ROI-AI can be both a bubble and a breakthrough: Mike Dolan

By Mike Dolan

LONDON, Nov 6 (Reuters) - Bubble or bonanza? AI could be both.

Artificial intelligence may well be the future of business and the wider economy, yet the sky-high stock prices it's generating may still represent an unsustainable bubble.

about whether frothy tech stocks are in is rooted in fears that AI may not live up to the hype and that business returns will therefore never justify the blistering investment we're seeing on advanced chips as well as the infrastructure and energy needed to support AI's vast processing needs.

will be written on the "AI hype" question in the coming months, with most likely concluding that "it's too early to tell".

Some early studies [https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf] say companies are not yet reaping significant returns from AI, even as an estimated $3 trillion of investment is expected to be ploughed into the technology over the next three years. But generating returns was always going to take time.

As to the consensus view? Maybe that's best left to AI itself.

In response to the question "Is AI a bubble?", Google's "AI Mode" concluded: "While the core AI technology is considered a genuine and potentially revolutionary development, there is significant debate about whether the current market valuations are sustainable or if the sector is experiencing a bubble-like period of over-exuberance and speculation."

Productivity gains and revenue growth over the coming years will ultimately determine AI's value, it adds, clearly reluctant to talk itself out of a job just yet.

Perhaps it's more useful to look at the lessons of past bubbles. History suggests that revolutionary technologies typically attract a flood of capital, not all of it "smart money".

So even if the technology endures and ultimately transforms the world, some of the priciest stocks from the bubble years can still be left high and dry. The companies may not go bankrupt, but over-eager valuations can mean their share prices take years to recover - if they ever do.

That certainly seems to be the conclusion of short-seller - of "The Big Short" fame - on today's AI darlings Nvidia and Palantir . He doesn't think they're bad companies. He simply thinks their stock is overpriced.

It's impossible not to compare today's valuation frenzy to the 1990s dotcom bubble in tech, media and telecom stocks that burst spectacularly in early 2000.

AI optimists routinely cite the long-term performance of those investors who bought some of the TMT survivors, even at the height of the frenzy.

After all, if you were lucky enough to buy Amazon , Apple or Microsoft at their peaks around the dotcom bubble, you're still looking at eye-popping returns now.

That said, you'd have had to wait up to seven years after the bust to break even.

For the bear case, AJ Bell's investment director Russ Mould homes in on networking equipment giant Cisco , one of the brightest stars of the late 1990s. Its stock shot up from $2 a share in 1995 to more than $80 at the peak of the bubble in 2000, doubling in price in 1999 alone.

Within 18 months, it was back at just $8 a share.

IN THE LONG RUN, WE'RE ALL DEAD

Cisco did eventually deliver as a business, as Mould points out. The company's bottom line has almost quadrupled in the past 25 years, with net income of more than $11 billion forecast this year alone on almost $57 billion of sales.

But its stock price still hasn't recovered its March 2000 peak, when the firm's market cap of $550 billion was 200 times its annual income. Cisco's share price may have doubled over the past 10 years, but it remains 10% below the intraday peak at the dawn of the new millennium.

Mould reckons Cisco is a lesson in how hard it can be to meet short-run expectations even if you do so in the long term.

And, importantly, it's a reminder that stretched valuations provide little protection if growth rates suddenly slow and sales disappoint - something that should be top of mind for investors in defence software star Palantir, which currently has a forward price/earnings ratio of 216 times.

Perhaps Nvidia best embodies the conundrum facing tech investors today.

On the one hand, the chipmaker's market cap - which recently eclipsed $5 trillion - today accounts for 16% of U.S. GDP, compared to some 5% for Cisco at its peak.

However, looked at another way, Nvidia's valuation can look downright reasonable.

"Nvidia's current rate of growth makes Cisco's look pawky by comparison, and the silicon chip specialist generates fatter profit margins and higher returns on capital than the telecom equipment giant ever did," Mould wrote, adding that Nvidia's stock is considerably better valued too at about 33 times forward earnings.

So can today's tech investors rest easy?

Depends on their holding period.

While the dotcom bubble is essentially a blip on a historical chart of the Nasdaq, it's easy to forget that the relentless bear market that followed lasted almost two years, and that it took about 15 years for the index to regain its peak.

Stocks always go up, right? As long as you hold on forever.

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. Plus, sign up for my weekday newsletter, Morning Bid U.S. [https://www.reuters.com/newsletters/morning-bid-us/]

Nasdaq took 15 years to recapture peak of 2000 https://tmsnrt.rs/4qB3utg [https://tmsnrt.rs/4qB3utg]

Cisco still shy of its peak from 2000 https://tmsnrt.rs/4otbawi [https://tmsnrt.rs/4otbawi]

Palantir's lofty valuations https://tmsnrt.rs/43dTL24 [https://tmsnrt.rs/43dTL24]

Nvidia's earnings boom keeps valuations in check https://tmsnrt.rs/4qJzbk5 [https://tmsnrt.rs/4qJzbk5]

(by Mike Dolan; editing by Jamie Freed)