Jordi Visser — Bubbles, Parabolas & Speed Crashes Investment Thesis

Source: Bubbles, Parabolas and Speed Crashes: The End of Human Market Structure, Jordi Visser (YouTube), May 2026.

The Framework: From Kindleberger Humans to Agentic Flow

Visser argues market microstructure has broken the “mania–panic–crash” playbook tied to emotional, slow-moving human participants. AI agents trading and routing capital supposedly optimize for P&L without nostalgia—shrinking behavioral half-lives while amplifying parabolic rallies and “speed crashes” resolved in months, not seven-year unwind analogies.

Lens	Legacy (human-centric)	Visser’s agentic-era read
Information	Lags masked supply–demand dislocations	Instant digital visibility (Strait-of-Hormuz “video-game” oil analogy)
Shocks	Broad manias fed by euphoria + debt	Bottleneck + inflation shocks priced fast; drawdowns fast
Style	FOMO / Druckenmiller-type regret loops	Dispassionate accumulation; passive benchmark drag resolves via forced catch-up flows

Investment Thesis #1: “Bubble” Critics Are Fighting the Wrong Book — Facts Read Like Earnings Expansion, Not 1999

The argument Visser contrasts today’s cohort with dot-com Cisco comps: cites ~$27.1% S&P YoY EPS growth (post–320 reporter sample ~20% avg surprise), ~88% YoY March semiconductor sales, ~$1.3–1.4T cloud backlog prints, aggregate PEG ~1.03 (via Ed Yardeni reference), Nvidia forward multiple near decade lows versus Cisco ~130× peaks, and consumer confidence nowhere near euphoria.

"How in God's name, this is an unbelievable move in year-over-year earnings growth, 27.1%. … every single one of them except for energy."

Contrarian element Bears anchor charts resembling 2000; bulls ride beats + revisions breadth (median-stock / smaller-name participation beyond megacaps, per curated slides referenced).

Trigger Subsequent quarter maintains upside surprise dispersion without growth estimate slashes—invalidates Kindleberger-style payoff bet.

Names Broad semiconductor/memory (Nvidia NVDA, memory leaders), capital goods lifting AI factories (Caterpillar CAT thematic via PMI channel).

Investment Thesis #2: Benchmark Arbitrage — Passive Isn’t Hedged Against the Concentrated AI Winners

The argument Cap-weight inertia (MSCI/pension footprints) mechanically undershoots the AI capex/leverage beneficiaries still repricing fundamentals; incremental $8T YTD-ish compute+model layer gains alongside $1.2T bleed from GDP-heavy service SaaS proxies embodies the pairs trade forcing allocators toward equipment + models.

"Year-to-date. The compute and model layers have added 8 trillion in combined market cap while the service apps that make up two-thirds of GDP today have lost 1.2. This is the benchmark arbitrage, guys."

Contrarian element Gamma + retail narratives underestimate indexed-fund structural mismatch chasing the same narrowing leadership.

Trigger Quarterly fund flow filings showing systemic Megacap 7/AI-factor overweight convergence vs benchmarks.

Names Nvidia, Taiwan Semiconductor (TSM), hyperscaler primes (Amazon AMZN, Alphabet GOOG, Microsoft MSFT) as distribution + compute oligopoly.

Investment Thesis #3: Agentic Inflection (Nov/Jan) — Token Demand Stair-Steps (15–50×), Bottleneck Migrates GPU → Electron

The argument Coding agents (Anthropic toolchain referenced) trigger digital employee explosions; compares $90T Jensen “physical upgrade cycle” meme. Tokens become consumable commodities tethered to electrons.

"Tokens are the food of these digital employees. … Tokens are power plus chips. End of story."

Contrarian element IQ-era focused only HBM silicon; agency era prioritizes power budget + grid hook + cooling BEFORE incremental flash DRAM bursts.

Trigger hyperscaler disclosures of PUE/power purchase escalation faster than silicon wafer adds.

Names Constellation Energy (CEG) / GE Vernova (GEV); Vertiv (VRT); Eaton (ETN); grid-scale battery storage thematic (named examples in narration).

Investment Thesis #4: Late Cycle = Bottleneckflation — Rotate Off Exhaustion DRAM Beta Into Lag Power + Monetary Alternatives

The argument Visser trimmed ~two-thirds of his memory-chip position—not due to DRAM demand collapse but technical exhaustion overlays versus later-cycle power / silver / crypto tails; aligns with flattening CPI path toward ~3.7% risking three-month bills < CPI (negative carry vs inflation parallels post-2022 regime where bitcoin rallied).

"DRAM is the ninth most traded ETF. It traded over 1 billion each in the past two days. Absolutely unheard of. This is … the reasons why I'm reducing my … micron significantly."

Contrarian element Memory tightness persists; tactical risk-reward rotates toward neglected bottlenecks (grid, turbines, electrons, metals).

Trigger DRAM spot/ETF volume spikes coincide with RSI clusters on subscriber exhaustion sheet.

Names Adds silver (SLV), Bitcoin (BTC) (plus Ether narration—no discrete ADR emphasized), Sterling Infrastructure (STRL) thematic (engineering builds), utilities (Vistra cited personally—consider CEG/IPPs if liquidity needed).

Investment Thesis #5: Macro Tape Confirms PMI → Logistics Revival — Freight, Copper, Foods Set Up Hawkish Surprise

The argument Cites March Logistics Managers Index ~69.9 (rapid transportation capacity tightening), aligns John Roque “frozen rope” lists clusters (power copper grid utilities NG transportation cap goods agriculture), warns breadth divergence (47 stocks within ~2% of highs albeit narrow concentration), fertilizer feed-through → grocer CPI base effects.

Contrarian element Equity indices print records simultaneous with historic negative breadth streak analogues—investors must differentiate bubble price from bubble fundamentals.

Trigger Subsequent employment mix persists (trade/transport job adds vs finance/information softness) → CPI stickiness pushes Fed jawboning hawkish.

Names Freeport-McMoRan (FCX) copper linkage; select upstream oil services thematic (shortage storyline); ** packaged agricultural inputs** (Albemarle ALB as materials basket proxy—prefer existing watchlist: FCX, CAT infra).

Investment Thesis #6: AI Liquidity Morphology — Faster Up/Faster Down Speed Crashes Define Next Decade

The argument Accept parabolic leadership + crash risk concurrently—difference versus historical cycles lies in recovery half-life (months) because policy + telemetry compress shock discovery.

"when AI agents dominate market forces … It's a speed crash. We go faster up, we go faster down, and it stays that way for the next decade."

Contrarian element Long-vol discretionary sellers misprice crash frequency × shallow duration.

Trigger Repeated VVIX/MOVE divergence vs equity highs while median stock RSI remains subdued.

Names Narrative-heavy bitcoin-beta equities (speaker discusses BTC ETF flows explicitly); thesis table sticks to SLV/BTC.

The Ecosystem Map (What Visser Is Watching)

Flows Passive benchmark → active catch-up (MSCI-heavy US watchers).
Fundamental guardrails Anthropic scale-up anecdote (order-of-magnitude FY growth commentary in video); hyperscaler interconnect revenue marks.
Positioning anecdotes Rotation sell memory / buy GPU + electrons + metals + crypto.
Tools Subscriber technical / peg / exhaustion fusion sheets (“140 IQ spreadsheet merge”) to rank themes not cross-sector raw PEG apples-to-oranges.
Conference Scheduled South Florida IA May 13–15 tokenisation talk (July rollout rumor).

Key Risks

Breadth extremes Repeated narrow rally parallels 1929 analogue datapoint cites—tail correlation surge if passive outflows collide.
Markup accounting loops hyperscaler VC marks (Google↔Anthropic etc.) can inflate optics intra-quarter (FT-style dependency article flagged).
Consumer durable collapse PCs/phones/auto/appliance anecdotes worst-than-GFC demand—GDP consumer share bleed unrelated to headline index prints.
Geopolitic oil choke Hormuz disruptions inflate goods inflation tightening financial conditions.
Policy error CPI reacceleration paired with Fed leadership change hawkish optics even if hikes off table.

Investment Opportunities at a Glance

Tier	Name / Category	Core Thesis	Conviction Signal
1	Nvidia (NVDA)	Still core compute leverage; multiples compress amid triple-digit YoY fundamentals narrative	Guidance + supply chain beats vs DRAM proxy
1	Constellation Energy (CEG) / GE Vernova (GEV)	Agentic bottleneck = electrons first; uranium/gas turbines monetize scarcity	Capacity retirements vs AI PPA backlog
2	Eaton (ETN)	Electrification + switchgear bottleneck stack	Orders inflection corroborating PMI logistics
2	Vertiv (VRT)	Liquid cooling scaling with multi‑MW racks	Thermal attach rate disclosures
2	Freeport-McMoRan (FCX)	Copper + silver beta from grid + inflation hedge rotation	Copper treatment charge / China stimulus
3	SLV / BTC	Negative bills vs CPI + institutional ETF share growth tail	ETF AUM ATH repeats / real-yield rollover
3	Caterpillar (CAT)	Late-cycle PMI cap goods uplift + power equipment channel	PMI transport & logistics follow-through

Monitoring Checklist

S&P earnings surprise breadth rolling above ~15% — Sustains “not Kindleberger earnings” pillar.
Semi billings YoY trajectory (~88% March anecdote)—Tests memory/power sequencing.
Anthropic/open-weight demand vs hyperscaler offload headlines — Validates $1.4T backlog glide path.
Logistics PMI & flatbed indices persistence > expansion thresholds — Freight bottleneck thesis stays live.
Exhaustion screen on memory names flashing — Timing tactical trims vs structural tightness (per Visser workflow).
Breadth divergence stats narrowing — Would signal crash velocity risk tempering marginal AI beta adds.

Bottom Line

Agentic workflows + indexed capital misalignment allegedly convert “bubble optics” into serial earnings beats — trade the fundamental tape, not 1999 anecdotes.
Token throughput economics collapse to watts + silicon → electron layer (utilities, turbines, transformers, cooling, metals) absorbs marginal capital after memory late-cycle exhaustion.
Benchmark arbitrage frames $8T add vs $1.2T SaaS bleed as capital rotation, not hallucination—passive reallocations lagging still create forced-flow fuel.
Inflation resets + negative short bill carry underpin silver/bitcoin overweight alongside industrial electron themes—risk assets reposition from pure beta chip lag.
Speed crashes substitute for multi-year panics: risk process must prioritize months-scale drawdown drills versus seven-year doom loops.