Get all your news in one place.
100's of premium titles.
One app.
Start reading
The Economic Times
The Economic Times
Devina Mehra

Double, Trouble, Toil and the AI Bubble: Why the boom may not be what it seems

In popular narrative, India's losing out as an economy and stock market because it doesn't have any AI or innovation story. But what is the AI story? Is it all that it's made out to be?

Take the stock market first. AI tech was the driver for global markets. In the last 3 yrs, you could get by in global investing just buying stocks of the 'Magnificent Seven' - Microsoft, Apple, Amazon, Meta, Alphabet (Google), Tesla and Nvidia. But these are not actually driving the global, or even the US, market any more.

Also Read: AI giant Anthropic confidentially files for US IPO

Leadership has moved from the S&P 500 index to the broader market. Russell 2000, the small-cap index for the US, is up 18% for the year up to May vs 11% for S&P 500.

It's no longer the Mag Seven driving even S&P 500. In 2023 and 2024, they accounted for over 50% of the S&P upmove. In 2025, this came down to 43%, and only two of them - Nvidia and Google - outperformed the index. In 2026, three of the seven are down.

The old economy sectors have made a tremendous comeback.

While the tech sector is up for the year, it's not companies setting up AI capacity, but semiconductor, hardware and electronic equipment sub-sectors that have rallied between 30% and 168% leading this run. So, the likes of Intel, AMD, Texas Instruments, Corning, ON Semiconductor, Micron, Qnity Electronics, SanDisk and Dell - which are on various points along the semiconductor and hardware chain - are all up between 60% and 600% in just 5 mths. These are the sub-sectors that have also driven Asian markets like South Korea and Taiwan.

AI mega-scalers are spending an astounding $750-900 bn this year in capex. Their stocks might be struggling, but all those suppliers supplying equipment and technology for their data centres are the ones doing well for now. But semiconductor and semiconductor equipment sectors, like most capital-intensive industries, are highly cyclical. Right now, they have both great demand and unbelievable pricing power. This may not be the case 2 yrs later when capex subsides.

Also Read: Young and unemployed? Remote work, not AI, may be the problem, study finds

So, what can we expect from the AI industry?

Most of the hype around AI is being created by companies in fundraising mode, either from private markets or IPOs. Others whipping up frenzy are early investors in companies like Open AI and Anthropic. There is no independent verification of what they are saying.

One Indian-origin Silicon Valley biggie was talking about how AI will bring down the cost of most services to zero. Zero?! AI is one of the most capital-, energy- and water-intensive industries of all time. Who exactly is going to pay for returns on these huge capital investments?

The real questions to ask are: how long will AI tech take to succeed? Which particular technology and companies will succeed? Will even those companies make adequate return on capital employed?

After the dotcom bust, it was thought that since the internet will be used by everybody, you couldn't go wrong investing in internet infrastructure companies. The internet is still carried across the world on undersea cables. Global Crossing laid many of these cables, which are still in use, but the company went bankrupt more than 20 yrs ago.

Between mega data centres and hiring or acquiring talent at mind-boggling numbers, the cost of being in the AI game is sky-high, even though 60-70% of usage is currently by low-paying students. There are also additional dangers with this growth being financed with debt, both by mega players and by others from whom they are leasing capacity.

Total debt for data centres is estimated to cross $1 tn by 2028. This is mostly against the security of fast-depreciating equipment and buildings, increasing risk in the financial system. Then there are a bunch of financial tangles between companies. Net of all the details, it's a case of capital funding by bigger companies like Nvidia, Meta or Microsoft coming back as revenue for them.

For now, the AI music is still playing. As Citibank's then-CEO Chuck Prince said before the 2008 mortgage crisis, 'As long as the music is playing, you've got to get up and dance.' This time, too, the bubble will burst. The only unknown factor is when.

For more news like this visit The Economic Times.
Sign up to read this article
Read news from 100's of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
One subscription that gives you access to news from hundreds of sites
Already a member? Sign in here
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.