II · Risk Matrix
Probability Matrix · All 10 Candidates
Risk, Upside, and Vehicle at a Glance
Claude's estimates. Not predictions — frameworks for sizing positions.
| Ticker |
Company |
5× Prob |
10× Prob |
Zero Risk |
Best Catalyst |
LEAP? |
| RELY |
Remitly Global |
45% |
20% |
Low |
Market re-rate + CEO execution |
Avoid |
| ODD |
ODDITY Tech |
30% |
18% |
Low |
Second brand launch proves platform |
Avoid |
| TEM |
Tempus AI |
25% |
12% |
Medium |
Data biz margin expansion |
Small |
| IOT |
Samsara |
22% |
10% |
Low |
International expansion |
Avoid |
| BRZE |
Braze |
20% |
8% |
Medium |
AI-native messaging wins enterprise |
Yes |
| SOFI |
SoFi Technologies |
18% |
7% |
Medium |
Galileo B2B scale |
Avoid |
| ASTS ★ |
AST SpaceMobile |
18% |
10% |
High |
Constellation deployment + revenue ramp |
Yes — 4/5 |
| RXRX |
Recursion Pharma |
15% |
7% |
High |
Partnership compound in Phase 2+ |
Avoid |
| NBIS |
Nebius Group |
15% |
6% |
High |
EU AI sovereignty demand |
Avoid |
| IONQ |
IonQ |
12% |
8% |
Very High |
Fault-tolerant QC demonstrated commercially |
Tiny only |
* MXL was not in the original 14-stock brief — added from the CRDO peer analysis. Grade C reflects current RSI >90 after 84% single-quarter move; becomes B on a pullback to $47–53.
Low zero-risk = real FCF, low dilution exposure ·
High = binary execution dependency, significant dilution history ·
★ = 5/5 model consensus
III · Ranked Candidates
Claude's Ranked List · May 30, 2026
10 Candidates by Conviction
Valuation-anchored throughout. Color badges indicate primary thesis type.
The most glaring valuation anomaly on this list. A company growing at 29% annually with $283M in free cash flow trading at 1.7× revenue — less than 4% of the global remittance market captured. The immigration-fear discount is real; the revenue growth disproves it every quarter. New CEO from Amazon and Apple.
Why This Could Fail
A meaningful expansion of the current U.S. remittance tax would structurally impair unit economics. If Congress moves from 1% to 10%, the entire remittance corridor re-prices. This is the single regulatory risk that could break the thesis completely.
Real FCF
Claude only — 1/5 models
Common stock
Resembles: MercadoLibre 2016
Why the market is wrong
Pricing in immigration-driven remittance collapse that hasn't shown up in the revenue. The fear is real and persistent enough to hold the stock down indefinitely — which is precisely the opportunity.
NBN · No fit
The remittance story is economic and demographic, not scientific. If the global payments angle interests you, it belongs in an economics channel, not Pop Science.
83% revenue growth in 2025. Proprietary multimodal clinical dataset built over a decade. Q4 EBITDA positive. 2026 guidance: $1.59–1.6B revenue, $65M EBITDA. NRR of 126% — existing pharma customers spend more every year. Perplexity ranked it #1 and chose it as the single-stock pick.
Why This Could Fail
A HIPAA enforcement action or data breach would be catastrophic to the trust the platform is built on. Also: health systems building their own data infrastructure could undercut the moat before monetization fully scales.
AI · Healthcare
3/5 models
Common stock
Resembles: Veeva 2015
Why the market is wrong
High absolute valuation and uncertainty about reimbursement paths keep most generalist investors away. The data moat is invisible until it compounds — and 126% NRR is the compounding signal most are missing.
NBN · Strong fit
AI and precision oncology is exactly the kind of mathematically serious, clinically grounded science the Pop Science channel is built for. An oncologist or data scientist who built the diagnostic layer would be a compelling guest.
Operating system for physical operations — trucking, construction, food distribution, municipalities. Switching costs are extraordinary: once routes, compliance, and insurance programs are built on Samsara, replacement is a multi-year project. Physical operations ≈ 40% of global GDP, almost entirely uninstrumented.
Why This Could Fail
Cisco, Trimble, or a hyperscaler decides physical operations is worth owning and competes on price. The moat is integration depth, but it can be bought or built by deep-pocketed competitors with patience.
Physical-world AI
2/5 models
Common stock
Resembles: Datadog 2021
Why the market is wrong
Persistently misread as an IoT hardware company rather than a software platform with extraordinary switching costs. The physical-world OS thesis hasn't been priced in.
NBN · Moderate fit
Physical operations intelligence is adjacent to complexity science and emergence themes. A conversation about instrumenting the analog world — sensors, logistics, the mathematics of flow — could work.
The only profitable, scaling AI-native consumer company on this list. Builds brands using proprietary skin-analysis algorithms trained on 100M+ consumer interactions. The market prices it as a beauty company. It is a consumer AI platform that happens to sell cosmetics. No other model surfaced it.
Why This Could Fail
Beauty trends are fickle and brand fatigue is real. If IL MAKIAGE loses its cultural moment, the underlying AI infrastructure doesn't save you. An Israeli-based company also carries geopolitical headline risk.
Consumer AI
Profitable
Claude only — 1/5
Resembles: Lululemon 2012
Why the market is wrong
The market sees an Israeli beauty brand with geopolitical headline risk. The AI consumer platform underneath — 100M+ interactions, skin-analysis algorithms, multi-brand architecture — is entirely invisible to most analysts.
NBN · No fit
Consumer AI is not a natural Pop Science subject. The underlying algorithm is interesting, but the guest would need to be a scientist, not a marketer.
First space-based cellular broadband connecting directly to standard smartphones. FCC approval secured. $1.2B+ in contracted carrier commitments. $3.5B cash. Universal 5/5 model consensus — and the only name where even Claude, which ranked it #5, didn't exclude it. The SpaceX IPO filing in late May drove a ~17% single-day spike. Entry price matters here more than any other name on this list.
Why This Could Fail
Space hardware is unforgiving. BlueBird 7 was already lost in a launch anomaly. A second major satellite failure before the constellation reaches operational scale could trigger a capital raise that dilutes equity holders severely. The TAM is real; execution is the entire question.
Space · Infrastructure
Dilution risk
5/5 consensus ★
LEAP: 4/5 models
Resembles: Early Rocket Lab
Why it's controversial
Near-universal consensus that it is overvalued relative to current revenue. The 5/5 model consensus is itself a mild contrarian warning. Entry price matters here more than any other name on this list.
NBN · Strong fit
Direct-to-device satellite connectivity involves orbital mechanics, RF physics, and deployable structures. The science story is legitimate and has not been well told at the popular level.
Customer engagement platform at the intersection of AI and customer data. LLMs are making its product dramatically more powerful. NRR above 115%. Re-rated down from 2021 highs, creating a genuine entry point. Primary risk: Salesforce, HubSpot, and Adobe all want this market.
Why This Could Fail
Adobe or Salesforce acquires a close competitor and bundles it into an existing enterprise relationship. Braze's moat is integration depth, but enterprise software markets can be disrupted by bundling strategies that make switching cost irrelevant.
AI · MarTech
Claude only
LEAP viable
Resembles: HubSpot 2018
Why the market is wrong
Dismissed as a 'nice-to-have' in enterprise budgets — the first cut in a downturn. The AI-native messaging thesis, which makes the product dramatically more powerful, has not been priced in.
NBN · No fit
Marketing automation is not a Pop Science subject regardless of the AI involved.
Most credible pure-play quantum computing company. The lottery ticket on the list. Over $20B in market cap on less than $100M in revenue — you are paying for a future that may not arrive in the five-year window. Included because the asymmetry is real, not because the probability is high.
Why This Could Fail
Quantum computing timelines have slipped repeatedly. Error correction at commercial scale may be a decade away. Cash burn continues and a dilutive capital raise at a lower price than today is a plausible scenario before any commercial milestone is reached.
Quantum · Deep tech
Very high zero risk
2/5 models
Tiny LEAP only
Why it's controversial
$20B+ market cap on $80M in revenue. The market is pricing a technology that may not commercially arrive in this decade. This is a bet on timing as much as on the technology.
NBN · Strong fit
Quantum computing is the highest-CY subject on this list. Trapped-ion architecture, error correction, the relationship between quantum and classical computation — a natural interview. Several strong books on QC have appeared in the last three years.
Using AI and high-throughput biology to industrialize drug discovery. Not trying to develop one drug — trying to rebuild how drugs are found. Proprietary dataset of billions of cellular images. Platform thesis: the output is a pipeline engine the entire pharma industry needs.
Why This Could Fail
Biology is hard and AI doesn't change the fundamental failure rate of drug candidates. If the first wave of partnership molecules fail in Phase 2 trials, the platform thesis collapses faster than the science would suggest. Cash runway is finite.
AI · Drug discovery
2/5 models
Common stock
Resembles: Early Illumina
Why the market is wrong
Three years of missed drug discovery timelines have exhausted investor patience. The platform thesis requires faith that biology can be predicted at scale — and that patience is now at a low.
NBN · Strong fit
AI drug discovery at the biology-computation interface is exactly the interdisciplinary science the channel is designed for. What does it mean to industrialize the search through chemical space?
Only digital financial services company building banking, lending, and investing simultaneously. Banking charter gives a cost-of-funds advantage neo-banks can't match. Consistently misread as a lender when the real asset is a growing deposit base and cross-sell engine.
Why This Could Fail
Banking is slow and regulated. Credit losses in a recession, student loan policy shifts, or a regulatory action on the banking charter could each independently impair the thesis. Being a bank while the market prices you as a tech company is a permanent tension.
Fintech · Banking
2/5 models
Common stock
Resembles: Square 2018
Why the market is wrong
Persistently misclassified as a lender rather than a banking platform. Rate sensitivity fears and student loan policy noise override the membership flywheel story every quarter.
NBN · No fit
Fintech and digital banking do not map naturally to the Pop Science channel.
European AI infrastructure — GPU cloud, AI tooling, data services — spun out of former Yandex. Non-American GPU cloud at a moment when European enterprise AI demand is growing and sovereignty concerns push customers away from AWS/Azure/GCP. Trades at a fraction of CoreWeave's valuation for similar positioning.
Why This Could Fail
GPU cloud is a commodity unless you layer software on top. CoreWeave, AWS, and Azure will compete on price. The Yandex heritage creates reputational friction in some markets. Geopolitical headline risk is permanent and unpredictable.
AI · Infrastructure
3/5 models
Common stock
Resembles: Equinix 2010
Why it's controversial
The Yandex heritage creates a reflexive avoidance reaction in many institutional investors. European AI sovereignty is real but underappreciated as a commercial driver.
NBN · Weak fit
AI infrastructure is a technology story, not a science story. Unless the conversation is about the physics of computation or thermodynamics of data centers, this belongs in a business channel.