Why
A list like this is a cheap sanity check on which "sounds like magic" bets are engineering-tractable soon versus still science fiction, and this year two of the ten intersect with decisions jay has reason to make anyway. Ethereum's default signature scheme, secp256k1 ECDSA, is not quantum-safe — a sufficiently powerful quantum computer could recover a private key from an already-public public key and forge transactions from that address. Account abstraction (ERC-4337) is the one place in Jayverse where the signature scheme is not fixed by the base protocol: a smart-contract account can validate however its code says to, which means Rabbit could in principle swap in a post-quantum verifier per account before it is forced to. World models matter for a separate reason — the game is a 3D environment and Rabbit already talks about physical/agentic extensions, and world models are the research line that lets a system learn "what happens next" from video and motion instead of text, which is the substrate that robotics and autonomous systems need and language models don't provide.
How it works
The ten, in one table
| # | What | Why now |
|---|---|---|
| 1 | Everything-to-grid energy — EVs, rooftop solar, home batteries as bidirectional grid resources (05:03) | New battery chemistries + distributed control absorb AI-datacenter and heatwave peaks (07:53) |
| 2 | Direct lithium extraction — adsorption filters, not multi-year evaporation ponds (09:11) | Hours instead of up to 2 years; modular, lower water use (12:07) |
| 3 | Passive radiative cooling — paints/tiles/films radiate heat to space, no electricity (13:30) | Claimed up to 40% less cooling energy; breaks the "AC heats the city" loop (15:44) |
| 4 | PFAS destruction — electrochemical electrodes break the carbon-fluorine bond (17:47) | Moves past containment to actual remediation and reuse (19:38) |
| 5 | Precision fermentation — reprogrammed microbes synthesize protein, fat, feedstocks in bioreactors (21:54) | Land/water limits ahead of a projected 9B-person 2050 (22:46) |
| 6 | Exosome drug delivery — natural vesicles carry therapeutics to target cells (27:38) | Crosses the blood-brain barrier, avoids immune rejection; Alzheimer's, Parkinson's, pancreatic cancer (29:18) |
| 7 | Personalized mRNA cancer vaccines — biopsy mutations to custom vaccine in weeks (30:57) | Replaces chemo's systemic toxicity with per-tumor targeting (32:16) |
| 8 | Quantum simulation for drug discovery — atomic-level folding/binding instead of approximation (36:20) | Pre-filters candidates against a claimed 90% clinical-failure rate (39:57) |
| 9 | World models — learn physics/causality from video and motion, not text (40:50) | Foundation for physical robotics, autonomous driving, climate prediction (42:21) |
| 10 | Lattice-based cryptography — high-dimensional lattices + noise resist quantum attack (44:34) | Blocks "harvest now, decrypt later" (45:51) |
The episode closes on four megatrends the ten are instances of: hyper-personalization (49:18, 49:25), localization of production (49:38, 49:51), doing more with less (50:24, 50:31), and AI-led science (54:15, 54:42).
World models: the substrate under the game, not a feature for it
An LLM learns statistical relationships between tokens; it has never seen an object fall. A world model is trained on video, depth and motion so it builds something closer to intuitive physics — what happens if this is pushed, dropped, or occluded — which is exactly the kind of prediction a physical robot, a self-driving system, or (at a much smaller scale) an NPC that has to behave believably in a 3D street needs and a text-only model doesn't provide. Jayverse already has a line on this: the lecun-world-models-jepa item covers Yann LeCun's JEPA architecture, which is one specific proposal for how a world model predicts in a learned representation space rather than pixel space. This item is the "why the whole field is betting on this now" framing that item didn't need to make; together they're one topic, not two.
Lattice cryptography: the signature-scheme question behind account abstraction
NIST standardized the first lattice-based post-quantum algorithms — ML-KEM for key exchange, ML-DSA for signatures — in 2024. "Harvest now, decrypt later" describes the encryption case: an adversary records ciphertext today and decrypts it once a quantum computer exists. Ethereum's exposure is a related but distinct threat: once an account's public key is visible on-chain (which happens the moment it signs a transaction), a future quantum computer could derive the private key via Shor's algorithm and forge later transactions from that address — not "decrypt the past" but "impersonate the account going forward." secp256k1 ECDSA, the scheme every plain Ethereum account uses today, has no defense against this built in.
Account abstraction is the one lever Jayverse has. Under ERC-4337, a smart-contract account's validateUserOp function decides what counts as a valid signature — it doesn't have to be ECDSA. That means a Rabbit account could ship a validator that checks an ML-DSA signature instead, and different accounts could run different schemes side by side, with no protocol fork. The honest constraint: ML-DSA keys and signatures run to several kilobytes against 65 bytes for a secp256k1 signature, and verifying a lattice signature on-chain costs meaningfully more gas than an ecrecover call — a real engineering cost, not a detail to wave away, and as far as jay's research goes, no production post-quantum signer is deployed on Ethereum mainnet today. This is a devnet-and-research problem before it's a shipping one.
AI-led science: the same shape as agentic engineering
The summary describes AI-led science as flipping the usual hypothesis-then-experiment order: instead of forming one hypothesis and running one physical experiment to check it, AI and simulation explore a much larger hypothesis space first, and the physical experiment becomes a verification step at the end rather than the primary search mechanism (54:15, 54:42). That's the same shape as how jay already works with agentic engineering: a spec or plan narrows the space of correct implementations before a line of code is run, and the test suite verifies rather than searches. It's worth naming as a pattern, not a coincidence — "let the cheap, fast process explore broadly and reserve the expensive, slow process for verification" recurs across domains and is worth watching for in every new WEF-style report, not just this one.
Where it lands in Jayverse
- Rabbit / Wallet: a signature scheme is not the same thing as a wallet feature. ERC-4337's per-account validation logic is the reason Rabbit could support ML-DSA-style accounts without touching the base protocol — but there is no production PQ signer on Ethereum mainnet yet, so this is a design note, not a roadmap item.
- Devnet: this is where a PQ validator contract actually gets tested. Before any lattice-signature verifier goes near Rabbit, it belongs on the Anvil devnet (chainId 313370) as a standalone
validateUserOpimplementation with gas benchmarks against the ECDSA path. - Theory: lattices, LWE, and homomorphic encryption are missing entries. Learning With Errors (LWE) is the hard problem ML-KEM and ML-DSA are built on; it belongs alongside the existing ECC and digital-signature notes (math-46) and the ZK entries (math-51, math-52) — lattice cryptography and ZK are both "hard lattice/algebraic problem in, proof or ciphertext out."
- Invest: five of the ten are sector themes, not a company thesis yet. Everything-to-grid energy, direct lithium extraction, and precision fermentation are Economics-tagged watch items — infrastructure and commodity plays, not names to act on from a 55-minute podcast summary alone.
- Eng: "harvest now, decrypt later" is a clean interview answer. It's a compact way to explain both why post-quantum cryptography matters today (not in some future when quantum computers exist) and, with the Ethereum angle above, a concrete example beyond the generic "TLS traffic" one most explanations use.
Verified and unverified
Verified on 2026-09-19: the WEF publishes an annual Top 10 Emerging Technologies report in partnership with Frontiers, and Radio Davos is its real podcast; NIST standardized the lattice-based post-quantum algorithms ML-KEM and ML-DSA in 2024; "harvest now, decrypt later" is a recognized threat model in cryptography; JEPA and world models are covered elsewhere on this site (the lecun-world-models-jepa item). Taken from the summary and not independently checked: the specific contents and framing of the 2026 list, the 40% cooling-energy figure, the 90% clinical-failure figure, the speaker's exact title, and all timestamps. Whether a production post-quantum signature validator exists on Ethereum mainnet was checked as part of this write-up and, as far as this research went, none does as of this writing.
Sources: YouTube — WEF Radio Davos, "The scientific advances ready to change the world: the Top 10 Emerging Technologies 2026" · related items: lecun-world-models-jepa (world models / JEPA) · Tech #62 (agentic engineering writes the boundaries) · Theory math-46 (ECC and digital signatures) · Theory math-51, math-52 (ZK).
Key expressions
| Expression | 뜻 · 쓰이는 자리 |
|---|---|
| WEF | World Economic Forum(세계경제포럼) · 다보스 회의를 여는 기구, 이 보고서의 발행처. "the WEF publishes an annual Top 10 Emerging Technologies report" |
| steering group | 방향을 정하는 운영·자문 그룹 · 목록 선정 주체를 가리킬 때. "an academic-and-industry steering group judged likely to scale" |
| mainstream (v./adj.) | 주류가 되다/주류의 · 기술이 틈새를 벗어나 일반적으로 쓰이는 단계. "likely to scale to mainstream within five years" |
| adsorption | 흡착(표면에 붙잡히는 현상, absorption과 다름) · 직접 리튬 추출의 핵심 메커니즘. "adsorption filters pull lithium from brine" |
| radiative cooling | 복사 냉각(전자기 복사로 열을 방출해 식히는 방식) · 전기 없이 냉각하는 기술. "passive radiative cooling — paints, tiles, window films radiate heat to space" |
| atmospheric window | 대기 창(특정 파장대가 대기를 통과해 우주로 빠져나가는 구간) · 복사 냉각이 작동하는 물리적 이유. "through the atmospheric window with no electricity" |
| PFAS | Per- and polyfluoroalkyl substances(과불화·폴리불화 알킬 물질, "영원한 화학물질") · 탄소-불소 결합 때문에 분해가 어려운 오염물질군. "PFAS destruction — electrochemical electrodes break the carbon-fluorine bond" |
| remediation | 정화·복원(오염된 환경을 되돌리는 작업) · containment(격리)와 대비되는 개념. "moves past containment to actual water/land remediation" |
| bioreactor | 바이오리액터(미생물·세포를 배양해 물질을 합성하는 용기) · 정밀 발효의 생산 장치. "synthesize protein, fats, chemical feedstocks... in bioreactors" |
| exosome | 엑소좀(세포가 분비하는 나노 크기의 천연 소포체, 세포 간 신호 전달용) · 약물 전달체로 쓰이는 이유. "exosome drug delivery — natural cell-signaling vesicles" |
| blood-brain barrier | 혈뇌장벽(뇌를 보호하는 생리적 장벽, 대부분의 약물이 통과 못함) · 신경계 질환 치료의 핵심 난제. "crosses the blood-brain barrier" |
| mRNA | messenger RNA(전령 RNA, 단백질 생산 지시를 세포에 전달) · 코로나19 백신으로 널리 알려진 기술, 여기선 맞춤 암 백신. "personalized mRNA cancer vaccines" |
| harvest now, decrypt later | 지금 암호문을 모아두고 양자 컴퓨터가 생기면 나중에 복호화하는 공격 시나리오 · 포스트양자 암호가 지금부터 필요한 이유. "harvest now, decrypt later" (45:51) |
| NIST | National Institute of Standards and Technology(미국 국립표준기술연구소) · ML-KEM·ML-DSA를 표준화한 기관. "NIST standardized... ML-KEM and ML-DSA in 2024" |
| ML-KEM / ML-DSA | Module-Lattice-based Key-Encapsulation Mechanism / Digital Signature Algorithm(격자 기반 키 교환/서명 표준) · NIST가 2024년 확정한 포스트양자 알고리즘 이름. "ML-KEM for key exchange, ML-DSA for signatures" |
| LWE | Learning With Errors(오류가 섞인 학습, 격자 암호의 기반 난제) · ML-KEM·ML-DSA가 안전성을 기대는 수학 문제. "Learning With Errors (LWE) is the hard problem ML-KEM and ML-DSA are built on" |
| ECDSA | Elliptic Curve Digital Signature Algorithm(타원곡선 전자서명 알고리즘) · 이더리움 기본 계정이 쓰는 현재 서명 방식. "secp256k1 ECDSA... has no defense against this built in" |
| Shor's algorithm | 쇼어 알고리즘(양자 컴퓨터로 소인수분해·이산로그를 다항 시간에 푸는 알고리즘) · ECDSA를 깨는 이론적 근거. "a future quantum computer could derive the private key via Shor's algorithm" |
| ERC-4337 | 이더리움의 account abstraction 표준(스마트 컨트랙트 계정이 자체 검증 로직을 갖게 함) · Rabbit이 서명 방식을 바꿀 수 있는 근거. "under ERC-4337, a smart-contract account's validateUserOp function decides what counts as a valid signature" |
| ecrecover | 이더리움 EVM의 내장 함수(서명에서 서명자의 주소를 복원) · ECDSA 검증의 표준 경로, 격자 서명 검증과 가스 비용을 비교하는 기준점. "verifying a lattice signature... costs meaningfully more gas than an ecrecover call" |
| homomorphic encryption | 동형암호(데이터를 복호화하지 않고 암호문 상태로 연산할 수 있는 암호 방식) · lattice cryptography와 결합되는 기술. "combined with homomorphic encryption, lets medical/financial data be computed on without decrypting it" |
| JEPA | Joint Embedding Predictive Architecture(결합 임베딩 예측 구조) · 얀 르쿤이 제안한 world model 설계, 픽셀이 아니라 표현 공간에서 예측. "JEPA... predicts in a learned representation space rather than pixel space" |