Is It A Bubble?

I have lived through several financial bubbles and read closely about many others. One might expect that the damage inflicted when previous manias burst would inoculate markets against repetition. Experience suggests otherwise. Memory fades quickly, and caution rarely survives contact with the prospect of extraordinary wealth generated by a technology that appears to promise fundamental change. Artificial intelligence now occupies that space.

The distinction that matters is not enthusiasm, which is natural when innovation is genuine, but excess. Bubbles form when enthusiasm hardens into certainty and prices come to reflect expectation rather than evidence. The belief that something is transformative can be correct, yet still give rise to damaging speculation if capital is deployed without regard to risk, valuation, or ultimate returns.

Over recent months, discussions with investors across Asia and the Middle East have repeatedly circled back to the same question. Is artificial intelligence becoming a bubble? I do not claim technical expertise in AI, nor do I pretend to possess special insight into short-term market movements. What I do have is a long view of cycles, and an understanding of how investor psychology tends to behave when confronted with novelty, complexity, and the fear of missing out.

It is important to acknowledge, at the outset, that bubbles are not entirely destructive. Research highlights a recurring pattern. When investors become captivated by a revolutionary idea, capital floods in at a pace that would never be justified under more sober assumptions. That surge of funding accelerates development, builds infrastructure, and brings forward innovation that might otherwise have taken decades. The cost is that much of the money is wasted or destroyed. The challenge for investors is not to deny the innovation, but to avoid being among those whose capital is sacrificed in the process.

There is near-universal agreement that AI has the potential to rank among the most significant technological advances in history. It is already influencing how businesses operate, how labour is deployed, and how decisions are made. Markets, too, have become increasingly dependent on it. A narrow group of AI-linked companies has driven a large share of recent gains in US equities, while spending on data centres, chips, and supporting infrastructure has become a notable contributor to economic growth.

What remains unclear is the shape and scale of future demand. Growth projections vary wildly, and in truth, nobody can reliably forecast how widely AI will be adopted, how quickly productivity gains will materialise, or where pricing power will ultimately settle. In such circumstances, speculation is unavoidable. Investors are being asked to commit capital today based on assumptions about markets that may not fully exist for years.

At the extreme end of the spectrum, signs of exuberance are difficult to ignore. Start-ups with limited operating history have raised funding rounds running into the hundreds of millions or even billions of dollars, sometimes without a clearly defined product or revenue stream. This behaviour is not new. It is a familiar feature of every major technology boom, from railways to radio, from telecoms to the internet.

Defenders of the current wave argue that comparisons with past bubbles are misleading. They point out, correctly, that AI is not a theoretical concept. Products are already in use, adoption is spreading, and revenues are rising quickly. Unlike during the dotcom era, many of today’s leading companies are profitable, well capitalised, and embedded in the global economy. Valuation metrics, while elevated, are not universally extreme. Nvidia, often cited as the emblem of the AI boom, trades on a forward price-earnings ratio that reflects strong growth expectations rather than pure speculation.

These arguments carry weight. It is also true that the phrase “this time it’s different” is heard during every bubble, and occasionally it turns out to be right. Sir John Templeton, who first impressed this idea upon me, observed that while the phrase is usually a warning sign, there are periods when structural change genuinely alters the investment landscape. The difficulty lies in distinguishing those moments from the many false dawns that preceded them.

Sceptics, meanwhile, see uncomfortable parallels with earlier episodes of excess. The most significant concern is uncertainty over who will ultimately earn acceptable returns from the enormous sums being invested. Hundreds of billions of dollars are being committed to AI-related infrastructure and development. The industry increasingly resembles a winner-takes-all contest, yet it is far from clear who the winners will be, how many there will be, or whether competitive pressures will erode profits even for the leaders.

Another shift deserves close attention. Much of the initial investment in AI has been funded from internal cash flow, particularly by large technology groups with strong balance sheets. More recently, however, the scale of spending has begun to push some participants towards debt financing. Debt is not inherently dangerous. Its impact depends on how much is used, the stability of the cash flows it relies upon, the quality of the assets being financed, and the margin of safety afforded to lenders. In an environment characterised by optimism and competitive urgency, maintaining discipline becomes harder.

History shows that bubbles rarely announce themselves clearly in advance. They can persist longer than sceptics expect, and they often deflate unevenly rather than collapsing overnight. Prices may pause, rotate, or correct selectively, giving the impression of consolidation rather than excess. It is only in hindsight that the boundary between rational optimism and irrational exuberance becomes obvious.

Given the scale of the opportunity and the depth of the unknowns, it is impossible to say with confidence whether current behaviour is already irrational or merely anticipatory. What can be said is that the risk of disappointment is real. Technological progress does not guarantee investment success. Many companies that help build transformative systems fail to deliver commensurate returns to shareholders.

The sensible response is neither full embrace nor outright rejection. Investors who avoid AI entirely risk missing participation in one of the most consequential developments of the era. Those who concentrate portfolios heavily in the sector, particularly at stretched valuations or with implicit leverage, risk significant losses if expectations are not met. The prudent course lies between these extremes.

Selectivity matters. Valuation matters. An honest assessment of uncertainty matters most of all. AI may reshape the global economy, but that does not absolve investors from the discipline required to survive cycles. Moderation, applied with patience and scepticism, remains the most reliable defence when genuine innovation collides with speculative excess.