Al Root

Chinese AI chatbots, data-center spending, the Jevons' paradox, and Sputnik moments have investors' heads spinning. So far, their reaction has been to sell first and ask questions later.

When the dust settles, there will be some opportunities in manufacturing stocks that have benefited from the artificial-intelligence trade. What has shaken everything up is that a Chinese AI company called DeepSeek released an AI model that performed almost as well as OpenAI's ChatGPT.

That is no big deal. But viral posts over the weekend suggested the chatbot was created for $6 million, a fraction of the cost of others such as OpenAI's ChatGPT, calling into question the capital spending plans of AI players such as Microsoft, Amazon.com, Alphabet, and Meta Platforms. A closer look at the situation shows the $6 million isn't correct, but it isn't clear how much DeepSeek spent.

Capex at the four U.S. tech companies is expected to total $261 billion in 2025 and $283 billion in 2026, up from $216 billion in 2024. That spending becomes revenue at the likes of Nvidia and countless manufacturing companies that help build data centers, or make electrical connectors to plug Nvidia chips into server stacks.

"The key metric to watch is cloud service providers' capex plans for 2025 and 2026," wrote BofA Securities industrial analyst Andrew Obin in a Tuesday report titled "DeepSunk? Implications of DeepSeek on multi-industrial stocks."

It is a clever title. And watching what the companies say about capital spending sounds easy enough.

Predicting capex in 2026, however, isn't simple, especially now. Even if demand from the big four falls, lower prices can lead to more use of AI.

That is essentially Jevons' paradox, named for the English economist William Stanley Jevons. Technological improvement -- in this case falling costs for useful AI models -- means more total demand.

The effect shows up everywhere. If steel gets cheaper, more steel gets used. More cars get sold when cars become less expensive. In technology, semiconductors and memory have gotten cheaper per unit of computing power, so everyone now has a powerful computer that can make phone calls in their pockets.

In the current instance, lower costs for AI computing could mean more than just the big four or five hyperscale data center players would be spending money.

Obin points out that while hyperscalers account for about 25% of total data-center capacity, so-called colocators make up about 50%. Those are companies whose computers are at server farms run by third parties. That business segment isn't growing like the hyperscalers, but if AI gets cheaper, it could.

That is the basis of Marc Andreessen's "Sputnik moment" tweet from Sunday. The Soviet Union launched its Sputnik satellite in 1957, marking the start of a space race between the U.S. and the Soviet Union. The U.S. federal space budget went from about $500 million a year to $10.5 billion in 1958, Obin points out.

Obin isn't changing his assumption for 2.8% annual growth in U.S. electricity demand from 2023 to 2030. That growth underpins demand assumptions for a host of companies that make electrical components, cooling technology, and grid equipment.

Growth of 2.8% might not sound like a lot, but it is way faster than in recent history. U.S. electricity demand growth grew less than 1% annually between 2013 and 2023.

What about the stocks? Obin has Buy ratings on shares of the data-center infrastructure provider Vertiv, the electrical equipment maker Eaton, and GE Vernova, a provider of power-generation technology.

Vertiv stock dropped almost 30% Monday, closing at $102.60. GE Vernova stock fell more than 20% to $330. And Eaton shares slid about 16%, ending the day at $311.55.

Obin's price target for Vertiv stock is $165. His targets for Vernova and Eaton are $485 and $410, respectively. Those prices imply an average gain of about 45% from Monday's closing levels.

That is attractive, but investors will have to wait and see if it is attractive enough amid all the recent AI turmoil.

The early answer is not yet. Vernova stock was down 1.7% in midday trading while the S&P 500 and Dow Jones Industrial Average were up 0.6% and 0.5%, respectively. Eaton and Vertiv stocks were both down close to 3%.

Write to Al Root at allen.root@dowjones.com

This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

January 28, 2025 12:06 ET (17:06 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.