AI PC sales slower than expected, but they’re still set to dominate the market

The adoption of artificial intelligence (AI) personal computers (PCs) has proven to be slower than expected. However, they’ll command the majority share of global PC shipments next year.

Last year, Gartner forecasted AI PCs to represent 43% of all PC shipments in 2025. The American business and technology insights firm expected the shipments to reach 114,2 million units by the end of this year.

However, now the forecast has been adjusted to just 31% of the total PC market and 77.8 million units. The sales number is 32% lower than initially expected.

The firm defines AI PCs as computers equipped with an embedded neural processing unit (NPU), a dedicated hardware (processor) accelerating artificial intelligence and machine learning tasks.

NPUs are integrated into most ARM architecture CPUs and are now also appearing in x86 processors. All new MacBook and Mac computers now come with NPUs. However, ARM adoption in Windows has been strained by software compatibility issues and a limited ecosystem.

“AI PCs are reshaping the market, but their adoption in 2025 is slowing because of tariffs and pauses in PC buying caused by market uncertainty,” said Ranjit Atwal, Sr. Director Analyst at Gartner.

Despite the struggles, Gartner expects AI PC sales to reach 54.7% of the market next year.

“Nevertheless, users will invest in AI PCs to ensure they are prepared for the growing integration of AI at the edge,” Atwal added.

The firm expects AI PC shipments to reach 143 million units next year and to become the norm by 2029.

In the business segment, users still prefer x86 and Windows, which is expected to make up 71% of the AI business laptop market in 2025, leaving 24% to ARM. ARM-based laptops are gaining a larger share of the consumer market.

Only two percent of software vendors prioritized investments in AI capabilities directly on PCs last year, but the situation is quickly changing. Gartner expects that 40% of software vendors will prioritize AI capabilities next year, and multiple small language models will run locally.

Running advanced AI tools directly on PCs enables faster response times, lower energy consumption, and reduced reliance on cloud services, helping keep user and business data private and secure.

Atwal believes that the future of AI PCs is in customization.

“It lets users configure their devices with the apps, features, and functions they want. The more users interact with a vendor’s AI PC, the better the vendor understands them, allowing for ongoing improvements and stronger brand loyalty,” he said.