The Next Black Swan

The largest technology companies are on track to spend somewhere north of $600 billion on AI infrastructure in 2026 — data centers, chips, and power. What changed recently is how it is funded. For a decade these were cash-rich businesses that built from internal profits. Now they are issuing debt at a scale not seen since the telecom build-out of 2001, with analysts at Morgan Stanley and J.P. Morgan projecting the sector may need to raise roughly $1.5 trillion in new borrowing over the coming years.

Three things make this fragile. The assets are short-lived — GPUs depreciate far faster than the buildings that once dominated corporate capital spending, so the accounting hit arrives quickly. The revenue is concentrated — much of the projected demand rests on a handful of AI labs that are still burning cash. And the financing is increasingly circular, with chipmakers, cloud providers, and model developers investing in one another. If the returns on this spending disappoint, the spending stops abruptly, and the earnings estimates that justify today's valuations have to be rewritten.

This is the re-rating risk we covered in "Clearly Cisco." The concern is not fraud; it is that a great technology can still be attached to a stock price that has run too far ahead of it. Because these same names dominate the index, an AI capex reversal would not stay contained in the technology sector.

This is the candidate most retirees already own without realizing it. The S&P 500 is no longer diversified in the way the word implies. Seven companies make up roughly 35% of its value, the top ten roughly 38%, and that small group generates the large majority of the index's economic profit. A retiree who believes "I'm in the index, so I'm diversified" may in fact be running a concentrated bet on a single theme.

The mechanism is reflexive. Passive index funds buy the largest companies mechanically, in proportion to their size, which pushes those companies higher, which draws still more money into them. That flywheel works beautifully on the way up. It runs in reverse on the way down, with forced selling concentrated in exactly the names that have the furthest to fall. Of every risk on this list, this is the one most within your control — and the one most worth examining before any shock arrives, not after.

The most dangerous concentration is the one you didn't choose and don't think you have.

Federal debt now stands near $39 trillion, and the annual interest cost has passed $1 trillion — more than the United States spends on national defense. The weighted-average interest rate on Treasury debt is rising as older, low-coupon bonds mature and are refinanced at today's higher yields, which means the interest bill keeps climbing even if rates simply hold steady. Several recent auctions have drawn soft demand, raising a question that used to be unthinkable: who is the marginal buyer of all this debt?

The Treasury market is the foundation the entire system is built on. It prices mortgages, it anchors corporate borrowing, and it is the asset retirees hold precisely for safety. If confidence in that demand cracks — through a failed auction, a downgrade spiral, or a prolonged fiscal standoff — yields spike and bond prices fall. The painful irony is that the holding many retirees own for stability becomes a source of loss. The Congressional Budget Office and the Committee for a Responsible Federal Budget have both flagged this trajectory as unsustainable on current policy.

Private credit — lending that happens through investment funds rather than banks — has grown into a market measured in the trillions, much of it built up over a period of easy money and limited regulation. The loans are not marked to market daily, so reported values can be stale and, at the margin, subjective. Early in 2026 the sector met its first serious liquidity test: a flagship non-traded fund received redemption requests well beyond its quarterly cap, forcing it to gate withdrawals.

The IMF and the Federal Reserve have repeatedly flagged a specific design flaw. Semi-liquid funds that let investors exit quarterly are holding loans that cannot be sold quickly, which creates a "first-mover advantage" — those who run for the exit first get paid out at stale valuations, leaving everyone who stays to absorb the eventual markdowns. Official assessments still judge the sector as not yet systemic. But opacity, leverage layered on leverage, and unclear connections between players are precisely the conditions under which a localized problem becomes a contagion that no one mapped in advance.

The dollar's status as the world's reserve currency is what lets the United States borrow as cheaply as it does. That privilege is not a law of nature. Jim Rogers has argued for years that the steady accumulation of debt and the long habit of money creation eventually force a reckoning in the currency itself — and that the warning signs tend to be ignored until they are not. Central banks have been buying gold at a notable pace, and gold's long advance is, in part, a market vote on monetary trust.

There is also a plumbing risk inside this theme. The unwind of the yen "carry trade" in the summer of 2024 produced a violent, if brief, global tremor — a reminder of how much leverage rests on a few stable currency relationships. If the dollar's advantage erodes faster than markets expect, the cost of nearly everything priced in dollars resets at once. These are macro and geopolitical claims, attributed to the voices making them; they are not a forecast from this firm.

The classic black swans of the modern era have been geopolitical. Today the most consequential pressure points include the semiconductor chokepoint around Taiwan, recurring conflict in the Middle East — which some strategists have already labeled a "gray swan" following military operations involving Iran — and the energy supply shocks that feed straight into inflation. A genuine escalation around Taiwan would be especially severe, because it would strike the exact AI-and-semiconductor complex that now dominates the index, layering a geopolitical shock on top of a concentration problem.

This firm does not make political predictions, and these scenarios are described as the risks that analysts and policymakers are tracking, not as outcomes we expect. The fiduciary point is narrower: a portfolio should be built so that it does not depend on any one of these tensions resolving peacefully.

This is the candidate that comes nearest to Taleb's original meaning, because its origin is not economic and it is genuinely hard to model. The machinery that markets depend on — clearing, settlement, payment rails, and the concentrated cloud infrastructure that now runs much of it — represents a single point of failure that no balance-sheet analysis captures. We explored this in "The Financial Outage."

A coordinated cyberattack, or a cascading outage at one of the handful of cloud providers that the entire economy leans on, could freeze transactions in ways that have nothing to do with valuations or earnings. The unsettling part is the overlap: the same small group of companies that dominate the stock index also operate much of this critical infrastructure. Concentration of ownership and concentration of operational risk have quietly become the same problem.