For decades, the tech industry has relied on the ability of semiconductor companies to wring more power out of computer chips, making the smartphones that fit in a hand today more capable than the computers that filled entire rooms 40 years ago.

While some experts worry that era of increased miniaturization is ending, IBM is saying not so fast.

The big tech company on Thursday released details of its next advance in chip manufacturing technology, which it says could keep that innovation going for another 10 years.

Using a novel approach to making smaller transistors that act as tiny switches in microprocessors and other chips, IBM said, the new production process can squeeze nearly twice as many transistors on a fingernail-size chip as the last technology it introduced in 2021. That will offer 50 percent greater computing performance and 70 percent better energy efficiency, the company said.

Both attributes are in hot demand, particularly in the race to build data centers for artificial intelligence. Power is a particularly severe constraint, with energy-hungry A.I. chips in some cases causing construction delays for builders that can’t secure affordable electricity.

“Everyone demands more performance, but no one wants to pay for the power,” Huiming Bu, an IBM vice president who leads chip research and development, said in a briefing with reporters.

IBM, though a pioneer in semiconductors, doesn’t make or sell chips anymore. But engineers at its laboratory in Albany, N.Y., still develop new technology for turning silicon wafers into chips, which it typically licenses to manufacturers.

Mr. Bu said the technology should be ready in the next five years, but he declined to disclose potential users. Past licensees have included Samsung Electronics and Rapidus, a Japanese company.

Industry analysts briefed on the IBM announcement were impressed, while noting that the approach would face competition.

“This is a big deal,” said Dan Hutcheson, an analyst at TechInsights. “It basically puts another 10 or 15 years on the road map.”

The chip road map is a shorthand for Moore’s Law, an often-cited tenet about progress in chips attributed to the Intel co-founder Gordon Moore. It projects that companies will pack twice as many transistors on a chip every year or two, boosting performance while lowering the cost per transistor.

Chip manufacturing has become so expensive that the cost benefits of making transistors smaller have vanished, leading industry leaders like Jensen Huang, chief executive of the chip giant Nvidia, to declare Moore’s Law dead. But computing speed and data storage capacity have continued to improve thanks to tinier transistors.

Miniaturization gains are frequently described in nanometers, or billionths of a meter, though that nomenclature is now less a precise measurement than a marketing term to distinguish technology generations. Taiwan Semiconductor Manufacturing Company, for example, is making chips rated at roughly two nanometers. Intel has a comparable technology described as 1.8 nanometers.

IBM described its new production process as 0.7 nanometers, the first to break the one-nanometer threshold. The underlying innovation, which IBM first described a year ago, is what the company calls a “nanostack” transistor.

It is the latest example of building transistors from three-dimensional structures — akin to tiny skyscrapers, rising from the surface of the chip — rather than simply shrinking transistors in lateral directions.

TSMC and Intel recently adopted a new 3-D structure, called a nanosheet transistor, which was also pioneered partly by IBM. IBM’s latest approach takes two wafers with nanosheet-style transistors and effectively glues one upside-down on top of the other, creating a structure that vertically links two types of transistors in a compact space, IBM said.

Imec, an influential research center in Belgium, is backing a next-generation technology that has attracted interest from multiple chip manufacturers, Mr. Hutcheson said. It builds three-dimensional structures layer by layer, a process that he said could create defects.

With IBM’s approach, “you get a much better transistor, with faster speed and lower power,” Mr. Hutcheson said. “It’s pretty revolutionary.”

Patrick Moorhead, an analyst at Moor Insights & Strategy, said questions remained about which partners might adopt IBM’s technology and whether it would hit the market first. But the announcement nevertheless suggests that proclamations of the death of Moore’s Law have been exaggerated.

“It is a sign that we are not going to run out of gas like we always think we are,” Mr. Moorhead said.