Title: Naveen Rao: 4D Computing, AI's Energy Wall & Beating Biology
Author / source: Naveen Rao (co-founder/CEO, Unconventional AI); All-In Podcast stage; Chamath Palihapitiya joins Q&A (~19:40+)
Source title: Naveen Rao: 4D Computing, AI's Energy Wall & Beating Biology
Source URL: https://www.youtube.com/watch?v=yAsrMA_ADPc
Video ID: yAsrMA_ADPc
Channel: All-In Podcast (@allin)
Published / upload: 20260921 (Mon Sep 21, 2026) — desk flag ~5:04 PM ET per brief
Duration: 22:34 (1354s)
Memo date: Tuesday, 22 September 2026 (America/Toronto)
Views / likes at pull: ~63,465 / ~646 (~39 comments)
Transcript: /workspace/youtube-transcripts/yAsrMA_ADPc.md · Brief: /workspace/youtube-transcripts/yAsrMA_ADPc_brief.md · Meta: /workspace/youtube-transcripts/yAsrMA_ADPc.json
Caption source: YouTube ASR only (en-orig json3 via yt-dlp). No manual English captions. Apply ASR locks below; do not quote raw ASR proper nouns without cleanup.
Source type: Conference / All-In stage talk + Chamath product Q&A. Motivated founder (Unconventional AI CEO pitching a new compute substrate). Channel / format: opinion tech podcast / conference stage — not investment advice.
Product: Institutional PM research map of claims on AI energy wall, von Neumann inefficiency vs biology, dynamical / “4D” computing, Unconventional AI prototype/roadmap, Jevons-scale demand, and product path (rack/DC system). Not advice. No buy/sell from this desk. Language such as “anti-doomer hardware bull / energy-wall / 1000× / Jevons” is source expression / research hypothesis, not an Erica/desk recommendation.
Source discipline: Primary = this transcript + brief + meta JSON only. Companions (Jensen All-In S7CrlFLAmEA, Asianometry x8QP9oXgahA) may be cited only as External / other desk for cross-reference — do not silently import their numbers into underwriting. Flag any other fact as External check needed.
ASR name / token locks (from brief):
| ASR / garble | Intended / note |
|---|---|
| Nirvana systems | Nervana Systems (sold to Intel) |
| platform rising GPUs | MosaicML (GPU infra / model training platform → Databricks) |
| Ali and team at data bricks | Ali Ghodsi / Databricks |
| unconventional AI | Unconventional AI (confirm branding) |
| map will | matmul / matrix multiply |
| Jeban's / Jevons | Jevons paradox |
| VM that sits somewhere | Product framed as managed rack/DC appliance (not classic VM) — Chamath prompt; Naveen reframes as DC rack/system |
| Uno | Image-gen model on oscillators (as named) — confirm public write-up External |
Sponsor noise (ignore as claims): IREN, Oracle, EY, Meta, Keel Infrastructure, Airwallex, PayPal, Google for Startups (and cold-open montage).
Guest pedigree (as stated on tape / brief): Founded first AI-chip co Nervana (2014) → sold to Intel, ran Intel AI group; then GPU infra / model-platform co (MosaicML) → joined Databricks (2023); Mosaic path now “~¼ of Databricks total revenue” (as-spoken); now Unconventional AI (rethink computer for power efficiency).
Takeaway 1 — Anti-doomer hardware bull: energy wall is real (~3 years); Unconventional bets ~1000× efficiency via dynamical / 4D compute to monetize watts and unlock Jevons-scale AI demand. Naveen opens as “the opposite of an AI doomer”; frames AI as transformational / next evolution of humanity; argues the binding constraint is energy, not floor space / networking / GPUs alone. Business case as-spoken: monetize each watt 1000× better than existing hardware. Source expression: anti-doomer hardware bull. Not a desk trade recommendation. Conviction on what was said: High.
Takeaway 2 — Quantified energy wall (source-locked numbers). Google alone: 3.2 quadrillion tokens / month (public, as cited). At ~10 J/token (stated as low end of spectrum) → ~12 GW for that one company’s AI services. US data-center power ~40 GW; US ~half world DC capacity → world <~100 GW. ~50% of token serving cost = energy; rest CapEx / hardware / floor. Scarce input shifted: floor space → networking → GPUs → power contracts. Run-out estimate: ~3 years if models + demand keep growing. 2030 AI market frame: call it ~$1T (maybe bigger) vs linearized energy supply → gap. All as-spoken; External check needed before model input.
Takeaway 3 — Biology as existence proof; von Neumann shuttling as the inefficiency. Human brain ~20 W; monkey-scale ~1 W (≈ phone); rats/bats mW; squirrel ~8 mW — “100+ squirrel brains on your phone.” Cortex moves ~16B bits/s (~13–14B neurons); GPU / high-end moves ~30T bits/s in/out of memory (outside chip); inside chip 10–100× more. Energy dominated by moving information. Computers since ENIAC-era still memory↔compute shuttle; optimized for speed, not energy. Moore’s-law efficiency from smaller transistors “largely ended.”
Takeaway 4 — Tech thesis: cut abstractions; dynamical computer; “4D computing.” Digital 0/1, then stacked abstractions, then neural nets on top — each layer lossy. Connect semiconductor physics more directly to neural networks (brain has no linear algebra / FP math). Dynamical systems theory (metronome sync, flocking, ant colonies) → circuits. Prototype image model Uno on oscillators (simulated, open-sourced as claimed). Sparsity: throw away connections → better scalability + more trainable (n² problem). Physical dynamical computer: company earnest Jan; taped out Jun 1; chip in lab with results; first public reveal of generated images. Energy: ~500 nJ/image vs GPU mJ order — many orders of magnitude. Architecture: compute+memory co-located (no classic memory interface); time as dimension + 3D die stacking = “4D computing.” Still claims existing model families work after model-layer port (not ops layer); no classic matmul implementation — time-varying behavior analyzable as state × transition matrix.
Takeaway 5 — Roadmap & product path (Chamath Q&A). Efficiency goal was 1000× in 5y → revised to ~3.5y (AI accelerated their science). Org stack: theorists (math / theoretical neuroscience) → models on real data → physical circuit design → board/system/product. Hardest org problem: bridging dynamical-systems theorists and chip builders who don’t talk. First product = new DC rack/system (tokens in/out over network; different guts); full product ~2 years. Software: Python libraries for time-varying stochastic elements (explicitly not CUDA analogy). Decade goal: beat biology → compute everywhere / robots; deployment shift from giant GW campuses → many small local DCs; enable billions of robots. Thermodynamic framing: today ~10 billion× away from limit; mammalian brains within 1–2 orders; hit 2D lithography limits in ~3.5y.
Takeaway 6 — Jevons paradox as demand upside, not demand destruction. 1000× cheaper compute → consume more than 1000× → “largest market humanity’s ever seen.” Desk inference (hypothesis only): efficiency breakthroughs, if real, are bull demand for intelligence / robots / local DCs, not a simple “GPU CapEx dies” story — but timing and who captures the watt-monetization are the open questions. vs Jensen All-In (S7CrlFLAmEA) — External companion: both anti-doomer / buildout-positive; Jensen emphasizes GPU ecosystem + NeoCloud GW; Naveen emphasizes substrate replacement if energy wall binds — do not import Jensen numbers. vs Asianometry (x8QP9oXgahA) — External companion: scarcity/glut framing elsewhere on desk; this tape is founder-level energy-wall → new machine — structure cross-ref only, no number import.
Takeaway 7 — Non-obvious angle for PMs. Headlines will hear “1000× chip” and “beat biology.” The underwriting spine is quieter: (i) power contracts already the scarce DC input and ~50% of token OpEx as claimed; (ii) if wall hits ~3y on current substrate, any credible path to orders-of-magnitude efficiency becomes allocation-relevant even at low prior; (iii) product is first a rack/DC appliance (tokens in/out), not a phone SoC — so near-term comps are infra/system vendors and hyperscaler attach, not consumer silicon; (iv) model-layer port + no classic matmul = ecosystem / software switching-cost risk as large as fab/process risk; (v) Jevons framing means success may increase total energy demand while reducing energy per token — utilities / local DC / robot theses can coexist with “GPU less efficient per watt” theses; (vi) motivated founder — treat as incentive-aligned but internally consistent with pedigree (Nervana → Intel AI → MosaicML → Databricks revenue claim).
Takeaway 8 — Why now + horizons. Why now: upload Mon 21 Sep 2026 (day after All-In conference cycle with Jensen stage companion); first public physical dynamical-computer reveal; ASR-only tape for desk; energy / power-contract narrative already in market conversation. Horizons: (Near, 0–6m) prototype credibility — independent measurement of nJ/image claims, Uno write-up, team hires, capital raises (External). (Medium, 6–24m) first rack/system product path (~2y as-spoken); model-layer port demos on existing families; power-contract attach pilots. (Long, 2–5y+) 1000× / 3.5y claim; 2D litho limit hit; local-DC / robot enablement; “beat biology” decade goal. Companion desk triangulation for frame conflict only — Jensen anti-pause buildout; Asianometry scarcity/glut — no silent number merge.
Takeaway 9 — Conviction. High that the source is a coherent anti-doomer / energy-wall (~3y) / 1000× dynamical-4D / Jevons-demand research hypothesis from a motivated serial founder. Medium that as-spoken macros (3.2Q tokens, 10 J/token → 12 GW, 40 GW US / <100 GW world, 50% energy share, 500 nJ/image, Mosaic ~¼ Databricks rev, 3.5y / 2y timelines) survive primary checks. Low as calibrated probability that Unconventional ships a rack that displaces meaningful GPU serving share on the stated clock — private company, early silicon, ASR-only demo claims. Hypotheses and watchlist only. No buy/sell.
Takeaway 10 — Incentive & process flags. Motivated founder pitching Unconventional AI on a conference stage; applause segments; first public chip reveal designed for narrative impact. Mosaic→Databricks “~¼ of total revenue” is as-spoken founder claim — External check needed. ASR-only: Nervana/MosaicML/matmul/Jevons locks mandatory. Packet routing: chat / file only — do not publish to here.now; do not email; not for auto-handoff unless AP directs.
Takeaway 11 — What this memo is not. Not a rating on NVDA, SMH, hyperscalers, utilities, or Unconventional (private). Not a claim that 500 nJ/image generalizes to frontier LLM serving. Not a silent merge of Jensen or Asianometry quantitative frames. Not investment advice. Thesis language is source-locked research hypothesis.
Takeaway 12 — Research next step from this summary alone. Treat energy wall + watt-monetization as the allocation research question; treat 4D dynamical claims as diligence agenda (measure, portability, product attach); keep Jevons as the demand-side hedge against “efficiency kills CapEx” shorts; tag all macros External check needed before any model.
Chapter-ordered from official description / meta JSON. Quotes ≤20 words. Attribution by speaker. Timestamps approximate (ASR).
(0:00–0:38) Intro montage / stage welcome. Hosts: Naveen Rao, co-founder/CEO Unconventional AI; “definitionally outlier founder”; built/sold two deep-tech cos; “I’m the opposite of an AI doomer”; “We need innovation on the hardware substrate.” Standing welcome. Ignore sponsor mid-rolls / cold-open music throughout.
(0:38–1:12) Anti-doomer frame. Naveen: “I’m the opposite of a doomer”; AI “one of the most transformational technologies”; “next level of evolution”; All-In as “anti-doomer conference.”
(1:12–1:45) Bio — early computers / EE / PhD neuroscience. Childhood computer ~1978 / early ’80s; programmed as puzzle; electrical engineer from sci-fi / intelligent-machine ambition; later PhD in neuroscience — “How do we make computers intelligent?”; world moved that direction.
(1:45–2:51) Nervana → Intel; then Mosaic path. Founded “first AI chip company” ASR “Nirvana systems” → lock Nervana (2014); hard to raise when AI not vernacular; sold “way too early” to Intel; started/ran Intel AI group. Post-2020: infrastructure for bigger models / LLMs; started ASR “platform rising GPUs” → lock MosaicML; after ChatGPT 2022 “best game in town”; joined Databricks 2023 with “Ali and team” → Ali Ghodsi.
(2:51–3:24) Mosaic economics claim; Unconventional pitch. Mosaic path “actually that’s a quarter of the total revenue of data bricks today” — as-spoken. Unconventional AI: “rethinking the foundations of how a computer works”; singular purpose power efficiency.
(3:24–3:57) 1000× timeline revised. Initially within 5 years to 1000× power efficiency; revised to three and a half years — “things have gone faster”; “solved very deep scientific problems quicker because of AI.”
(3:57–4:45) Org stack top-to-bottom. Theorists (math PhDs, theoretical neuroscience) → concepts that move less information → models trained/evaluated on real data → physical circuit architects/designers → board/system/product. Segue: “So, is energy really a problem?”
(4:45–5:37) Google tokens → GW math. One company (Google, public): 3.2 quadrillion tokens/month. Assume 10 joules/token (low end) → 12 gigawatts. US puts ~40 GW into data centers; US ~half world DC capacity → world under 100 GW. “12 gigawatts is going into one company just for AI services.”
(5:37–6:13) Run-out / gap graphic. Models bigger + demand growing → energy up; “run out of energy pretty fast, in like 3 years or so.” Exponential AI market vs linearized energy: call it trillion-dollar market in 2030, maybe bigger; gap is the problem; solve with technology.
(6:13–6:46) Scarce input shift; 50% energy. DC thinking: floor space → networking → GPUs → energy first (get power contract, then fill with GPUs). “About 50% of the cost of serving a token… is energy”; rest CapEx/hardware/floor.
(6:46–7:18) Watt monetization business case; brain 20 W. Get power contract → monetize every watt; “monetize that 1,000x better than existing hardware.” Biology proof: human brain ~20 watts.
(7:18–7:51) Animal brains / phone comparison. Monkey-scale ~1 W (phone ~1 W); rats/bats milliwatts; squirrel brain eight milliwatts; “over a hundred squirrel brains on your phone”; precise motor behavior.
(7:51–8:55) Understand by creating; bit-rate contrast. Quote frame: don’t truly understand until we can create it. Most energy in computing = moving information. Cortex ~16 billion bits/s (~13–14B neurons); GPU/high-end ~30 trillion bits/s memory in/out; inside chip 10–100× more — drives energy demand.
(8:55–10:24) History of computers; ENIAC; Moore end. Mechanical → analog → digital 1930s–40s; operation similar today — memory outside, compute, shuttle bits; built for speed not efficiency. ENIAC (1945) artillery trajectories faster than human computers; today sell “twice the speed.” Transistor counts up, frequency/single-thread stalled; efficiency from smaller transistors “largely ended” (Moore). Need rethink: “cut out the middle man.”
(10:24–11:07) Abstractions are lossy. Digital 0/1 is abstraction of transistor continuum; stacked abstractions → neural nets on top; each lossy. Simplify: abstraction of semiconductor physics connected to neural network. Brain: neurons, “no linear algebra… no floating point math”; physics gives rise to intelligence — mimic with semiconductor.
(11:07–12:14) Dynamical systems in nature. Computation in nature: bird flocks, ant colonies; dynamical systems theory — emergent properties from simple component rules; brain works this way; building circuits from these ideas. Metronome example: multiple metronomes on rolling plank synchronize via physics.
(12:14–13:54) Metronomes → Uno image model. Can such a system do computation for generative AI? Released model Uno — image generation on oscillators; simulated; open-sourced as claimed; state-space trajectories conditioned on class (airplane/car/bird); actual generated images shown.
(13:54–15:33) Sparsity as holy grail. Dense all-to-all = n² connections (10→100; 1000→1M). Sparsity: throw away connections, rescue (even improve) behavior; more trainable; works in simulation and physical systems; “more efficient… more scalable… more performance.”
(15:33–16:06) First physical dynamical computer. “First time… talking about this publicly”; “first physical dynamical computer ever built”; company earnest January (no team initially); taped out June 1; chip back in lab with results; first images from such a computer. Applause.
(16:06–16:39) 500 nJ/image claim. Beyond images: sequence modeling / language models possible as claimed; ~500 nanojoules per image vs GPU millijoules order; “many orders of magnitude more efficient”; “doesn’t move information around”; “proof positive.”
(16:39–17:46) Von Neumann vs dynamical; 4D computing. CPU→GPU→compute-in-memory still von Neumann (memory↔compute shuttle). Dynamical computer: compute and memory co-located; no memory interface; each element is memory. 4D computing: time/dynamics + physical 3D die stacking (vertical + planar).
(17:46–18:52) Thermodynamic limit; beat biology; local DCs / robots. Intelligence per watt; thermodynamic limit never exceedable; mammalian brains within 1–2 orders; today ~10 billion× away. In 3.5 years hit limits of 2D lithography; company goal beat biology this decade → compute everywhere / robotic forms. Shift: big GW campuses → many small local DCs; environmentally friendlier, local, adaptive; enable billions of robots.
(18:52–19:40) Jevons paradox close. AI ~trillion-dollar market; disrupt by 1000× → Jevons paradox (ASR “Jeban's”): drop cost → consume more than the drop; 1000× cheaper → consume more than 1000×; “largest market that humanity’s ever seen.” Applause; Chamath: “extremely unexpected.”
(19:40–20:29) Ecosystem / product / timing. Chamath: need fabs, packagers, ecosystem beside you (nod to Jensen earlier — External companion context only); path from early version to something people use? Naveen: within 2 years to full product. Product = new data center product first — whole rack/system; tokens in/out through network cable; “inner guts… completely different.” Chamath “VM” prompt → Naveen reframes as managed DC rack/system (ASR lock).
(20:29–21:35) Porting models; matmul. Chamath: existing model families / KV cache / reductive abstractions? Naveen: sliding scale of better vs pain; make move compelling; port at model layer not operations layer; existing models will work but “fair bit of compute” for transition. Matmul (ASR “map will”): can characterize as matmul but doesn’t implement as matmul; time-varying behavior; each timestep analyzable as state × transition matrix.
(21:35–22:34) Team span; Python libs. Team = dynamical-systems theorists (field ~100 years) + chip builders — “they don’t talk to each other”; facilitating that span “one of the most challenging things.” CUDA analogy denied: Python libraries expressing time-varying stochastic elements. Chamath close: ambitious; thanks.
Source-locked chain (power → DC → substrate → models → apps/robots):
Power / energy contracts (scarce input). As-spoken: today’s DC scarce input is the energy contract first, then fill with GPUs/infra. ~50% of token serving cost = energy. US ~40 GW DC / world <~100 GW; Google AI alone framed at ~12 GW under 10 J/token. Binding constraint narrative: ~3-year energy wall if growth continues.
Data-center shell / networking / floor (demoted but still necessary). Historical scarce inputs: floor → networking → GPUs. Still needed to host racks; Unconventional first product is a whole rack/system sitting in that shell with tokens over network.
Incumbent compute substrate (GPU / von Neumann stack). CPUs → GPUs → compute-in-memory — still memory↔compute shuttling; optimized for speed; energy dominated by moving bits (~30T bits/s memory I/O vs cortex ~16B). Pedigree contrast: Nervana (early AI chip) → MosaicML (GPU scale infra) → now escape that substrate.
Unconventional dynamical / 4D substrate (claimed alternative). Oscillator / dynamical-systems circuits; sparsity; compute+memory co-located; time dimension + 3D stacking. Prototype silicon taped out Jun 1; ~500 nJ/image claim. Org: theorists → models → circuits → boards → product.
Model layer / software port. Not CUDA; Python libraries for time-varying stochastic elements. Port at model layer (compute-heavy transition); no classic matmul implementation (state × transition characterization). Existing model families claimed workable after port — ecosystem attach risk sits here.
Serving economics / watt monetization. Business case: monetize each watt 1000× better. If true, power-contract holders + Unconventional attach capture surplus; if false or late, GPU stack continues to clear scarce watts at lower intelligence/watt.
Deployment topology (long horizon). Source vision: from giant GW campuses → many small local DCs; enable billions of robots; “beat biology” decade goal; compute everywhere.
Demand multiplier (Jevons). 1000× cheaper intelligence → consume >1000× → “largest market humanity’s ever seen.” Efficiency success ≠ demand destruction in this narrative.
[Energy / power contracts] → [DC shell / network]
↓
[Incumbent GPU / von Neumann] ↔ [Dynamical / 4D Unconventional rack]
↓
[Model-layer port / Python libs]
↓
[Token serving / agents] → [Local DCs / robots / “compute everywhere”]
↓
[Jevons: more total intelligence demand]
Incentive note: Founder of Unconventional AI has every reason to narrate an imminent energy wall and a 1000× substrate breakthrough. Pedigree (Nervana, Intel AI, MosaicML, Databricks revenue claim) adds credibility and sales skill. Treat as motivated but coherent; separate claim classes (public Google token figure vs private nJ measurements vs timeline promises).
Causal chains. S = source claim; I = desk inference (hypothesis). Companion citations = External / other desk only — no number import.
S7CrlFLAmEA) — External: Jensen also elevates land/power/shell; Naveen quantifies wall with Google/GW math — cross-frame only; do not import Jensen figures. x8QP9oXgahA) — External companion: scarcity/glut debate is a different tape; use for opposing/companion frame only; no number import. Probabilities are desk research hypotheses for organizing diligence — not forecasts or trade tickets. Motivated founder speaker.
Assumptions: Energy wall narrative confirmed by hyperscaler power constraints within ~3y; Unconventional (or peer dynamical/analog approaches) ships rack product with verified orders-of-magnitude energy/token gains; model-layer ports for at least one production family complete; power-contract holders adopt alternative racks for inference slices; Jevons lifts total token/robot demand.
Winners (hypothesis): Unconventional / similar substrate startups (private); power owners who re-monetize watts at higher intelligence/MW; local/micro-DC and robot-adjacent infra; selective packaging/3D-stack suppliers.
Losers (hypothesis): Pure “GPU CapEx forever linear” theses that ignore watt scarcity; inefficient serving stacks; late NeoClouds without power or efficiency path.
Leading indicators: third-party nJ or J/token audits; named hyperscaler/NeoCloud pilots; 2y product ship evidence; rising intelligence-per-MW disclosures.
Assumptions: Power remains scarce and ~token energy share stays material; GPU/packaging efficiency and software continue incremental gains; Unconventional prototype is scientifically interesting but product attach slips past 2y; model-port friction high; 1000×/3.5y misses but 10–100× research progress continues; Jevons mostly shows up as more GPU tokens, not rack displacement.
Winners: Diversified AI infra (power, interconnect, incremental GPU efficiency); hyperscalers with locked PPAs; tooling that reduces bits moved (sparsity, better serving).
Losers: Narrative-only “beat biology” equity stories; CapEx without energization; pure pause theses that ignore demand.
Leading indicators: utilization vs energized MW; Unconventional fundraising vs missed milestones; incumbent efficiency roadmaps; token growth vs DC GW growth (External).
Assumptions: 500 nJ/image does not generalize to frontier LLM serving; fab/yield/3D-stack or noise/precision issues block scale; theorists–chip-builder org fails; labs refuse model-layer ports; energy supply linearizes up (gas, nuclear, interconnect) enough to defer “3y wall”; founder narrative premium collapses on missed 2y product.
Winners (relative): Incumbent GPU ecosystem and CUDA software moat; utilities/developers who still energize conventional DCs; skeptics of analog/dynamical compute history.
Losers: Unconventional and peer “new machine” narratives; investors who underwrote 1000× on stage demos; any book that shorted watts assuming efficiency kills demand (Jevons fail and wall deferred is different bear).
Leading indicators: failed customer PoCs; adverse technical replications; silence after Jun tape-out hype; large new conventional GW energized on schedule (External).
Hypotheses for research queues — not trade tickets; no buy/sell.
| Asset / cluster | Thesis hook from source | Metrics / catalysts to watch | Risks |
|---|---|---|---|
| Unconventional AI (private) | Dynamical/4D computer; 1000×/~3.5y; rack product ~2y; 500 nJ/image prototype | Tape-out follow-ons, third-party energy metrics, rack LOIs, capital raises, fab partners | Founder narrative; demo≠product; org-span; ASR-only claims |
| NVDA / GPU ecosystem | Incumbent von Neumann/GPU substrate; Chamath nods Jensen/ecosystem need; bits-moved energy critique | Efficiency roadmap, serving J/token, attach to power-constrained DCs | Displacement if alternative racks work; also beneficiary if Jevons lifts total tokens before displacement |
| AMD / custom silicon (hyperscaler ASICs) | Alternative paths to watt monetization without Unconventional physics | Inference ASIC deployments, J/token vs GPU | Same wall; different architecture bets |
| TSM / advanced packaging / 3D stack names | “4D” uses physical 3D die stacking + time | CoWoS/SoIC-class capacity, stacking yields | Cycle, customer concentration; Unconventional may or may not be meaningful volume |
| Hyperscalers (GOOGL explicit; MSFT/AMZN/META orbit) | Google 3.2Q tokens public cite; power contracts first; ~50% token cost energy | Token growth, DC GW, PPA/interconnect, energy share of COGS | CapEx air-pocket; custom silicon; regulation |
| NeoClouds / DC REITs / shell developers | First Unconventional product = DC rack/system; topology → many small local DCs long-term | Energized MW, utilization, willingness to host non-GPU racks | Financing; stranded shells; GPU-only designs |
| Utilities / IPPs / nuclear / gas peakers | Energy wall ~3y; watt monetization; Jevons may raise aggregate demand | Interconnection queues, AI offtake PPAs, local-DC load growth | Policy, permitting, overbuild |
| Databricks (private) / Mosaic path | Pedigree + “~¼ of Databricks total revenue” as-spoken | Mosaic contribution verify (External); talent spinouts | Founder claim inflation |
| Intel (pedigree only) | Nervana sale; Naveen ran Intel AI group | Historical context only — not a thesis from this tape | N/A from primary |
| Robot / edge compute OEMs (long-dated) | Billions of robots; beat biology; compute everywhere | Only after rack credibility — watch as optionality | Decade vision; timing |
External / other desk cross-read only: Jensen All-In (S7CrlFLAmEA) for anti-doomer buildout / NeoCloud/power frame; Asianometry (x8QP9oXgahA) for scarcity/glut frame — structure only, no number import.
| Claim | Speaker | Status |
|---|---|---|
| Google 3.2 quadrillion tokens/month | Naveen | As-spoken (cites public) |
| ~10 J/token (low end) → ~12 GW | Naveen | As-spoken math |
| US DC ~40 GW; world <~100 GW | Naveen | As-spoken |
| Energy wall ~3 years | Naveen | Estimate |
| ~$1T AI market 2030 (maybe bigger) | Naveen | Frame |
| ~50% token serving cost = energy | Naveen | As-spoken |
| Monetize watts 1000× better | Naveen | Business case |
| Brain ~20 W; monkey ~1 W; squirrel ~8 mW | Naveen | As-spoken |
| Cortex ~16B bits/s; ~13–14B neurons | Naveen | As-spoken |
| GPU ~30T bits/s memory I/O; 10–100× inside | Naveen | As-spoken |
| 1000× in 5y → revised ~3.5y | Naveen | Roadmap |
| Full product ~2 years (DC rack/system) | Naveen | Roadmap |
| ~500 nJ/image vs GPU mJ | Naveen | Prototype claim |
| ~10 billion× from thermodynamic limit; brains within 1–2 orders | Naveen | Framing |
| Mosaic path ~¼ of Databricks total revenue | Naveen | Founder claim — verify |
| Company earnest Jan; tape-out Jun 1 | Naveen | As-spoken chronology |
IREN, Oracle, EY, Meta, Keel Infrastructure, Airwallex, PayPal, Google for Startups
S7CrlFLAmEA) — /workspace/pm-memos/2026-09-14-allin-jensen-huang-doomer-hoax.md x8QP9oXgahA) — /workspace/pm-memos/2026-09-14-asianometry-silicon-valley-starving-compute.md/workspace/youtube-transcripts/yAsrMA_ADPc.md /workspace/youtube-transcripts/yAsrMA_ADPc_brief.md /workspace/youtube-transcripts/yAsrMA_ADPc.json /workspace/pm-memos/2026-09-22-allin-naveen-rao-4d-energy.mdArtifact only. Markdown saved to the path above. Do not publish to here.now. Do not email. Do not auto-handoff to Compound or Jordi. Deliver in chat for AP / Erica desk packet routing. Thesis language (anti-doomer hardware bull / energy wall ~3y / 1000× dynamical-4D / Jevons demand) is source-locked research hypothesis, not a desk recommendation. Motivated founder. ASR locks enforced. No buy/sell.
— End of PM RESEARCH MEMO —
Desk copy · not a trade recommendation · Erica · 22 Sep 2026