Back Research Notes The Fragile Immune System: AI, Concentration, and VaR Shocks Published on October 14, 2025 By Jordi Visser In a concentrated economy where wealth, profits, and innovation are clustered in a few dominant sectors and companies with low debt, the old playbook of credit-driven recessions no longer applies. With governments and central banks saddled with debt and deficits conditioned by decades of crises to intervene at the first sign of contagion, downturns are delayed or stopped rather than allowed to run their natural course. Yet as with all systems in nature, control has limits. As the naturalist John Muir once said, “When we try to pick out anything by itself, we find it hitched to everything else in the Universe.” In markets, this interconnectedness means that suppressing volatility in one area often amplifies fragility somewhere else. The more concentrated the system, the more it resembles a body that seems healthy on the surface but whose immune system has quietly weakened from lack of exposure. That concentration of strength across markets, wealth, and corporate power creates a paradox. On one side, we see bubbles in innovation, productivity, and valuations; on the other, stagnation and silent recessions for the majority of people and sectors. This duality mirrors biological imbalance: an organism that overdevelops certain cells at the expense of others becomes vulnerable to systemic failure. When such an economy finally “gets sick,” corrections can be extreme rather than mild just like an immune system that, after years of suppression, overreacts when finally tested. That is the risk today: a system that has substituted resilience for constant rescue, one that has traded adaptation for protection, and one that may face its most severe illness not from neglect, but from overcare. Right now, as I highlighted in my recent video , I feel like the risk of the market getting sick for a few weeks has risen. No, this is not me believing AI is in a bubble. However, I think it important to share a few thoughts on both the positive and negative sides of AI and how it will impact the market structure as the “bubble” grows. In my opinion, there is no denying that AI will replace jobs and anxiety for workers will only worsen as time passes. It’s not speculation, it’s already showing in economic data and it will accelerate as agents, autonomous vehicles and humanoids grow. Like every major innovation in history, it will eventually create new forms of work and opportunity, but this time it’s different because AI isn’t about helping physical labor or software to help efficiency through human interaction, it’s about intelligence, creativity, and cognition. It’s about solving problems that humans alone cannot. AI has already conquered chess and Go, outperformed students on standardized tests, and now even beats humans in math Olympiads. To dismiss it as hype or a bubble is to misunderstand its inevitability. As intelligence becomes computable, its scale will follow the same exponential path as every other digital transformation. There will be extraordinary fortunes made along the way for companies positioned in the buildout, the infrastructure, the chips, the data centers, the software layers and investors who recognize this will have a generational opportunity. However, while the total revenue pool for AI will eventually justify the capital spending, I don’t believe the current market leaders will all earn enough to balance out their investments. Yes, I do not think the capex will end up being a good decision for all the spenders. I believe the AI revenues will dwarf any estimates out there but I do not believe they all will proven to be smart decisions. For many or even most, the returns on spending will compress as competition from startups and private firms expands. In the era ahead, “price-to-sales” will become what “debt-to-equity” once was, the measure of financial or at least stock multiple vulnerability. A company without debt but with an extreme price-to-sales ratio is still fragile because it’s reliant on revenue growth that may not materialize. As innovation diffuses, margins shrink. As someone over 55 who fully embraces AI as the greatest personal and societal innovation of our lifetimes but also knows there are no free lunches, I see both the promise and the peril: the promise of abundance and problem-solving, and the peril of a market that mistakes exponential change for infinite growth. All of this is to say that as AI accelerates, the economic and market immune system, the equivalent of a diversified microbiome begins to weaken. In a world where intelligence, productivity, and capital are concentrated, the system becomes more efficient but also more brittle. Instead of traditional recessions, what we are likely to experience are VaR shocks —sudden and violent repricings of risk. Hedge fund gross leverage is near all-time highs, which means the system is collectively betting on stability. But every form of leverage is ultimately a bet on something: a relationship, a pattern, a correlation that continues to hold. When those relationships change abruptly, the immune system gets tested and today’s system has been engineered to avoid even mild fevers. In credit markets, the danger is clearest. Private credit has exploded in a world where the broader economy remains uneven and heavily supported. These structures depend on a slow, predictable credit cycle with defaults rising gradually, liquidity absorbing shocks. But if weakness emerges suddenly, especially through fraud or structural cracks, the adjustment will not look like a typical credit downturn; it will look like a VaR event. Similarly, in market-neutral strategies, leverage depends on the covariance matrix on stable relationships between volatility, correlation, and liquidity. AI itself threatens that stability by changing how capital moves, how liquidity clusters, and how correlations form. As machine learning systems identify the same signals and react in the same milliseconds, the market’s diversity of behavior, its microbiome, shrinks. The immune system, once adaptive and resilient, now shows the early signs of exhaustion. When the next shock comes, it may not be deep because of fundamentals, but because the body has forgotten how to fight. In my latest video, I highlighted a cluster of “immune-system” warnings that, taken together, raise the odds of a VaR shock into month-end. Odds rising does not mean it will happen, but I think the risk of it is rising by the day. Crowded hedge-fund longs have ripped relative to the most crowded shorts; equity-specific volatility is being bid (seen in VIXEQ) even as index volatility (the VIX) lags; credit spreads are widening post First Brands and Tricolor and business development companies (BDCs) and publicly traded private equity firms are showing early signs of contagion risk. Overlay this with the AI-driven market concentration and the fresh escalation in U.S.–China trade tensions, especially around AI dependent rare earths, and the probability of a deal by the assumed Trump–Xi meeting at APEC looks increasingly less certain. Despite this, betting markets still price an 80% chance of resolution by November 10th, which means any deviation from that assumption could spark further repricing as uncertainty builds. My gut tells me these odds and people’s lack of worry over this dispute is a representation of the TACO PTSD that exists. Only recently do I think the last of the tariff fear holdouts capitulated. Nobody wants to deal with another bout of this. Positioning only amplifies this fragility. We’re entering earnings season with buyback windows closed, vol-control and CTA models extended long, and retail momentum still driving flows. These are mechanical exposures, algorithms buying strength and selling weakness, creating a system dominated by reflex rather than discretion. If credit cracks suddenly or AI-driven correlations inver