LAS VEGAS, 9 January 2026 — Four years after ChatGPT propelled artificial intelligence into the mainstream, nearly every exhibitor at this year’s Consumer Electronics Show (CES) in Las Vegas showcased an AI-powered device, underscoring the industry’s shift from software toward so-called “physical AI” spanning robotics, humanoids and autonomous driving technologies.
The annual gathering of global technology and automotive giants highlighted how AI is moving from cloud-based applications into the physical world. Alongside big-tech roadmaps for future homes and lifestyles, dozens of start-ups and lesser-known companies competed for attention with AI-enabled gadgets designed for tasks ranging from hair cutting to emotional support.
For investors and analysts, the challenge was identifying which AI-enhanced hardware could resonate with consumers and scale into multi-billion-dollar businesses.
“AI is really driving a whole innovation and demand cycle,” said Chris Bergey, head of Arm’s unit overseeing PCs, phones and other AI devices. Arm told Reuters it has reorganised to create a physical AI unit to expand its presence in the robotics market.
Humanoids draw crowds, but challenges persist
Interest in humanoid robots surged at CES as companies pitched human-shaped machines as the next frontier of automation. Robots from LG Electronics and others played poker, folded paper into pinwheels and danced with attendees.
Yet the slow pace of these demonstrations highlighted ongoing hurdles, including processing power, battery life and handling scenarios beyond programmed instructions. As a result, experts do not expect affordable, widely usable humanoid robots to be available anytime soon.
AI hardware and software converge
Chinese technology group Lenovo hosted a high-profile event at the Las Vegas Sphere featuring Jensen Huang of NVIDIA and Lisa Su of Advanced Micro Devices. Lenovo unveiled its Qira AI voice assistant platform, designed to operate across devices and integrate services such as those from travel firm Expedia Group.
Elsewhere, Meta Platforms said it had enhanced its Ray-Ban Display and Neural Band products with features including live recording, while Alphabet introduced its Gemini AI model for televisions and home devices.
Alongside these announcements were more playful and utilitarian offerings, from AI-powered three-dimensional avatars that speak in multiple voices to a pocket pet designed to evolve with its owner. One company demonstrated an AI dry-cleaning device priced at US$599, while another showcased AI-assisted hair clippers.
After reviewing AI-enhanced lawnmowers, massage chairs and mattresses, Seaport Research analyst Jay Goldberg noted that many of these products were once marketed simply as “smart” devices, suggesting that the AI label may now serve as a branding tool.
Chips under the hood
While many AI applications still rely on the cloud, costs are pushing more processing onto devices, increasing demand for specialised chips.
“Unless you’ve been living under a rock over 2025 you probably heard that AI is getting really expensive,” said Aravind Srinivas at an Intel launch event.
Intel’s PC chip chief Jim Johnson echoed concerns over escalating cloud-computing costs. The company unveiled its Panther Lake AI chip for laptops, its first product using the next-generation 18A manufacturing process. AMD also announced new processors designed for AI-enabled PCs.
AI-powered computers promise gains in speed, battery life and security, but whether consumers will embrace them remains uncertain.
“I think the general consumer is still not aware of what an AI PC really means or what it can do for them that they can’t do today,” said Ben Bajarin, chief executive of Creative Strategies. “Hopefully that will become more clear over time.”
