In this week’s video, I break down why we are now in the midst of what Elon Musk called the “supersonic tsunami”, and why the acceleration phase of AI is rewriting market structure in real time. Over the past week, the disruption spread beyond SaaS into insurance brokers, wealth management platforms, commercial real estate services, and trucking stocks. 115 S&P 500 names fell at least 7% over a rolling 8-day window, many near 52-week highs, a dispersion pattern we haven’t seen since the dot-com rotation in 2000.
The critical development: recursive self-improvement is no longer theoretical. OpenAI’s GPT 5.3 Codex helped build itself, and Claude Opus 4.6 shipped with a 5x context window expansion over 4.5 just weeks prior. Meanwhile, free Chinese models like MiniMax M2.5 are benchmarking near Opus 4.6 at 1/20th the cost. The model velocity war is accelerating the deflationary spiral in anything built on code, while the hyperscalers face a trap, rising RPOs they can’t fulfill due to physical bottlenecks, growing capex guidance, and competitive erosion from open-source alternatives.
My turbulence model has fired 12–15 signals in five weeks versus 20–25 in the prior 28 months, and it’s happening while hedge fund gross leverage sits near all-time highs. Credit is showing signs of weakening, leveraged loans have broken below the 100-day moving average, HY-to-IEF is trending lower, and BDC indices remain depressed. The covariance matrix is under stress across assets, not just equities. I believe the probability of a meaningful unwind event is closer to 25% while the market positioning suggest much lower.
The positioning: be long scarcity, short abundance. That means energy, materials, chemicals, small caps, foreign stocks (particularly Brazil), and Bitcoin once software stabilizes. Names like Eaton and Chevron are breaking out of multi-year bases as PMIs just posted their biggest upswing outside of COVID. The IWM-vs-QQQ trade remains the key expression. Software isn’t dead, it’s worse: the cost of building a $20–100M ARR SaaS product has fallen below $10,000 in compute, creating a supply explosion that compresses margins for incumbents. The hyperscalers relative to the S&P are approaching their lowest levels since 2023. This is the creative destruction cycle Joseph Schumpeter described and it’s moving at supersonic speed.
Timestamps
- (00:00–03:20) Supersonic tsunami thesis: AI has reached the acceleration phase with recursive self-improvement now confirmed. Paywall launch, the framework for why software is a value trap, and why crypto’s utility function will grow as AI agents become the consumers.
- (03:38–06:30) Speed and structural power: Why this isn’t a typical rotation as AI is cannibalizing its own winners. The disruption spread to insurance brokers ($250B market cap hit from a $1.5B startup), wealth management, commercial real estate, and trucking in a single week.
- (06:30–09:00) Dot-com parallel: 115 S&P names fell 7%+ in 8 days near all-time highs. Turbulence model and covariance matrix stress. Hedge fund gross leverage at extremes, with pod shop de-risking at 100th percentile.
- (09:00–12:25) Leverage and systemic risk: Why gross leverage matters more than net, how the IWM-vs-QQQ trade is critical to the risk framework, and why the rotation from tech into small caps/materials creates a size mismatch problem.
- (13:06–14:34) SaaS as a value trap: Software is now the 5th most expensive sector (was 1st). The “buying cheap software” narrative ignores the deflationary pressure of exponential AI progress where coding is effectively free.
- (14:34–18:26) Recursive self-improvement: Opus 4.5 to 4.6 in five weeks, GPT 5.3 Codex building itself, Eric Schmidt’s warnings from July confirmed. The inflection point from AI assisting humans to AI generating infrastructure for AI.
- (18:43–21:58) Model velocity war: MiniMax M2.5 matches Opus 4.6 at 1/20th the cost. Chinese models going free. Mac Minis sold out. Open-source agents turning hardware into digital labor. The K-shaped economy drives entrepreneurs to free models.
- (23:16–25:26) Jaws of disruption: No moats survive, companies gapping down from 52-week highs signals momentum breakdown. Crowding + AI disruption = a new factor that doesn’t exist in quant models.
- (26:08–29:16) Macro warning signs: 10-year rates falling despite 7% nominal GDP. Payrolls driven by education/health. Wage pressure declining. True core inflation at 1.17% vs. reported 2.6%. The S&P can’t break out when its largest components are declining.
- (29:16–34:45) The trade: Long scarcity (IWM, energy, materials, chemicals), short abundance (QQQ, software, hyperscalers). SaaS supply explosion means 10x more competitors chasing same customers. Andre Karpathy’s “vibe coding” as a death certificate for software’s scarcity premium.
- (35:01–41:13) Hyperscaler trap: RPOs at $1.2T and growing but capacity can’t keep up. Capex guidance rising. Chinese models eroding market share. ChatGPT share fell from 70% to 45%. Microsoft relative to S&P back to 2020 levels. Elon Musk: “Software guys are becoming hardware guys.”
- (41:35–44:57) Physical bottlenecks: Transformers, substations, copper, silver, rare earths, optical fiber, HBM, any one can delay revenue. Data center cancellations growing. Dario Amodei’s overbuilding-vs-underbuilding trap.
- (46:40–53:47) Risk framework: Turbulence model, credit contagion (leveraged loans, HY/IEF, BDC weakness), financials joining the selloff. Leadership transitioning to unhealthy mix of staples/utilities/energy. Growing probability of a deleveraging event this year.
- (53:47–59:22) Positioning: Long MSCI World ex-US (flat from 2007–2024, now accelerating). Long Brazil (EWZ) with falling Selic rate + rising CRB industrials. Chevron and Eaton breaking out of multi-year bases. Chemicals as the next theme (Corning/Meta deal). PMIs posting biggest upswing since pre-COVID.
- (59:22–1:04:40) Bitcoin: Correlated with software decline but showing lower beta on the downside. If software waterfalls further, BTC could test $40K, but Fed rate cuts and dollar weakening would set up the next leg higher. Dario Amodei’s pessimistic piece and Anthropic safety researcher quitting signal growing existential awareness. The creative destruction cycle demands scarcity positioning.