Late-2026 backdrop: Fed easing cycle underway with the dollar softer, AI capex still the dominant capital-flow story but visibly hitting physical constraints (power, water, transformers). Geopolitics — Russia LEU ban approaching, China REE/ZPMC decoupling, Ukraine/Israel/Taiwan defense procurement — is driving forced-substitution trades across energy, materials, and defense components. This favors picks-and-shovels behind the AI/defense/decoupling megatrends over the megacap incumbents themselves.
Published September 30, 2026
This batch is unusual: Perplexity validation failed across all 20 themes, so ranking rests on initial thesis quality + Grok's X-narrative reads, nearly all of which returned 'pre_formation / silent.' That is itself a signal — these are genuinely off-radar ideas, not consensus trades, which favors alpha but demands stricter thesis discipline. Cross-theme patterns are strong: (1) AI infrastructure's next-order bottlenecks are shifting from GPUs to power distribution (Theme 11), water (Theme 2), and financialization of compute (Theme 6); (2) a defense-tech cluster (7, 14, 18) all hinge on 2026 DoD Replicator-2 and allied procurement cycles; (3) biotech_genomics is over-represented (Themes 5, 10, 15, plus 19) — 4 of 19 non-rejected themes (~21%), acceptable but flagged. Highest-conviction tier1 picks are ones with named sub-$5B pure-plays and hard 2026 regulatory/procurement catalysts: uranium enrichment (Centrus), solid-state transformers (Ideal Power), prediction market infra, and ASC surgical robotics. Absent Perplexity corroboration, position sizing should skew smaller and monitoring windows tighter.
🔒
Our strongest early-curve themes come in the Pro newsletter — including specific instruments and buy zones.
Premium · coming soon Timing: early
GPU price volatility is real and hyperscaler hedging demand is nascent but growing; however, no listed sub-$5B pure-play exists yet — Together AI/Silogen private, CME hasn't launched a GPU-hour contract. Watch as adjacent thesis until listed vehicle emerges.
- +Spot GPU prices ranged 30-60% intra-quarter in 2025-26
- +CoreWeave / Nebius earnings volatility highlights hedging gap
- +Together AI, Vast.ai, Silogen all scaling compute-brokerage models
- −No listed pure-play — thesis remains venture-only until IPO wave
- −Hyperscalers self-hedge via long-term contracts, killing derivatives demand
- −GPU price volatility normalizes as Blackwell/Rubin supply expands
Timing: early
The bottleneck in loitering muni production is not airframes but MEMS-IMUs, seekers, and fuzes; however, listed sub-$5B pure-plays are thin — MRCY is the closest but not pure. Thesis is real, tradeable expression is mostly via primes and Kratos/AeroVironment.
- +Ukraine consuming 10K+ loitering munis/month; Israel and Taiwan scaling orders
- +DoD Replicator-2 budget line item lands FY2027
- +AVAV Switchblade backlog stretched; component pinch acknowledged
- −Ukraine ceasefire compresses near-term demand
- −Component pinch resolved via multi-sourcing faster than expected
- −MRCY execution issues persist despite thesis
Timing: early
PLTR Maven owns the incumbent narrative; the sub-$2B edge-LLM wedge is real but tradeable pure-plays are scarce (Rebellion Defense private, Applied Intuition private, HawkEye 360 SPAC'd). Play mostly captured via smaller defense-AI adjacents and PLTR itself as a hedge.
- +DoD Replicator-2 FY2027 explicitly budgets edge inference
- +CDAO edge-AI pilot RFPs expanding 2026
- +Applied Intuition defense pivot accelerating with Anduril partnership
- −PLTR captures all edge AI budget via Maven expansion
- −Anduril IPO absorbs public-market attention and capital
- −Replicator-2 budget markup gets cut in FY2027 appropriations
Timing: early
2027 NDAA mandate to remove Chinese ZPMC crane software is a hard forcing function; retrofit control-system market goes to Western vendors. Konecranes (KCR.HE) is the natural incumbent beneficiary, but no US-listed sub-$5B pure-play cleanly captures it — Identiv-class industrial cyber names are proxies.
- +2027 NDAA compliance deadline forcing 2026 procurement decisions
- +CISA guidance on ZPMC software expected H2-2026
- +Port of LA/Long Beach and NY/NJ RFPs anticipated 2026
- −NDAA compliance deadline extended or waivers granted
- −Chinese vendors offer discounted retrofits that ports accept
- −Konecranes captures market but stock already reflects it
Timing: early
Bioleaching is a credible long-dated alternative to conventional REE refining, but commercial pilots are 2027-28 events. Phoenix Tailings and Cyclic Materials are private; MP is the only listed pure-play and it's the incumbent, not a bioleaching play. Real thesis, wrong tradeable vehicles today.
- +China 2025 REE export controls sharpening substitution urgency
- +DOE ARPA-E bioleaching program milestones in 2026
- +Cornell/Idaho National Lab microbial pilot data expected 2026-27
- −Bioleaching remains sub-scale for a decade — thesis right, timeline wrong
- −No listed bioleaching pure-play — plays are conventional REE proxies
- −China lifts export controls, deflating urgency
Timing: early
Retinal biomarkers for Alzheimer's are gaining FDA breakthrough traction and Medicare screening code creation is a plausible 2026-27 catalyst. RetiSpec and Neurovision remain private; Notal Vision and Optomed provide adjacent exposure but no clean sub-$5B US-listed pure-play.
- +GLP-1 for cognition trials increasing screening demand
- +FDA breakthrough designations for retinal AI biomarkers 2025-26
- +Medicare preventive screening code review cycle 2026
- −CMS screening code delayed to 2028+
- −Retinal biomarker sensitivity/specificity data disappoints
- −RetiSpec/Neurovision remain private, blocking pure-play expression
Timing: early
KOSA and EU AI Act child provisions are real regulatory catalysts, but named companies (Bark, Aura, Angel) are private or too small to trade cleanly. Watch for IPO/SPAC activity; thesis premature for allocation.
- +KOSA implementation timelines in 2026
- +EU AI Act child-safety provisions active 2026
- +State-level age verification laws (UT, TX, CA) creating compliance market
- −KOSA struck down or delayed by courts
- −OpenAI/Google absorb the category via native features
- −No tradeable pure-play emerges in the window
Timing: early
Vaccine approach sidesteps consumer feed-additive backlash, but ArkeaBio and Hoofprint Biome are early-stage private companies; NZ/Ireland mandates are 2027+. Real long-term thesis, no near-term tradeable pure-play.
- +NZ agriculture emissions pricing scheme 2027 target date
- +Ireland methane reduction mandates advancing
- +Bill Gates / Breakthrough Energy portfolio validation
- −Vaccine efficacy data underwhelms
- −NZ/Ireland mandates delayed
- −No listed pure-play emerges
Timing: early
Achieve Life Sciences (ACHV) cytisinicline PDUFA is a real binary catalyst; category exists but demand for youth-specific pouch cessation Rx unproven at scale. Interesting speculative play.
- +ACHV cytisinicline PDUFA action expected 2026
- +PMI Zyn US volumes still growing 30%+ YoY, expanding addressable population
- +FDA youth tobacco cessation guidance expected 2026
- −PDUFA rejection or delay
- −Cessation demand insufficient to justify commercial launch
- −ACHV dilution ahead of launch
Timing: early
Single-pilot cargo cert path is real but FAA/EASA timelines are 2028+ and Reliable Robotics/Merlin remain private. ATRO (Astronics) captures avionics adjacency but not pure single-pilot exposure.
- +Regional pilot shortage worsening into 2027
- +Reliable Robotics FAA STC milestones progressing
- +Merlin Labs Air Force cargo autonomy demos
- −FAA single-pilot cert timeline slips to 2029+
- −ALPA political opposition halts progress
- −Reliable / Merlin remain private through window
Timing: early
RBI and PIX auth mandates are real but BioCatch is private; TWLO is the target of the thesis, not the beneficiary. No clean tradeable expression today.
- +RBI 2026 additional factor of authentication rules
- +Brazil PIX security enhancements
- +SMS 2FA fraud losses accelerating in EM
- −BioCatch IPO absent, no pure-play
- −SMS remains sticky despite fraud
- −Regulatory implementation delayed
Timing: early
Live-service shutdowns accelerating in 2026 could drive middleware M&A, but Improbable spin-outs and matchmaking vendors are private. Thematic hypothesis without listed pure-play.
- +Concord/Suicide Squad-style live-service failures continuing
- +Publisher cost-cutting mandating outsourcing
- +Unity/Epic services expansion validating category
- −Publishers keep infra in-house
- −No M&A wave materializes
- −Unity execution issues persist
Timing: early
Water is a real emerging AI infra constraint but Watergen and SOURCE Global remain private/sub-scale. No tradeable public pure-play; captured only via large-cap water incumbents.
- +UAE/Saudi sovereign AI buildouts consuming record water
- +G42/Humain data center water disclosures Q4-2026
- +Gulf sovereign contracts to Western water tech firms accelerating
- −AWH costs remain uneconomic vs. desal
- −No listed AWH pure-play emerges
- −Gulf projects revert to conventional desal
Timing: early
H5N1 zoonotic risk is real but biosurveillance funding remains politically contested; Ginkgo pivot is unproven and Biobot private. Watch item, not action.
- +H5N1 dairy cattle spread continuing 2025-26
- +USDA/WHO surveillance program renewals 2026
- +Ginkgo biosecurity segment ramp
- −H5N1 fades from headlines, funding cut
- −Ginkgo execution track record poor
- −Biobot remains private
Timing: early
Lottery — 60-70% failure probability. 802.11bf ratification is a real standard-based catalyst and Cognitive Systems/Origin Wireless are named credible pre-IPO pure-plays, but adoption timelines are highly uncertain and no listed microcap cleanly captures the thesis today. If WiFi-sensing displaces even 5% of the installed IoT sensor market, adjacent semis benefit disproportionately.
- +802.11bf ratified 2025; chipset support rolling out 2026
- +HPE/Aruba enterprise WiFi refresh cycle 2026-27
- +Cognitive Systems OEM partnerships expanding
- −802.11bf adoption stalls at niche vertical use cases
- −Cognitive Systems / Origin remain private through window
- −Privacy backlash blocks enterprise deployment