Erica · PM Research MemoAll memos

PM RESEARCH MEMO

Title: The Bond Market “Crash” Is a Trap — AI Agents Are About to Ignite Crypto
Author / source: Jordi Visser (solo YouTube; channel @JordiVisserLabs / Visser Labs)
Source title: The Bond Market “Crash” Is a Trap — AI Agents Are About to Ignite Crypto
Source URL: https://www.youtube.com/watch?v=er5mqvbDQU8
Video ID: er5mqvbDQU8
Published: 2026-09-06 (YouTube upload calendar day; America/Toronto; exact clock time not in yt-dlp/page metadata)
Duration: 46:39 (2799s)
Views: 58,491 (as of transcript retrieval)
Memo date: Tuesday, September 8, 2026 (America/Toronto)
Transcript: /workspace/youtube-transcripts/er5mqvbDQU8.md · Brief: /workspace/youtube-transcripts/er5mqvbDQU8_brief.md
Caption source: YouTube English automatic captions (ASR) only (yt-dlp timedtext; yt-dlp --write-auto-sub returned 429 at retrieval). Timestamps below are cue timestamps from the retrieved timedtext, not interpolated.
Source type: Macro / thematic education talk. Speaker disclaimer throughout: not investment advice; educational; disclosures on site. This memo contains no trade recommendations from this desk.
Product: Regime map and research hypotheses for allocation research. Not advice.
Source discipline: Primary sources are this transcript and its brief only. Companion Jordi / Alma / SV / Cboe memos on this desk are different sources — do not silently merge numbers. Speaker-stated figures are used as spoken and attributed; they are not independently verified (External check needed).

How to read this document: Restatements of the talk are Source. Interpretive links and underwriting judgments are Inference. Any fact not spoken in the transcript is External check needed. Handicapping language (~5% probability → ~5% allocation) is source expression / research hypothesis, not an Erica recommendation.

ASR name locks:

ASR Intended
Bonval / bonvs bond vol
Besson / Bessant Scott Bessent
infant (P&L / costs) inference
Grockbot / Grock Grok bot / Grok (xAI)
Moonay MoonPay (likely)
viser-labs / visserylabs visser-labs / Visser Labs
waitings weightings
50inute 50-minute
IO macro nexus AI / macro nexus (likely)
Liupold / Liupole unclear proper name (possible Leopold / situational-awareness figure; do not lock ticker)
Enthropic Anthropic
Chat GBT ChatGPT
Salana Solana
Marll Marvell
Tro (RWA) unclear (RWA issuer/manager; do not lock)
Jensen Yuang / Jensen Senuang Jensen Huang
PERS PERPS (perpetual futures)
MCI World MSCI World
socks SOX (semiconductor index)
RAIA / RAAS / FAS RIA / RIAs / firms (context)
Open Face OpenAI / Hugging Face podcast episode (title as spoken)
Astra Astra (model name as spoken; identity External check needed)
Clarity Act Clarity Act (US crypto market-structure bill, as framed)

EXECUTIVE SUMMARY


SOURCE-ACCURATE SUMMARY

Chapter-ordered with cue timestamps. Short quotes only where they carry the claim (≤20 words).

Framing — empowerment, agents, crypto focus (00:00–02:50)

  1. (00:00–01:03) Three-day weekend / last true summer day. Final four months of the year = “most important point in AI where the AI agents are growing rapidly.” Video is education / empowerment, not a guarantee of money. If you understand agentic AI and crypto, “you are at the most important part of the disruption.”

  2. (01:03–02:50) Old path (good school → job → raises → retire) “is not the way the world works anymore.” New world: learn anywhere; collaborate with AI; now send agents off to build, then staff-meeting. Young people who learn agents can build businesses; investing via crypto / tokenization / tokenized products — “not through speculating” but agents building diversified tokenized portfolios with “a good Sharpe ratio.”

Bond fears are the wrong theme (02:50–07:17)

  1. (02:50–03:39) Major theme: “stop with the bond fears.” Same cohort that called AI a bubble now says bonds explode higher. Chance yields go up: yes. 10y to 5.5%: “sure.” Will Bessent fight it: “Absolutely.” Why obsess over a range-bound move that already happened?

  2. (03:39–04:29) 30y ~5.23%, could make highest since 2004 soon — “don’t believe the hype.” If crisis: bond vol would be elevated; it is not. Junk spreads at “tightest level ever”; CCCs widening — he expects more AI-driven business failures → some spread widening in weak credits, not systemic junk blowout.

  3. (04:29–05:32) Inflation swaps 1y/2y/5y/10y ~2.48–2.52%, same level, matching core CPI ~2.48. Daily Trueflation falling fast — he insists it has value (real prices). 20yr+ total return YTD −3% vs −32% in 2022.

  4. (05:32–06:26) JGB 30y: May last year breakout to ATHs; +117 bp since; Zero Hedge “biggest bond market is imploding” (May 20); Nikkei +52% YoY — “cost yourself a 70% rally.”

  5. (06:26–07:17) Robin Brooks: selloff since 2022, not new; won’t become crisis because governments won’t let it; about debt levels + good growth; completely global; inflation break-evens asleep. Bessent: expects Japan to boost yen; “I know what the Japanese are planning”; US–Japan coordination; yen strengthened this week.

Equity / growth facts vs bond obsession (07:17–15:50)

  1. (07:17–09:02) S&P PEG lowest since ~1998 (PE~20 / growth~25 example). Weekly revisions strong. Time to be short stocks was 2022 when rates rose and Fed said don’t own stocks; now president and Bessent want rates not higher, “run it hot,” stocks higher.

  2. (09:02–10:41) Recession signals absent vs ’08 analogs. YoY: S&P +18, Nasdaq +24, Russell +23, SMH +88. Global: MSCI World +16, EM +30, World Bank Index +29 — banks would not lead if bond/credit crisis. Revenues/EPS revisions “through the roof”; valuations “inside the bands at below 20” because earnings rising.

  3. (10:41–12:16) Global profit margins rising — “shouldn’t long equities be a better thing?” Pre-recession usually see flattening 1–2 years prior. Oil ~$86 Dec; sixth forward +22% YoY = healthy spending signal. Redbook 5-week YoY avg 9.6%; last August week back up; overlay with personal spending; consumption ~67% of economy. Possible Fed +25 bp not a reason to avoid stocks while earnings/margins grow.

  4. (12:16–14:08) S&P had first real correction of year after Iran-war / AI-trade peak; three-month ROC hit zero around “Liupold” blow-up then bounced; AI “bottoming” in his opinion. Adv/dec 9-day RSI hit 25 (March low analog). Nvidia approaching ATHs; chip complex follows Nvidia as “head of the circular finance.” If technician bearish semis and bullish yields and bearish Bitcoin → possible “endgame bias.”

  5. (14:08–15:50) SMH below 50-day / trending down = fact; Nvidia “up here” = fact until breakdowns. Consolidation, not end. S&P/QQQ/Russell above 200-day, roughly unchanged Jun→Sep. Tech-momentum new lows = long-side deleveraging, not shorts covering like prior episode; “stuck in mud”; expects eventual highs, maybe after midterms; made buys (e.g. Micron) in weakness for a year-out higher view. “Most of my weightings are towards crypto and I continue to add to crypto.”

Thematic portfolio internals & circular AI economics (15:50–21:11)

  1. (15:50–18:22) Open YouTube remains; thematic + crypto sleeve + prompts moving to subscribers. Two blah weeks then improvement in breadth of 100-name book. ATR% (avg 14-day true range of 100 names) falling weekly = positive (fewer stops; retail less involved). Believes thematic AI bottom in; will restart spotlights; Nvidia now as large as Marvell for him; relative strength while stuck in mud.

  2. (18:22–20:20) “Interest rates are a rounding error on inference P&L.” Anthropic, OpenAI, Nvidia matter most. Model price −10% → ~3 margin pts; rates +200 bp → ~1 margin pt. Concentrated economy; risk = model-price compression (not yet). Circular: model pricing supports compute → Nvidia spend → Dell/Marvell/Micron.

  3. (20:20–21:11) Nvidia quarter blowout then recovery = positive; Dell +9%, AI server revenue tripling (was doubling). Non-US compute shortage. Astra best on benchmarks; AGI-era article: ~40% → ~100% on knowledge-worker job capability; agents, not overnight mass unemployment.

Agents, workflows, Andreessen merge (21:11–34:21)

  1. (21:11–24:03) Must-listen podcast: Artificial Intelligence Show / OpenAI–Hugging Face episode — part hype, part agentic reality. Expect significant fiat-world hackings; crypto app hacks yes; Bitcoin protocol hacks no until quantum. Agents in teams of ~1,200; some “sacrificed their own goals for the collective” — changes mental model from AI assistant to AI organization. Browser agents / Codex: no time to wait. Business leaders giving everyone Anthropic chat ≠ workflow redesign → fall behind AI-native firms.

  2. (24:03–26:42) Bill Gates shifted view because agents move faster than human adoption. Takeaway: prepare for AI and crypto; next three years “insane.” OpenAI/Anthropic: next year ≈ four years of prior AI progress. Crypto + tokenization = compounding vs old 401k 10% world. Gavin Baker podcast: Grok-style agents as “new consumption shock” (timestamp cite 17:15–20:34 of that podcast).

  3. (26:42–29:12) Path: chatbots → coding agents → personal/workflow agents → financial agents (where crypto fits). Jensen Huang knowledge-brain prompt; Grokbot personal workflow agent (read/summarize/compare/recommend/ask permission). Grok now buy/lend crypto via MoonPay; Stripe Link purchase anywhere — try Grokbot.

  4. (29:12–30:55) Built Grok agents (Jarvis → Pattern Scout, Universe Clerk) for daily post-close pattern screen of “AI macro nexus 100” via Yahoo Finance — education on using agents, “not telling you to buy anything.”

  5. (30:55–32:38) Andreessen Apr podcast (Latent Space): held back then because payments weren’t ready; now: “AI is the crypto killer app.” Crypto phase-1 was human embrace; agents need stablecoins/tokenization. BTC = ecosystem (like S&P = US economy). “Why Bitcoin Matters” (2014) was preparation. Bitcoin = “purest AI hedge” because AI destroys businesses; new firms small/entrepreneurial. Failure/retry = trading mindset.

  6. (32:38–34:21) “Not humans, but AI agents will be the biggest users of crypto” → billions of machine consumers; equate crypto to tokens; crypto infra should rhyme with AI infra token expansion. Clarity Act: matters for pensions/insurance capital and preventing four-year policy rollback; not the sole existence condition. “Crypto didn’t lack a use case. It lacked a native user. AI creates it.”

Crypto tape, tokenization, double debasement, handicap (34:21–46:39)

  1. (34:21–36:01) BTC at technical level, consolidating — “very bullish.” Silent IPO: ideologues bailed/reduced; many missed the move, feared return to “60s,” expected October bottom — “I don’t think you’re going to get the chance.” ETH & SOL: 200-day turned up; consolidating; “three horsemen” weekly plus 46-name crypto portfolio (6 stocks / 40 tokens) launching this month on subscriber side.

  2. (36:01–37:54) QTD: ETH +56, SOL +38, BTC +36 vs S&P +3, Nasdaq −3, SMH −14. Expects crypto higher into year-end; earnings grow ~30%/yr while “these things” grow at half — double debasement story: crypto takes higher % of fiat assets over next five years via AI disruption + government preventing higher rates / managing K-shape. Robinhood soars on Wall Street upgrades; L2 / stock-token trades; $1.5B in six weeks — “looks kind of like the token expansion in AI.”

  3. (37:54–39:49) RWA ~18×; phase-1 collateral/money markets (BlackRock, Franklin Templeton, others); next tokenized equities / sports teams; South Korea ~Feb 4, 2027; Japan cited prior week — global capital race. Catalysts: Iran; SpaceX and Anthropic perpetual futures; 24/7/365 trading without futures rolls. Tokenization roadmap through 2030: portfolios, wallets, democratized ownership, IP trades.

  4. (39:49–42:18) Why he left hedge-fund path: entrepreneur, use every AI tool, understand crypto better than two years ago. Convergence now (not 3–6 months ago) — why charts look like that. AI trade stuck / vol exploded / run already happened; Micron may double next year but “a double’s not 10 times.” Policy wants markets higher and rates in check; yen coordination same week AI mud / Liupold; tape broke out; agents rising (Astra, Grokbot). Clarity vote → “immediate next leg” in his view. G20 / Jackson Hole. Paper: Age of Abundant Intelligence and Scarce Bitcoin — two presses; S&P multiple compression from AI terminal-value change + government debasement. Market not pricing Fed tightening cycle.

  5. (42:18–43:52) Until breakout (level spoken as “82” — asset ambiguous in ASR; treat carefully), no full chase; above it, chase factor. Fed meeting extreme importance: if market rises after a hike or after no hike, trend continues; midterm seasonality game. Substack → more behind subscriber paywall; monthly free thematic for loyal Substack. Papers: silent IPO over; Bitcoin, AI and the Collision of Time (first week of August); Handicapping the Future.

  6. (43:52–46:07) Handicapping close: father / horses — value not winner; no edge → sit out. BTC as 100-to-1 in that frame — small size can still pay. Annie Duke disciplined thinking; Munger pari-mutuel; PTJ great trades before consensus (his “third wave”); Stan: good idea + no positioning = opportunity. If ≥5% chance agents/tokenization drive crypto ecosystem, ~5% allocation is his opinion of a starting point; scale with probability; not advice. Out if lows taken until back above 200-day. Disclosures; visser-labs.com for RIAs; AI+crypto YouTube for subscribers end of month.


SYSTEMS MAP / VALUE CHAIN ANALYSIS

Mark inference vs source throughout.

Key players (Source, roles as assigned)

Node Role in his map Source locus
Official sector (Bessent, Japan MOF/BOJ, G20) Fight disorderly yield rise; yen coordination; run-it-hot equity preference 03:39–07:17, 40:34–42:18
Bond-crisis narrative cohort (X, technicians, Zero Hedge-style) Wrong thermometer; endgame bias overlap with AI-bubble / BTC-bear 02:50–06:26, 13:13–14:08
Credit market (IG junk tight; CCC widening) Systemic junk calm; AI disruption hits weakest credits 04:29–04:29
Inflation swaps / core CPI / Trueflation Anchor ~2.5%; argues against inflation-scare bond crisis 04:29–05:32
Anthropic, OpenAI, Nvidia Concentrated AI P&L core; model price ≫ rates 18:22–21:11
AI agents / Astra / Grok / browser agents Demand shock; AI organizations (teams ~1,200); native crypto users 21:11–29:12, 32:38–34:21
Crypto rails (BTC, ETH, SOL, stablecoins, wallets, L2s) Killer-app surface; BTC = ecosystem index / purest AI hedge 30:55–36:01
Tokenization / RWA / Robinhood L2 TradFi bridge already printing volume; RWA collateral → equity/IP 36:59–39:49
Pensions / insurance / Clarity Act Capital seal of approval; policy-reversal risk 33:29–34:21
Public AI equities (SMH, Micron, thematic 100) Stuck in mud / digestion; still long-term constructive but smaller weight than crypto 14:08–18:22, 40:34–41:28

Flow (Source, sequenced): Bond levels rise on global debt + growth → crisis narrative → he checks thermometers (vol, junk, inflation) → thermometers cold → governments coordinate to cap disorder → attention should shift to AI agents needing native digital money → stablecoins/tokenization/wallets → machine demand for crypto rails → while AI equity multiples/vol digest, crypto QTD and tokenization KPIs already move → size via handicapping (probability ≈ starting weight), invalidate on lows/200-day.

Bottlenecks, leverage points, pricing power

  1. False bottleneck (bond “crash”). Source: calm bond vol, tight junk, anchored inflation, 2022 vs 2026 total-return contrast, JGB scare vs Nikkei. Inference: media/technical focus on 30y level misprices crisis probability.

  2. True sovereign bottleneck (debt + growth repricing). Source: Brooks — debt levels, good growth, global, governments won’t allow crisis. Leverage: Bessent/Japan coordination, Clarity/G20 rhetoric. Inference: PMs fading all higher yields may be wrong on level but right that disorderly spike is fought.

  3. AI P&L bottleneck (model price, not rates). Source: −10% model price = 3 margin pts vs +200 bp rates = 1 pt; watch Anthropic/OpenAI/Nvidia. Inference: rate-spike equity fades that ignore model pricing are fighting the wrong risk factor.

  4. Next demand bottleneck (agentic payments). Source: Andreessen killer app; agents need stablecoins/tokenization; crypto lacked native user. Leverage: MoonPay/Grok, Stripe Link, Robinhood L2, RWA issuers, Clarity for pension capital. Inference: standards/plumbing race (US vs Korea/Japan) matters more than BTC maximalist vs ETH tribal fights for this thesis.

  5. Attention/positioning bottleneck (silent IPO). Source: ideologues reduced; many waiting for 60s/October; good idea + no positioning. Inference: under-ownership is part of the edge claim — crowdedness would change the handicap.

Upstream / downstream (Inference on Source)

Where constraints create investment hypotheses (not recommendations)

H1 (Source-led): Bond-crisis trades that require spiking bond vol / blowing out junk / un-anchoring inflation are fighting present thermometers. Research the level vs crisis distinction before sizing duration views off 30y headlines.

H2 (Source-led): If agents are the native crypto user, research stablecoin settlement, wallet/agent permissions, tokenized RWAs/equities, and L2 stock-token volume — not BTC-only SoV narratives.

H3 (Source-led): AI equity digestion can coexist with crypto outperformance (his QTD table). Relative-value research across SMH vs BTC/ETH/SOL is the coherent expression of his weightings claim.

H4 (Source-led): Model-price compression is the AI hard invalidation; rate +25–200 bp is not.

H5 (Source expression): Handicap sizing (p% probability → ~p% starting weight) is a process hypothesis for books with zero crypto — report, do not execute from this desk.


SECOND AND THIRD-ORDER EFFECTS

Each chain: [Primary] → [Second-order] → [Third-order] → [Investment relevance]. Links tagged.

1) Bond thermometers stay cold while levels high (rates, credit, equity risk premium)

2) Model price ≫ rates for AI P&L (Anthropic, OpenAI, Nvidia, semis)

3) Assistant → organization → financial agents (labor, software, payments, crypto)

4) AI mud + crypto QTD divergence (cross-asset allocation research)

5) Clarity / global tokenization race (policy, pensions, geopolitics of capital)


SCENARIO FRAMEWORK

He assigned an explicit handicapping metaphor (~5% / ~10% probability examples → matching starting weights) but did not publish a full bull/base/bear probability set for macro paths. Numeric scenario weights below are analyst inference for research planning only, except where his handicap language is quoted as Source.

Bull — “Trap confirmed; agents ignite rails”

Analyst inference probability: roughly 30%.

Base — “Governments contain yields; crypto works relative; AI digests”

Analyst inference probability: roughly 45%. Closest to his adjectives: bond fears overdone but yields can drift (even to 5.5%); AI stuck in mud / time needed; crypto already working QTD; add to crypto; handicap small starting weights for skeptics.

Bear — “Thermometers break and/or tape invalidation”

Analyst inference probability: roughly 25%.

Invalidation hierarchy for a PM dashboard (mixed Source/Inference):

  1. BTC (and by extension his crypto stance) takes out the lows — stay out until above 200-day — Source, hard.
  2. Bond vol + junk + inflation swaps all signal crisis together — Source thermometers, joint break = trap thesis fail (Inference on jointness).
  3. Model-price compression marking Anthropic/OpenAI — Source primary AI risk.
  4. Agentic payment/tokenization KPIs stall into 2027 while he claimed convergence is “now” — Inference timing risk on killer-app clock.

COMPANY / ASSET WATCHLIST

No ratings. No buy/sell from this desk. Positions/weightings are source expression.

Bitcoin (BTC)

Ethereum (ETH)

Solana (SOL)

Robinhood (HOOD) — named in 46-name crypto portfolio

Nvidia (NVDA)

Marvell (MRVL) / Micron (MU) / Dell (DELL)

Anthropic / OpenAI (private) / “Astra”

RWA / stablecoin–agent rails / long bonds / broad equities


DILIGENCE QUESTIONS & RESEARCH AGENDA

P1 — Must-do before any allocation-research decision

  1. Thermometer verification (P1). Rebuild: 30y yield ~5.23; bond vol (MOVE or his series) “down here”; HY OAS “tightest ever”; CCC vs BB/B divergence; inflation swaps 1y–10y ~2.48–2.52 vs core CPI 2.48; Trueflation path. Data: Bloomberg / BLS / Trueflation. If thermometers already hot, trap thesis is late.

  2. Tape verification (P1). BTC/ETH/SOL vs 200-day and vs “lows”; QTD +56/+38/+36 vs +3/−3/−14 (S&P/Nasdaq/SMH). Confirm whether “82” breakout level is BTC, ETH, or another index. Expert: house crypto quant.

  3. Robinhood $1.5B / L2 (P1). Confirm six-week tokenized stock-trade notional and L2 status. Without this print, tokenization-“now” claim weakens to roadmap rhetoric.

  4. RWA “18×” (P1). Define starting base, assets included (MMF-only?), and current TVL mix. Expert: digital-asset data vendor.

  5. Policy calendar (P1). Clarity Act status; Bessent–Japan yen coordination actions vs talk; Fed meeting he flags as extreme importance. External check needed on live status (memo date 8 Sep 2026; talk filmed around 6 Sep window).

P2 — Needed to underwrite the agent × crypto merge

  1. Model-price tape (P2). OpenAI/Anthropic/API price histories; margin sensitivity vs his 10%→3pt and 200bp→1pt rules of thumb. Expert: AI infra strategist.

  2. Astra / AGI % claim (P2). What model is “Astra”? What benchmark stack maps to “40% → 100% knowledge-worker” ? Do not underwrite AGI from ASR alone.

  3. Agent-payment KPIs (P2). Grok–MoonPay volumes; Stripe Link agentic purchase stats; browser-agent commercial deployment. Private/partial — triangulate.

  4. 46-name crypto sleeve (P2). When subscriber video drops later this month, refresh watchlist; until then treat 6 stock / 40 token split as process color only.

  5. Andreessen primary (P2). Re-listen Latent Space ~Apr 4 segment 50:00–55:00 for killer-app / payments wording fidelity vs his paraphrase.

P3 — Process and identity cleanup

  1. ASR identities (P3): Liupold/Liupole, Moonay→MoonPay, Tro in RWA list, “82” level, Open Face episode exact title. Audio-verify before LP-facing quotes.

  2. Handicap discipline (P3): If using his probability↔weight frame as research process, document explicit probability, invalidation (lows/200-day), and that desk recommendations are separate from his opinion.

  3. Companion memo hygiene (P3): Do not merge this talk’s QTD figures or 5.23% 30y with Alma/Bessent/SV/Cboe memos without attribution — different as-of stamps.


RISK ANALYSIS

Thesis risks (the framework is wrong)

Timing risks

Execution / data risks

External / regime risks

What would change the research view (actionable)

  1. Joint spike in bond vol + HY OAS + inflation swaps — reopen bond-crisis workstream.
  2. BTC losing lows — pause nexus/agent-ignition sizing research per his rule until 200-day reclaim.
  3. Documented >10% frontier model-price cuts with margin damage — AI circular-finance risk elevates over rates.
  4. Tokenized-volume / agent-payment KPIs failing to print by early 2027 — killer-app clock slips; prefer SoV-only research framing over rails.
  5. Clarity failure with simultaneous decline in global (Korea/Japan) tokenization progress — capital-formation path impaired on multiple tracks.

APPENDIX: SOURCE DISCIPLINE LOG


Prepared Tuesday, 8 September 2026 (America/Toronto) for internal PM-research use. Educational summary of a public talk; not investment advice. Cue timestamps from retrieved ASR timedtext.

Desk copy · not a trade recommendation · Erica · 8 Sep 2026 · HIGH batch