Significant shifts in financial markets happen with great frequency, but one recent move deserves everyone’s attention. Major UK pension funds, overseeing more than £200 billion for millions of workers, are looking into other sectors more actively, away from US equities. Their concern? The AI-fuelled tech rally may be far more fragile than it appears.

It’s a structural warning that echoes a familiar historical pattern: moments when entire booms concentrate value, expectations and capital into a narrow slice of the economy.

Yet when these long-horizon funds signalled they were scaling back US equities, markets barely blinked. The S&P 500 still hovered near record highs, AI stocks kept climbing, and analysts waved it off as routine portfolio housekeeping.

However, pension funds are steadfast institutions; they operate across decades, not quarters. The decision to disengage from the AI-led rally instead indicates a structural market fragility that today’s market darlings rest on an uncertain and increasingly exposed foundation.

The current equity rally is among the most concentrated in recent decades. According to a recent measure, the top 10 companies in the S&P 500 now account for approximately 40 per cent of the index’s total market capitalisation, which is historically high. Further analysis finds that the top five firms alone represent around 27 per cent of the index.

When index returns depend disproportionately on a handful of names, market breadth shrinks. This amplifies systemic risk because a shock to one or two of these firms can ripple through the entire benchmark.

The present harks to the past events when investors became enamoured with the progress and technologies promising to reshape economies, often prematurely.

• Railway Mania (1840s): Capital poured into railroads before revenues materialised. When enthusiasm waned, many early investors were left with unprofitable ventures, even though railways ultimately transformed transport and commerce.
• California Gold Rush (1848–55): Thousands chased gold, but the gains were for those who sold picks, shovels and supplies. Today, the “picks and shovels” are the chip makers, data-centre infrastructure providers and energy suppliers.
• The Age of Oil (20th Century): Oil reshaped geopolitics and created chronic instability. Fortunes depended on geography and politics. Today the semiconductor supply chain that underpins AI is equally concentrated and vulnerable.
• Biotech (1990s): Early enthusiasm preceded commercial returns. Many investors lost patience before the sector matured. AI may be more transformative, but revenue streams are equally nebulous.
• The “Nifty Fifty” (1970s): A concentrated group of “can’t-miss” growth stocks ended poorly when valuations collided with reality. Today’s “Magnificent Seven” bears similarities.

Across these episodes, the pattern is consistent: rapid amplification of expectations, narrowing of bets, and belief that a small group of companies can defy valuation norms.

A narrowing of the rally results in the firms leading the AI boom to trade at steep multiples relative to the broader market. Some of the “top 10” companies have forward earnings multiples more than three or four times higher than the S&P 500 average. This implies decades of near-perfect execution. In such a concentrated market, high valuations increase vulnerability because any underperformance may affect entire indices.

Long-horizon investors, such as pension funds, are concerned about this gap between lofty expectations and realistic execution. This is a sufficient reason alone to pause, recalibrate and reduce exposure and potential shocks.

Unlike previous advancing markets, today’s AI frenzy is tightly linked to wider issues affecting the global semiconductor supply chain.

At the centre is Taiwan Semiconductor Manufacturing Company (TSMC). As of 2024, TSMC accounted for 62 per cent of global foundry revenue. In the second quarter of 2025, TSMC’s share of the global foundry market reportedly reached 70.2 per cent, with quarterly revenue of $30.24 billion.

This concentration creates a single point of vulnerability. Any disruption in Taiwan, whether geopolitical, regulatory or related to supply chains, could have global consequences. Governments are not passive observers. Policies designed to achieve domestic chip sovereignty, enforce export controls or manage trade tensions all heighten risk. Long-horizon investors see this as the soft underbelly of the whole structure.

The data-centre expansion to support AI requires more than capital. It depends on power supply, cooling, logistics, and stable infrastructure at scale. As demand grows for AI, energy and infrastructure constraints may slow deployment.

Expansion is often financed with debt. In a higher interest rate environment, the margin for error is reduced. If demand falls short or capital expenditures overrun, leveraged firms could face losses, with effects that cascade through the broader market. These structural constraints are important for investors responsible for decades-long obligations.

Pension funds’ reallocations reflect multiple concerns:

1. Narrow market breadth as a few firms dominate equity benchmarks.
2. Valuations detached from near-term earnings potential and growth premiums are high.
3. Geopolitical risk concentrated in one vulnerable supply chain node. Advanced chip production is clustered in one company and geography.
4. Energy and infrastructure constraints so data-centre growth depends on more than capital.
5. Debt-fuelled corporate expansion amplifies downside risk in a high-rate environment.

For institutions with decades-long liabilities, the current reward-for-risk profile may no longer be acceptable. Investors are not avoiding technology; they are mitigating structural fragility.

AI is not overstated as a technological phenomenon – indeed, its long-term potential is significant. History shows, however, that markets often price the right technology at the wrong price at the wrong time. Railroads, oil, semiconductors, biotech, and the internet all reshaped economies but produced catastrophic losses for those who mistimed their investments.

Today’s market treats a decades-long technological revolution as if it were a short-term earnings cycle. Pension funds, with the responsibility and horizon to act cautiously, are providing the perspicacity that markets often lack.

At Arbra Partners, we help clients look beyond the market cycle to the structural forces that are shaping risk and return. Our focus is on interpreting long-horizon signals and positioning capital to stay resilient amid concentrated rallies, geopolitical tension and shifting global supply chains. Whatever direction clients choose, we ensure their portfolios remain robust and aligned with the realities of the market of today as well as tomorrow.