There’s much anticipation among enterprises and consumers alike for AI PCs, the evolution of computing that integrates dedicated AI hardware and software into everyday machines.
But as with any new technology, it’s unclear when AI PCs will finally have their moment. Top analyst firms, including Gartner and IDC, certainly have high expectations for the technology, based on new reports out this week.
Gartner forecasts that 43 million AI PCs will be shipped this year. Similarly, IDC predicts that 57 million units will be shipped in 2024, and by 2028, AI PCs will represent 92% of all PC shipments worldwide.
However, “while AI has been a buzzword of late, it has yet to be a purchase driver among PC buyers,” said Jitesh Ubrani, research manager for IDC’s worldwide mobile device trackers division. “Businesses certainly recognize the importance of AI, though many struggle to see the immediate use case and instead are opting for AI PCs as a means to futureproofing.”
An ‘onslaught’ of AI PCs
Gartner predicts that by 2026, AI laptops will be the only choice of laptop available to large enterprises (compared to less than 5% availability just three years prior). Further, AI PCs, including both desktops and laptops, will represent 43% of all PC shipments by 2025 (compared to 17% this year).
IDC also predicts that shipments of AI PCs will grow from nearly 57 million units in 2024 to more than 258 million in 2028. Still, the firm tempered its short-term assessment of the market, calling it “modest.” But, Ubrani noted, long-term, there will be an “onslaught” of AI PCs as NPUs are integrated into lower-tier hardware, and producing processors without NPUs becomes cost prohibitive.
North America, major countries in Asia Pacific (APAC), and some western European countries will lead AI PC adoption, IDC predicts, although the firm points out that this will largely be driven by the need for new PCs, rather than need for AI PCs specifically.
Gartner Director Analyst Mika Kitagawa noted that most enterprise-grade laptops will have an NPU core by 2026, which will be a major push from the supply side. From the demand side, on-device AI will be a standard requirement for the enterprise, as more AI functionality shifts from cloud-based to on-device-based; NPUs will speed up on-device AI tasks.
Also, while in 2025, Windows x86-based AI laptops will lead the business segment, that dominance eventually will reduce over time, Gartner said. However, this will not occur in the next five years.
While Arm-based laptops have benefits — they are usually quiet, because they do not need a fan to cool the processor, and have low power consumption, thus extending battery life, there are software incompatibilities on Arm-based Windows devices. This is because not all business applications are natively written for Arm and must use an emulator, Kitagawa explained.
“This could create limited user experiences,” she said. “The majority of enterprises will keep using x86 in the next five years.”
We’re not quite at the tipping point yet
A big reason behind the shift to AI PCs is the arrival of language models that are “distilled, fine-tuned and pruned” into small language models (SLMs) and tiny language models (TLMs) that are able to operate on AI PCs where large language models (LLMs) can’t, explained Jon Nordmark, CEO at enterprise AI platform company Iterate.ai. AI PCs can operate without needing to connect to the internet, and an SLM can be utilized when the computer is plugged in; users can opt to leverage a TLM when on battery to save power.
For companies and consumers, operating AI on a PC is more sustainable, Nordmark explained. Each query or prompt requires about 10 times less energy compared to on the cloud. Privacy is also a big part of demand.
“The AI PC generative AI experience can really be an oracle that users can ask complex business questions and trust to help navigate key decision making,” said Nordmark. This could include, for instance, finding effective SEO keywords, or getting advice on a specific business decision.
AI PCs could “substantially change computing” as they enhance tasks such as real-time language translation, advanced content creation, and workflows, leading to improved productivity, creativity, and personalization, agreed Matt Hasan of aiResults, which specializes in AI for marketing and customer lifecycle optimization (CLO).
With AI hardware built-in, he pointed out, users can run complex algorithms such as image recognition or natural language processing (NLP) locally, providing better speed, security, and the convenience of offline processing.
Still, “while AI PCs are exciting, they may not become the norm as quickly as some predict,” said Hasan. Challenges include the high cost of integrating AI chips, not yet proven everyday use cases, and concerns about data privacy and AI misuse. But “the current hype does suggest that AI-powered computing will eventually become an integral part of the future, even if the transition takes longer than the optimists claim.”
Instead of being passive tools, AI PCs are designed to anticipate needs, automate repetitive tasks, and make decisions based on user habits, said Jon Morgan, CEO and editor-in-chief of business advisory firm Venture Smarter.
“It’s a major leap from the static machines we’ve been using, making computing far more personalized and efficient,” he said. The allure is that AI integration isn’t just about “raw power,” but enhancing experience and adapting to individual workflows.
“We’re not quite at the tipping point yet, but I don’t think it’s far off,” said Morgan. “Once the early adopters demonstrate the practical benefits, the shift could happen faster than expected.”
