The global economy is in the largest capex cycle ever, with $5 trillion by the end of the decade

Alphabet, Amazon, Meta, and Microsoft have been plowing hundreds of billions of dollars a year into AI. But the Iran war and closure of the Strait of Hormuz have highlighted another factor in capital spending.

“I do believe this is the the largest capital cycle the global economy has ever experienced—that being the energy transition,” Eli Horton, a senior portfolio manager for TCW’s equity products, told CNBC on Thursday.

He cited three drivers to the capex boom: energy security, rapid growth in electricity demand, and ongoing efforts to decarbonize.

Together, they will combine to unleash close to $5 trillion of spending by the end of this decade, Horton estimated, adding that he’s confident the trend will endure across multiple decades.

After nearly two decades of stagnating, U.S. electricity demand is growing again due to a revival of domestic manufacturing, electrification of the economy, and more recently the AI frenzy, he explained.

While skeptics may doubt the longevity of AI capex, Horton said the latest quarterly earnings from the hyperscalers show durability and accelerating revenue.

The stunning rally in Caterpillar’s fortunes exemplify this thesis as its main businesses—construction equipment, mining equipment, and power generation—look to ride the capex wave, he added. GE Vernova is another example as demand for its gas turbines has soared after years of lagging growth.

“Now they’re sold out to 2030,” Horton said. “There’s three companies in the world that make them. They have a lot of power at the table. It’s important to recognize that.”

At the same time, AI capital spending, while overlapping with the energy transition somewhat, is also galloping ahead full speed.

Analysts at Bank of America last month estimated that hyperscaler capex will top $800 billion this year alone, up 67% from 2025.

They also see next year’s total possibly reaching $1 trillion, backed by continued improvement in revenue and cash flow this year.

To be sure, a significant chunk of those astronomical amounts is due to higher chip prices. Much of that flood of hyperscaler dollars will flow to chipmakers that supply data centers with computing capacity.

“For AI semis vendors, we expect pricing power and margins to generally hold, as major compute/networking vendors could pass on cost increases to customers,” BofA said.