25 · sourced by GPT-5.6 · reviewed by none yet
Disqualified: A one-ticker theme: WAY is the only real expression, and ORCL buries Cerner inside a conglomerate. Flagged tickers: WAY · ORCL
CMS-0057-F makes prior authorization and payer-provider data exchange machine-readable on a primarily January 1, 2027 API deadline. Bureaucracy does not disappear; requests, denials, attachments, status checks and resubmissions become software transactions.
Source verdict — review pending: This is a compliance deadline with real transaction density, not another vague healthcare-AI pitch. WAY is the clearest public workflow map, but direct payer-EHR integration can commoditize the rail before investors see clean incremental margins.
Emerging · 2026–2032 · Theme hunt · 2 reviewers · known 55 / priced 30 / gap +25 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Disqualified under the desk mandate: the clean public-market expression is primarily short SCI/CSV death-care exposure, not an asymmetric long opportunity. The mortality mechanism may be valid, but the available long offsets are too diversified or indirect. Flagged tickers: SCI · CSV
The CDC just reported 2025's US death rate at a record low — 689.2 per 100k, −4.6% year over year — and reinsurers (RGA, Munich Re, Swiss Re) project GLP-1-class drugs could cut all-cause mortality up to 6–9% over coming decades. Death care is pure inverse-mortality exposure: SCI's Q2 core funeral volumes fell 1.7%, management's own explanation was 'much softer mortality' — and the stock rallied to within 6% of its all-time high at 22x because pricing and preneed sales offset it. A structural volume headwind, essentially unpriced, with the short entry offered near the top. This resurrects the disqualified life-vs-annuity theme's mechanism through the side door that fixes its disqualification: SCI and CSV are pure mortality exposure, not diversified carriers.
Verdict: Known 45, priced 15 in the short direction. The record-low-mortality headline is public; its translation into a decade of funeral-volume decay is in nobody's SCI model — the street writes up pricing power instead. Honest warning pre-registered: this is a slow-burn short against a quality compounder with momentum at its back; the thesis needs 4–8 quarters of volume misses, and CDC attributes 2025's drop mainly to overdose declines, not GLP-1 — the drug-driven leg is still building. Size accordingly; CSV is already crushed (wrong entry side) and MATW is a conglomerate mid-breakup.
Sectors · 2026–2029 · Meta frontier · 2 reviewers · known 66 / priced 42 / gap +24 · sourced by Gemini · reviewed by GPT-5.6
Disqualified: Semis family DQ — crowded vehicles, no clean entry: CAMT already trades like an advanced-packaging winner while its glass-specific revenue is undisclosed, and LPKF is an execution-heavy small cap. Advanced packaging is a well-covered semis narrative now; adoption timing is qualification- and yield-dependent, so the isolatable edge doesn't justify building a position. Flagged tickers: CAMT · LPKF
The source thesis says large multi-chiplet AI packages are pushing organic substrates into warping, thermal, and interconnect-density limits. Glass substrates require new through-glass-via processing and additional metrology, creating a materials and equipment bottleneck.
Verdict: Glass substrates have credible technical advantages and production tooling exists, but broad volume adoption remains qualification- and yield-dependent. CAMT already trades like an advanced-packaging winner while its glass-specific revenue is undisclosed, and LPKF remains execution-heavy; this is a monitorable inflection, not yet a validated earnings bridge.
Geographies · H2 2026 · Theme hunt · 2 reviewers · known 70 / priced 47 / gap +23 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Duplicate: its only expression (INDA) is already the India leg of the EM twin-deficit sorting theme, with the same catalysts and falsifiers. Carried there. Flagged tickers: INDA
50% US tariffs over Russian crude, >80% oil import dependence at $100, and a pressured rupee form the maximum-pain configuration — and maximum pain is what produces resolution. Both sides need the deal: the US wants a China+1 anchor, India needs export access and oil diplomacy. INDA prices the squeeze, not the resolution. The other branch — doubling down on discounted Russian barrels into secondary-sanctions escalation — is live too, which makes this event-driven rather than slow-burn.
Verdict: Known 60, priced 30. The tariff fight is well covered; the deal path is not in the price. INDA at −10.4% YTD and −0.37 desk z is a market paying for the squeeze and assigning roughly nothing to softening US posture that is already on the record.
Emerging · 2026–2029 · Theme hunt · 2 reviewers · known 50 / priced 27 / gap +23 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: The verification infrastructure is almost entirely private and arrives via M&A premiums; RBLX sits on both sides of the thesis and TRU is a diversified bureau. No isolatable public expression. Flagged tickers: RBLX · TRU
Australia's under-16 ban has produced six months of real enforcement data: 4.7M accounts removed or restricted, under-16 account ownership down from 49.7% to 31.3%, five platforms under investigation with penalty decisions due mid-2026 against AUD $49.5M fines. The infrastructure layer is compounding — Yoti's revenue grew 62%, facial age estimation is regulator-approved in four major markets, NIST has begun global benchmarking, and the White House made age assurance a baseline requirement for AI. The UK's OSA and US state laws make Australia the template, not the outlier.
Verdict: Almost nothing here is priced. Age assurance and agent verification are converging into one identity layer — the rails proving 'this user is 16' also prove 'this agent may transact' — doubling the demand on infrastructure that is almost entirely private (the absence of a public pure-play is the signal: this arrives via M&A premiums). The engagement paradox is unpriced too: platforms purging minors will report better ad economics, flipping the narrative from threat to cleanup on the first 'users down, revenue fine' print. RBLX sits on both sides — defendant in a live age-verification securities suit, potential structural winner if compliance becomes its moat.
Sectors · 2026–2031 · Meta frontier · 3 reviewers · known 78 / priced 55 / gap +23 · sourced by Meta AI · reviewed by GPT-5.6, Claude Fable 5
Disqualified: Overcrowded and fully known: sovereign-AI / export-control semis is saturated coverage, and every vehicle (semicap, memory, accelerators) is the same mega-cap complex the whole market already owns for adjacent reasons. The theme's own verdict flagged that domestic substitution 'dominates semiconductor research' and that sovereignty can be violently priced locally. No differentiated entry for us — hard to want to build a position in names at these levels. Flagged tickers: NVDA · AMD · MU · AMAT · LRCX · ASML
Export controls and national-security policy are creating parallel AI stacks rather than clean decoupling. Nations will selectively localize memory, accelerators, cloud control, data residency, and critical tools while remaining interdependent. The economic result is duplicated capex, guaranteed domestic demand, and less fungible regional supply.
Verdict: Consensus is 66 known / 40 priced: a +26 gap. Sovereign AI is real, but “Nvidia monoculture is unchallenged” is already a straw man—export controls and domestic substitution dominate semiconductor research. CXMT’s reported debut surge also means sovereignty can be violently priced locally. The edge is scenario-specific: license policy, China revenue exposure, HBM qualification, duplicate regional capex, and which tools remain irreplaceable.
Geographies · 2026–2028 · Theme hunt · 2 reviewers · known 82 / priced 60 / gap +22 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Too few beneficial tickers: one country ETF, and the actual trade — intra-German dispersion between rearmament winners and energy-drag losers — is not constructible in liquid US-listed names, by the verdict's own admission. Flagged tickers: EWG
Defense spending rises from €82B to €109B by 2027 (≈€130B with Ukraine) under an amended debt brake — a genuine fiscal impulse arriving exactly as energy re-inflation cuts 2026 growth to 0.5%. The index nets to nothing: defense, construction and software absorb stimulus while energy-intensive export industry keeps shrinking and China retaliation aims at autos. EWG flat at +0.03 desk z all year is not the absence of a theme — it is two themes cancelling.
Verdict: Known 80, priced 50. Defense primes are fully found; the offset view — fiscal impulse quietly cancelling the energy drag, making Germany outperform a 0.5% consensus — is half-priced at best, and the dispersion trade underneath is barely constructed anywhere.
Geographies · 2026–2029 · Theme hunt · 2 reviewers · known 52 / priced 30 / gap +22 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Too few beneficial tickers: three blunt country ETFs proxy a plant-level story whose actual winners (Chinese OEMs' European transplants and their local suppliers) are not US-tradeable; transplant capex is a rounding error inside these funds. Flagged tickers: THD · EWG · EWY
The EU's drafted 'overcapacity instrument' — its own Section 301 — meets a China running record $412B monthly exports off 30% of world manufacturing against 13% of world consumption. Tariff walls do not stop that surplus; they relocate and redirect it. Relocation enriches transplant hosts (Hungary, Turkey, Morocco, Mexico, Brazil). Redirection sends the blocked surplus into the Global South — where Thailand's Japanese-owned auto base, the Detroit of Southeast Asia, gets hollowed. Germany is hit twice: competitor and first retaliation target. Korea autos quietly gain EU share.
Verdict: Known 40, priced 20 — the least-known theme of the hunt, and the one where the desk's own tape argues back: THD is the hottest country on the board (+0.48z, +22.9% YTD) while this thesis calls its auto base structurally short. Either tourism and rate cuts dominate for years and the thesis is early to the point of wrong, or the market is paying headline momentum and ignoring a hollowing it has not priced at all. Resolve that before any position: decompose THD.
Sectors · 2026–2028 · Theme hunt · 2 reviewers · known 75 / priced 55 / gap +20 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Overcrowded and fully known: the memory supercycle is the most-told story in the market — Gartner's 130% price projection, OEM hike confirmations, and HBM scarcity are front-page consensus, and MU has already been marked as the obvious vehicle. The theme's own verdict conceded it is 'priced in the memory makers'; the surviving second-order angles (durables CPI, refresh cycles) are too diffuse to isolate a position. Hard to want to build here at this point in the cycle. Flagged tickers: MU · WDC · SNDK
Gartner projects combined DRAM+SSD prices up 130% by end-2026, lifting PC prices 17% and phones 13% while shipments fall 10.4% and 8.4%. Memory goes from 16% to 23% of a PC's bill of materials; the sub-$500 PC segment 'disappears by 2028.' Lenovo, Dell, HP, Acer and ASUS have all confirmed 15–20% hikes and spec downgrades. Two decades of consumer-hardware deflation are ending because AI datacenters are eating the memory supply — and durable-goods CPI turning positive walks straight into a Fed where 17 of 18 officials already lean hawkish.
Verdict: Priced in the memory makers; unpriced as a regime. The second orders the market has not worked through: durables CPI as a new inflation input (a macro loop into the yen and rates themes), OEM margin resets and brand-power sorting, lengthening refresh cycles favoring installed-base software and repair/refurb over units, and edge-AI slowing — which keeps inference in the cloud and feeds the datacenter themes this ledger already tracks. The shortage is self-reinforcing across the stack.
Emerging · 2026–2032 · Theme hunt · 2 reviewers · known 70 / priced 50 / gap +20 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Not enough ticker diversity or thesis specificity. The theme is mainly SpaceX, RKLB, and ASTS; the public names are already-known space trades getting beaten down. Flagged tickers: SpaceX (private) · RKLB · ASTS
Starship finished its 12-flight test program in June and begins payload delivery later in 2026, targeting a flight every 7–10 days. Internal economics land around $200–500/kg to LEO against Falcon 9's ~$2,720/kg customer price — an order of magnitude, even if aggressive targets miss. Cheap mass to orbit reprices everything designed around expensive mass.
Verdict: Starship is famous; the downstream repricings are not started. Satellite design flips from mass-efficiency to mass-abundance — the exquisite rad-hard component chain (legacy space primes' margin pool) loses to COTS volume. GEO operators enter terminal managed decline. The rural telecom moat erodes: Starlink at 10x-cheaper deployment plus direct-to-cell undercuts rural cable/fiber build ROI and squeezes ASTS between SpaceX's scale and its own capital needs — this desk's scanner already carries ASTS on the short side. And SpaceX explicitly lists orbital AI compute as a Starship application: at a few hundred dollars per kilogram, the electricity-scarcity theme acquires its second duration cap, after the Gulf.
Emerging · 2026–2033 · Frontier signal · 3 reviewers · known 75 / priced 55 / gap +20 · sourced by Grok 4.5 · reviewed by GPT-5.6, Claude Fable 5
Disqualified: The recommended public tickers—especially LNZA and DNA—are too suspicious, and the underlying companies are hard to buy. Flagged tickers: LNZA · DNA
Programmed microbes and gas-fermentation systems are moving from specialty proteins toward broader chemicals and industrial molecules. The long-run shift is production that depends more on strain performance, feedstock, fermentation, and purification than on acreage, weather, animals, or conventional petrochemical routes.
Verdict: Consensus is 70 known / 45 priced: a +25 gap. The technology is real; the basket trade is not. LNZA and DNA already show how correct platform narratives can destroy equity when scale-up and financing fail. The surviving edge is product-level: contracted offtake, fermentation yield, downstream purification, plant utilization, and cost parity. ADM, IFF, and DD are potential enablers or incumbents, but the theme is unlikely to move their consolidated earnings soon.
Geographies · 2026–2030 · Frontier catalyst · 1 reviewer · known 50 / priced 30 / gap +20 · sourced by GPT-5.6 · reviewed by none yet
Disqualified: All four proxies are diversified giants with zero isolatable methane-MRV exposure, by the verdict's own admission. Four tickers that all fail isolation is the same disease as two. Flagged tickers: BKR · EMR · HON · TDY
From January 1, 2027, EU importers must report MRV equivalence for relevant oil and gas supply contracts concluded or renewed from August 4, 2024. Cargo quality expands from chemistry and delivery terms into producer-level methane evidence.
Source verdict — review pending: The investable layer is measurement, verification and contract traceability—not a simplistic long clean gas trade. Public proxies are diversified, so disclosed methane-MRV orders matter more than a low awareness score.
Geographies · 2026–2029 · Frontier catalyst · 1 reviewer · known 55 / priced 35 / gap +20 · sourced by GPT-5.6 · reviewed by none yet
Disqualified: A thin OTC ADR plus two conglomerates where cable repair is invisible. Investable only if the pilot scales — a watch-item, not a theme. Flagged tickers: KBGGY · TEL · TDY
Europe’s €40 million cable-hub call is a small but explicit shift from ordinary telecom maintenance toward pre-positioned repair capacity. Vessel access, trained crews, spare cable, fault localization and port readiness can become strategic-reserve services.
Source verdict — review pending: The vulnerability headline is known; repair mobilization economics are not. This becomes investable only if the pilot scales into multi-year vessel, monitoring and hub contracts.
Emerging · 2026–2035 · Theme hunt · 3 reviewers · known 27 / priced 10 / gap +17 · sourced by Claude Fable 5 · reviewed by GPT-5.6, Grok 4.5
Disqualified: Not enough publicly tradable tickers to play the theme: MET and PRU are too diversified as carriers to specifically target this actuarial exposure. Flagged tickers: MET · PRU
The reinsurers have done the math: RGA finds incretins could cut mortality up to 8.8%; Munich Re, across 41 million insured lives, finds lower all-cause mortality among GLP-1 users in diabetic and non-diabetic populations alike; Swiss Re projects up to −6.4% US all-cause mortality by 2045. The industry has stated the asymmetry plainly — lower death claims for life insurers, longer payouts for annuity and pension providers — and almost no equity analyst has translated it.
Verdict: The widest known/priced gap on this ledger. The cleanest unbuilt trade in financials: long mortality-heavy life books against longevity-heavy annuity books — XLF internals no one splits this way. The pension-risk-transfer boom of the past five years was priced on pre-GLP-1 tables, making PRT-concentrated balance sheets the specific losers. And the strangest loop: insurers become GLP-1 payers — subsidizing a $50/month pill to avoid a death claim is arithmetic — adding a durable payer to pharma volume that pharma analysts don't model.
Emerging · 2026–2028 · Theme hunt · 2 reviewers · known 72 / priced 55 / gap +17 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Disqualified under the desk mandate: the clean public-market expression is primarily short DKNG and FLUT—not an asymmetric long opportunity; HOOD and CME upside is too indirect. Flagged tickers: DKNG · FLUT
Combined Kalshi and Polymarket volume hit $44.8B in June — Kalshi alone $31B, up 70% month over month, with sports at 80% of its volume. The substitution is directly measured: Kalshi's daily actives grew 36% in late June while DraftKings and FanDuel lost 36–41% off their peaks in the same window. The engine is structural: CFTC event contracts pay none of the 20–51% state gaming taxes sportsbooks do.
Verdict: DKNG's decline is priced as competition; the tax asymmetry underneath is underappreciated. The gap forces an innovator's dilemma — launch event contracts and dilute the taxed book, or hold and lose share. The state-tax revenue hole guarantees a legislative war, so the theme is a timed trade, not a secular one. Downstream: event volatility becomes a hedgeable asset class (CME entry, brokerage distribution — HOOD's prediction leg is validated by every number here), and the marginal 0DTE dollar and marginal parlay dollar reveal themselves as the same dollar, linking options-volume revenue to the same pool.
Emerging · 2026–2029 · Theme hunt · 2 reviewers · known 68 / priced 51 / gap +17 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Cannot find a well-priced company on the list: SPHR (Sphere), VIK, RCL, and LYV are already near all-time highs. Revisit if those tickers pull back. Flagged tickers: SPHR · VIK · RCL · LYV
Live experiences are the spending category digital abundance cannot produce, and the demand is booked, not forecast: Live Nation has sold 107M tickets (+11%) with a record $6.6B in event-related deferred revenue (+22%) and 85% of large venues booked; Sphere's segment revenue grew 69%; experiences spending is growing 2-3x goods and adding 0.3-0.5pp to GDP. But the boom is K-shaped, and a weakening labor market sharpens the K: new-grad unemployment is 9.7% with entry-level postings at a 37-year low, luxury hotel RevPAR grows 3-7% while economy declines, and Delta's premium cabin out-earned its main cabin for the first time in history. The trade is not 'long experiences' — it is long the luxury tier against the mass tier, inside the same demand chain.
Verdict: The category is known; the internal dispersion is the unpriced part, and the price tape already proves it — the luxury tier trades within 5-7% of 52-week highs (LYV, ABNB, VIK, H) while the secondary and mass tier is wrecked (STUB −62%, SEAT −80%, FUN −45%, TRIP −28%) despite serving the same demand. Three second orders matter: AI-driven entry-level displacement pushes status spending toward IRL scarcity (this theme is the intimacy-recession's IRL layer, promoted); the 2026-28 US mega-event pipeline (World Cup → Super Bowl → LA Olympics) underwrites demand visibility no other consumer category has; and funflation is itself an inflation input — recreation-services pricing power feeds the same hawkish Fed loop as the memory tax. The oil/conflict tail from the Geographies section is the demand risk to travel-linked experiences specifically — Viator's March slowdown on Middle East conflict was the live demonstration.
Sectors · H2 2026–2027 · Theme hunt · 2 reviewers · known 47 / priced 30 / gap +17 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Disqualified under the desk mandate: the clean public-market expression is primarily short CF/MEOH as the nitrogen windfall unwinds, not an asymmetric long opportunity. The commodity-divergence thesis remains useful research, but it does not produce a clean eligible long. Flagged tickers: CF · MEOH
The February war shut Hormuz — ~25% of global urea and 27% of ammonia exports transit it — and urea spiked from ~$400 to $850/t, handing nitrogen producers a windfall (CF nearly doubled off its December low). The commodity has already given most of it back: urea is ~$420, −40% from the peak, as Hormuz management talks progress and China relaxes curbs. The equities have not followed — CF sits just −16% from its March high at ~10x windfall earnings the forward curve says are ending. The Aug 1 framework, with 'total opening of the Hormuz Strait' explicitly on the table, is priced as a crude-oil event; its fertilizer second order — Iranian ammonia and Qatari urea returning to a normalizing market — is in no nitrogen equity model.
Verdict: Known 35, priced 15 — the freshest gap of the hunt and the mirror image of the phosphate theme: same complex, opposite policy direction. Phosphate is policy-TIGHT (China ban); nitrogen is war-premium UNWINDING (Hormuz reopening). NTR is the name caught in the middle — its nitrogen windfall unwinds while its phosphate/potash legs benefit — which is exactly why it leads neither theme. The short expression is CF (purest nitrogen, closest to its high, reports Aug 5 pm at peak-windfall estimates) with MEOH as the methanol analog on the same Iranian-supply-return mechanism.
Geographies · H2 2026–2028 · Frontier catalyst · 1 reviewer · known 85 / priced 70 / gap +15 · sourced by GPT-5.6 · reviewed by none yet
Disqualified: Event passed: the exemption died 2025-08-29 and a year of repricing followed — the verdict itself granted 'little narrative obscurity.' One clean beneficiary (EXPD) is too thin a benefit vector. Flagged tickers: EXPD · FDX · UPS
The EU’s €3 handling charge on low-value imported items turns parcels previously waved through into classified, valued and paid customs records. Direct-from-China economics weaken while brokerage data and local warehousing gain importance.
Source verdict — review pending: This is a real customs-data shock with little narrative obscurity. EXPD may gain brokerage complexity while FDX and UPS lose low-value volume; the clean trade is relative and operational, not “all logistics wins.”
Emerging · 2026–2030 · Theme hunt · 2 reviewers · known 62 / priced 52 / gap +10 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Only two tickers, both already-crushed losers; the second-order winners are private identity rails, and the tradeable promotion — the live-experience IRL layer — was itself disqualified. Flagged tickers: MTCH · BMBL
Dating apps are the first consumer category being structurally shrunk by AI. Bumble's paying users fell 21% in Q1; Tinder has declined eight straight quarters. The category's response is capitulation — Bumble is retiring the swipe itself for an AI matchmaker. And the stigma data (2 in 5 singles would refuse to date an AI-companion user) means companion AI grows as a substitute for the courtship funnel, not a feature inside it.
Verdict: The apps are priced (crushed); the second order is not. Agentic matchmaking is deflationary to its own revenue — an agent that finds three good matches destroys the engagement-monetization model that paid for it. AI-generated profiles make proof-of-human verification dating-critical infrastructure (the same identity rails as the age-gated internet and agent-commerce themes). And the macro tail — dating apps mediated a large share of US couple formation; their decay plus substitution is a decade-scale drag on household formation nobody models in housing demand.
Emerging · 2026–2030 · Frontier catalyst · 1 reviewer · known 45 / priced 35 / gap +10 · sourced by GPT-5.6 · reviewed by none yet
Disqualified: The only pure play is a high-risk vehicle by our own verdict (contract conversion, gross margin, cash runway), and the remaining names are conglomerates where sensing is a rounding error. Flagged tickers: INFQ · LMT · HON
Portable optical clocks and quantum sensors can solve GPS-denied timing and navigation before fault-tolerant quantum computers create broad commercial value. Military procurement supplies the field-validation bridge.
Source verdict — review pending: DARPA’s Tiqker transition is a better commercialization receipt than quantum-computing qubit headlines. INFQ is the cleanest public map and the highest-risk one: contract conversion, gross margin and cash runway matter more than technical demos.
Geographies · 2026–2028 · Frontier catalyst · 1 reviewer · known 70 / priced 60 / gap +10 · sourced by GPT-5.6 · reviewed by none yet
Disqualified: By its own verdict, 'the policy is more under-modeled than the obvious beneficiaries are underpriced' — no underpriced expression means no ticker we benefit from; IBM/NTNX are partial maps at best. Flagged tickers: IBM · NTNX · NET · DDOG
From January 12, 2027, the EU Data Act removes switching charges, including data-egress charges required for a genuine cloud exit. Migration economics improve, but continuous inter-cloud traffic and architectural lock-in remain.
Source verdict — review pending: The rule weakens one lock-in mechanism, not cloud lock-in itself. IBM/Red Hat and NTNX are plausible portability maps; NET and DDOG are expensive abstractions, so the policy is more under-modeled than the obvious beneficiaries are underpriced.
Sectors · 2026–2029 · Theme hunt · 2 reviewers · known 65 / priced 57 / gap +8 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: The expressions are at their highs (XLRE at its high; DLR and EQIX within 3%) — the live-experience precedent applies. The classification-arbitrage observation survives as a note; revisit on a real pullback. Flagged tickers: XLRE · DLR · EQIX
XLRE ranks near the top of the desk's sector board while the 10-year sits above 4.7% — a combination rate-sensitive REITs should not produce. The resolution is composition: datacenter REITs are carrying the index on AI leasing (Digital Realty's strongest quarter ever, 1.2GW pipeline 60% pre-leased; Equinix spending $4.1B across 46 projects) while the office half stages a genuine flight-to-quality recovery (most leases signed in a decade to open 2026). Passive real-estate allocations are now buying AI infrastructure wearing a REIT label.
Verdict: The index-level move is half-priced; the two second-order reads are not. One: sector-classification arbitrage — XLRE's defensive, income character is quietly gone, and allocators holding it for that character own something else now. Two: the internal dispersion — commodity office and levered traditional REITs stay broken at 4.7% while datacenter and prime-gateway names carry the average over them. Index-level signals, including this desk's own z-score, blend two unrelated businesses.
Geographies · 2026–2028 · Theme hunt · 2 reviewers · known 75 / priced 72 / gap +3 · sourced by Claude Fable 5 · reviewed by GPT-5.6
Disqualified: Single-ETF expression and a deliberately flat 60/60 score — a timing call, not a known/priced gap. The direct memory expressions already live in the memory-tax theme. Flagged tickers: EWY
The worst memory shortage in 15 years — spot DRAM +700% y/y, capacity captured by HBM, SK Hynix sold out through 2026, Micron exiting consumer — lands on a country index that is roughly 40% Samsung plus Hynix. EWY is a duopoly royalty stream wearing an ETF: exports, the won, and the fiscal balance all lever to DRAM pricing power. The short leg of the same theme is the memory payers — device assemblers and hardware-margin stories (this desk's NTDOY thesis is that leg).
Verdict: Known 60, priced 60 — the narrowest gap of the hunt, deliberately scored flat: direction is paid, duration is the only live question. Desk z at −0.73 with +57.7% YTD reads as consolidation after capture. If the sold-out-through-2027 claims hold, the pause is the entry; if contract prices roll over, the cycle has already topped. This one is a timing problem, not a thesis problem.
Emerging · 2026–2028 · Frontier catalyst · 1 reviewer · known 55 / priced 70 / gap -15 · sourced by GPT-5.6 · reviewed by none yet
Disqualified: The only negative known/priced gap on the ledger (55 known / 70 priced) — the market has it more than priced — and the cleanest proxy is over-owned near highs. Flagged tickers: ULS
The 2026 UL 9540A and NFPA 855 regime pushes large-scale fire testing, gas characterization, spacing, ventilation and emergency-response design deeper into BESS permitting and insurance. The gating asset becomes an acceptable safety case, not cell supply.
Source verdict — review pending: The bottleneck is under-discussed while the cleanest proxy looks over-owned. Treat this as a diligence theme for testing, engineering and recurring inspection—not permission to pay any multiple for ULS.