Goldman Sachs refutes three major concerns about the AI bubble

Ben Snider, the chief U.S. equity strategist at Goldman Sachs Research, dismissed concerns about an artificial intelligence (AI) bubble in a recent interview and said he will continue to focus on three major themes in AI trading.

Three Major Concerns

First, there is a widespread market concern that hyperscale data center operators may suddenly scale back their huge expenditures. If a company believes its return on investment is not satisfactory, it may significantly cut its spending expectations. People think that if investors "reward" such a move, other companies may follow suit.

However, Snider believes the risk of such a shift is small. He said that the growing spending on AI construction should continue to drive this trading.

Another concern outside is that despite the strong profit growth in the AI and technology sectors, their valuations are quite cheap. This indicates that investors may be reluctant to pay a higher premium for these stocks because they doubt how long the profit boom can last.

For Snider, this is a counter - intuitive argument.

He said, "Interestingly, investors see both over - valuation and under - valuation as warning signs."

"I'm always relieved when I hear bearish views," he continued. "It's an affirmation for me that there is still a certain equity risk premium in the market. If one day everyone thinks the outlook is bright and there's nothing to worry about, that day clearly indicates that the market is over - valued."

The last worrying issue is that various valuation indicators point to different signals. The Shiller P/E ratio is close to historical highs, hovering around the peak levels during the dot - com bubble and in 2021. Coupled with the more than 20% increase of the S&P 500 index in the past year, people can't help but wonder if its upside is limited.

But other valuation indicators present a different picture, such as the forward 12 - month P/E ratio. Snider simply pointed out that although the S&P 500 index has risen significantly in the past 12 months, its P/E ratio is lower than it was a year ago.

This indicates that it is still profit growth, rather than P/E expansion, that is driving the market. Snider said that the most attractive part of the market remains the companies that have performed well in recent months and years.

Three AI Themes

Snider highlighted three sub - sectors in AI trading that he is currently most focused on.

First, he continues to focus on AI infrastructure stocks because their valuations are relatively low. This includes companies that produce semiconductors, servers, and AI networks.

"For most areas of the semiconductor industry, including memory chips, the valuation multiples have hardly expanded, which indicates that the market is still skeptical about the semiconductor industry," he said.

Second, Snider said that the power infrastructure theme is "extremely attractive." This benefits both from the continuous investment in the AI field and the impact of the Iran war on power and energy. Energy infrastructure has been the pillar of the "HALO (Heavy Assets, Low Obsolescence)" trading in 2026, and infrastructure construction is the key to meeting the endless power demand of the AI industry.

Third, Snider is bullish on hyperscale cloud service companies that have invested heavily in AI construction in cloud - platform data centers. Such as Amazon, Microsoft, Meta, Alphabet, Oracle, and IBM.

These stocks have performed well since the end of 2022, but since the beginning of this year, many of them have underperformed as investors have shifted to stocks that benefit from hyperscale data center spending.

"They've just had one of the worst months on record. So, from a profit perspective, their valuation multiples are very low. In terms of P/E ratio, they are trading at the lowest levels in the past decade, very similar to the trading situation at the end of the bear market in March 2020 or 2022," Snider added.

This article is from the WeChat official account "Caixinlian She", author: Huang Junzhi. It is published by 36Kr with authorization.