Repeats to additional subscribers without any changes. The opinions expressed here are those of the author, an investment strategist for Panmure Liberum.

By Joachim Klement

LONDON, May 29 (Reuters) - Investors should be asking not whether the AI boom will end, but what will happen to markets when it does and where safety may be found.

I have previously argued that the business models of OpenAI, Anthropic and other AI darlings look increasingly unsustainable and that hyperscalers such as Meta META.O and Amazon AMZN.O face daunting maths to make their data centre investments pay off. Yet, for now, the investment surge continues to accelerate.

Annual investment in technology equipment and software reached $1.5 trillion last year, about 70% more than during the inflation-adjusted peak of the late-1990s technology, media and telecommunications (TMT) bubble, better known as the dotcom boom.

While I do not expect the AI boom to come to a crashing stop any time soon, growth will stall at some point. Every tech boom since the end of World War Two has been followed by a drop-off in tech investment.

After the cybernetics boom ended in 1962 and the speculative U.S. "Go-Go" stock market boom of the late 1960s faded in 1969, tech investment dropped 5%. Similarly, after the TMT bubble burst in 2000, it declined by 18.6% over two years, according to the Bureau of Economic Analysis.

WHAT MIGHT A PULLBACK LOOK LIKE?

Given how enormous AI investment currently is, a pullback could have serious and broad implications for markets and economies globally.

My models indicate that we don't need a repeat of the TMT crash to push the U.S., Britain and the euro zone dangerously close to recession. Just a 5% drop in U.S. tech investment would significantly harm these three major economies, my analysis shows, with real GDP in each declining by up to one percentage point in the year after such a shock.

The U.S. economy would likely recover relatively quickly, while the less dynamic euro zone would struggle for longer.

But equity markets on both sides of the Atlantic would likely not take such a shock in stride. To gauge the market risk, I modelled three possible endings to the AI capex boom: a mild correction, a normal recessionary downturn and a full TMT-style bust. In the first, the boom rids itself of short-term excess and drops 4.5% before recovering. I estimate that the U.S. stock market .SPX would enter a correction, with a drawdown of around 15%. European markets .STOXX, meanwhile, would likely drop more than 20% and enter a new bear market.

Why would Europe fall further if the AI correction starts on Wall Street? In a slowdown or recession, U.S. investors tend to repatriate foreign holdings and move into safer assets. European investors may also pull money home, but the net effect is negative for the region, adding selling pressure to its equity markets.

In the second scenario - a typical tech-driven recession with a 6% investment drop - the U.S. and Europe would both likely enter recession, compounding the selloff in stock markets. That could mean a drop of more than 20% in the U.S. market and more than 30% in Europe.

In the third scenario - a full TMT-style crash - drawdowns could exceed 50%. The best position to take in this case would likely be cash, but fund managers and other investors who want to stay in equities would instead need to find calmer waters. But where might those be?

LOOKING FOR A HEDGE

A reversal would likely hit European equities harder on average — yet the best hiding places may also be on the continent.

One sector that looks particularly attractive is European construction and infrastructure .SXOP. Germany has committed some 12.5% of its GDP to infrastructure over the next decade, spending that should continue regardless of what happens in the AI space. In fact, if a recession did hit, such spending may even be accelerated to provide fiscal stimulus to an ailing economy.

In addition, European pharmaceuticals and food – which have little exposure to AI and are typically fairly "recession proof" – could also outperform their U.S. equivalents in such a scenario.

The reason, once again, comes down to investor flows. U.S. investors, hardest hit by a tech crash, would presumably be forced to sell more of their other holdings to cover losses or margin calls, driving broader market declines and larger drawdowns even in defensive sectors. European defensives would likely face less of that forced selling pressure.

This does not mean these sectors would be immune to a wider selloff, but the key is to find spots that potentially offer better relative value in this scenario. While the AI boom may seem unstoppable, it will lose steam at some point, and investors should prepare today rather than trying to fix their portfolios in a panic.

(The views expressed here are those of Joachim Klement, an investment strategist for Panmure Liberum.)

Enjoying this column? Check out Reuters Open Interest (ROI), your essential new source for global financial commentary. Follow ROI on LinkedIn, and X.

And listen to the Morning Bid daily podcast on Apple, Spotify, or the Reuters app. Subscribe to hear Reuters journalists discuss the biggest news in markets and finance seven days a week.