The opinions expressed here are those of the author, a columnist for Reuters.

By Jamie McGeever

ORLANDO, Florida, May 27 (Reuters) - As U.S. hyperscalers' investment in AI keeps rising, so too do concerns that the eye-popping outlay will fail to generate adequate returns. Does that mean the stock market is in a tech-inflated bubble that is bound to burst like it did after the dotcom mania of the late 1990s? Probably not.

That doesn't mean the parallels aren't raising some red flags.

For starters, the current AI capex boom, which includes investment in data centers, power, equipment and software, is not only the biggest spending splurge since the 1990s. It's also the biggest in history.

AI-related capex is expected to reach roughly $800 billion this year, according to analysts at Goldman Sachs and Morgan Stanley, who both recently revised up their forecasts. The Morgan Stanley team also raised its 2027 AI capex outlook 17% to an eye-popping $1.12 trillion.

As a reminder, hyperscalers' capex bill in 2024 was only $260 billion, according to Morgan Stanley.

This massive investment is rapidly whittling away Big Tech's enormous cash mountains. Analysts at PIMCO reckon capex will absorb 94% of hyperscalers' operating cash flows over the next two years, up from 40% in 2023.

This has led to a surge in borrowing. Big Tech's new debt issuance this year is already around $135 billion – more than last year's total – and is set to rise further. AI capex will increasingly be debt-financed.

Today's "we can't afford to fall behind" mania is certainly reminiscent of the environment in the late 1990s when aggressive competition and large investment outlays led to significant overcapacity in fiber optic cables, as Lotfi Karoui, a managing director at PIMCO, points out.

After that bubble burst in the early 2000s, it was widely agreed that too much capex had been spent in too short a time, even though internet usage exploded and has never looked back.

Fast forward to today. If demand for AI in the next few years fails to match the mushrooming supply, we'll have a problem – even if AI does end up changing the world.

PARALLEL SIGNS

But there are also differences that suggest a repeat of 2000 isn't in the cards.

Firstly, many of the big telecoms in the late 1990s were either barely profitable or actually losing money. As a result, many became very highly levered and were heavily reliant on foreign investment.

The story is rather different today. The balance sheets of AI titans remain quite strong - even after the recent surge in debt issuance - and their profit margins continue to be sky-high.

In fact, the U.S. technology sector is currently one of the least leveraged of all sectors, particularly when zeroing in on hyperscalers. If required, there appears to be ample room for further Big Tech borrowing before the alarm bells start to ring.

Let's imagine that 94% of operating cash flows over the next two years are eaten up by capex, as Karoui and his colleagues expect. That would still be lower than telecom companies' comparable ratio in every year of the dotcom era. Indeed, this figure exceeded 200% and 150%, respectively, in the final two years of the boom.

Equity investors today are also displaying more restraint than they did back then. Tech valuations measured by forward price-to-earnings ratios are well below the peaks of 2000, and hyperscalers' relative performance against the S&P 500 in this cycle has been much more stable than the Nasdaq's rollercoaster ride in the dotcom era.

"AI is in the midst of a capex boom with genuine risks: uncertain monetization, potential overbuild, shortening asset lives, and growing reliance on debt,” says PIMCO's Karoui. "But for now, it is a more disciplined and far more financeable cycle than the late-1990s telecom boom."

And if the seemingly constant upward revisions to capex forecasts are any guide, this cycle could actually be in its early phase, with plenty more room to run. Legitimate concerns about overinvestment and overcapacity are for the medium term, not today.

(The opinions expressed here are those of the author, a columnist for Reuters)

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.