AI, Hormuz and Credit: Echoes of Bubbles

AI, Hormuz and Credit: Echoes of Bubbles

Earnings are driving Wall Street higher, but tech concentration, energy and private credit are reopening old vulnerabilities.

There is a paradox defining this market phase: the world’s most important energy chokepoint remains effectively blocked, Brent is still hovering near $100, and yet Wall Street continues to hit record highs. Reuters reported that the Strait of Hormuz has been “effectively shut” since the beginning of the conflict in late February, while the IEA has described energy markets as heading into “troubled waters,” with millions of barrels per day removed from global flows.

The reason behind the rally is no mystery: earnings. According to FactSet, as of May 1, 63% of S&P 500 companies had already reported first-quarter results; 84% had beaten EPS estimates, aggregate earnings growth stood at 27.1% year over year, and the leading sectors were Communication Services, Information Technology and Consumer Discretionary. Even more importantly, the “Magnificent 7” were seeing earnings growth of 61%, compared with expectations of 22.4% at the end of March.

This is what distinguishes 2026 from 1999: today the market is not buying only dreams, but real profits, high margins and the cash flows of megacap companies. Yet this is precisely where the risk begins. In 1999, the story was the internet; today, it is artificial intelligence. In 2007, the fragility was hidden in subprime mortgages and securitisations; today, a growing share of risk is concentrated in private credit, which is less transparent, less liquid and increasingly intertwined with banks, insurers, funds and retail investors.

The technical comparison is less reassuring than it might seem. The S&P 500 is trading at 20.9 times expected earnings over the next twelve months, above its five-year average of 19.9 and its ten-year average of 18.9; its trailing P/E is 28.5. We are not at the extreme multiples reached by the Nasdaq during the dot-com bubble, but Shiller’s CAPE ratio has returned to a historically elevated zone: Multpl places it at 41.69 as of May 7, compared with an all-time high of 44.19 in December 1999.

The strongest resemblance to 1999, therefore, is not the S&P 500’s forward P/E, but the combination of three factors: a dominant technology narrative, concentrated returns and very stretched long-term growth expectations. Reuters reports that on May 6 the S&P 500 and Nasdaq closed at new records, with the PHLX Semiconductor Index up 62% year to date and AMD jumping almost 19% after issuing guidance above expectations.

In short, the market is making a very specific bet: that AI will turn enormous capex into recurring revenues and durable margins. That is possible. But the “1999” signal is that investors today are paying not only for innovation, but for its almost perfect monetisation. Goldman Sachs has recalled that between 1995 and 2000 the Nasdaq quintupled and reached P/E valuations of around 200 times; today we are far from that extreme, but the acceleration dynamic around the dominant theme is similar.

The second signal is concentration. When a handful of stocks account for a disproportionate share of market gains, the market becomes more fragile: all it takes is a shift in sentiment toward semiconductors, hyperscalers or data centres to turn an orderly rally into a violent repricing. FactSet shows that Alphabet, Amazon and Meta represented 71% of the net increase in S&P 500 earnings in the week under review; in the same report, Nvidia and Micron were the main contributors to growth in the Information Technology sector, which would fall from 50% to 27.6% if they were excluded.

The third signal, closer to 2007 than to 1999, is credit. The Financial Stability Board estimates private credit at between $1.5 trillion and $2 trillion at the end of 2024, with growing interconnections among asset managers, banks, insurers and private equity. Direct bank credit lines to private credit funds amount to around $220 billion in data collected by FSB members, while commercial estimates range from $270 billion to $500 billion.

Here, the analogy with subprime should not be overstretched, but nor should it be ignored. In 2007, the problem was not only that borrowers were deteriorating: it was that risk had been transformed, layered, sold and often poorly understood. IOSCO found that delinquency rates on subprime ARM mortgages rose from less than 4% to more than 10% by September 2007, while CDOs and structured instruments ended up concentrating risk rather than dispersing it.

In today’s private credit market, we see rhymes, not replicas. The FSB points to borrowers with lower credit quality and higher leverage than comparable public markets; according to some estimates, leverage among private credit borrowers is 5-6 times debt/Ebitda, but can approach 7 times once Ebitda adjustments are stripped out. In addition, the use of PIK instruments — interest paid by increasing the debt rather than in cash — affects around 12% of loans and has grown since 2022.

The deterioration is no longer merely theoretical. Fitch, according to Reuters, recorded a record default rate of 9.2% in 2025 among US corporate private credit borrowers in its monitor, up from 8.1% in 2024; the sample mainly concerns middle-market companies, often with Ebitda below $100 million and debt below $500 million.

Systemic risk today does not necessarily run through an investment bank overexposed to CDOs, as it did in 2007. It runs through a more diffuse ecosystem: semi-liquid funds with redemption windows, insurers searching for yield, retail BDCs, NAVs valued quarterly, banks providing credit lines and private equity sponsors that can decide whether or not to support already highly indebted companies. The FSB explicitly refers to valuation opacity, sector concentration in technology, healthcare and services, the growing use of private ratings and data gaps that make it difficult to monitor risk at the system-wide level.

The Hormuz factor adds a macro dimension that was almost absent in 1999. The EIA notes that in 2024 around 20 million barrels per day passed through the Strait, equal to about 20% of global petroleum liquids consumption, as well as roughly one-fifth of global LNG trade. The IEA estimates for 2025 almost 20 million barrels per day of oil flows and only 3.5-5.5 million barrels per day of potential alternative capacity.

This means that the equity market is simultaneously pricing in an AI earnings boom and geopolitical normalisation. If the Strait reopened quickly, the rally would find a macro justification: lower energy prices, lower expected inflation and a Fed freer to cut rates. But if the closure were prolonged, the risk would be the opposite: compressed margins outside tech, higher real rates, pressured consumption and a more fragile private credit market. The IEA has already recorded an 85 million-barrel drawdown in observed global inventories in March, with 205 million barrels drained outside the Middle Eastern Gulf as flows through Hormuz were choked off.

The conclusion is that the market is not “1999” in a strict sense: today’s big tech companies generate real earnings, cash and competitive advantages. Nor is it “2007” in classic form: private credit does not mechanically replicate subprime mortgages, and direct bank exposure appears more contained. But the analogous signals are there: elevated valuations, narrow leadership, an all-consuming technology narrative, opaque credit, hidden leverage, illiquid instruments sold as stable-yield products and a possible exogenous shock — energy — capable of rapidly changing the cost of money.

The real question is not whether we are already in a bubble. It is whether the market is pricing in a world in which everything has to go right at the same time: AI monetises, Hormuz reopens, inflation falls, the Fed cuts, defaults remain contained and margins do not normalise. History teaches that markets can remain euphoric for longer than sceptics imagine. But it also teaches that when valuations, leverage and opacity add up, the problem does not begin with the first stock that falls: it begins with the first asset no one can price anymore.