ChatGPT has transformed AI from a niche, behind-the-scenes tool to one of the fastest-growing consumer products in history. That trend has further accelerated this year, as the use of tokens, the fundamental building blocks of any AI query, has skyrocketed this year thanks to widespread enterprise adoption and the rise of agent AI platforms like OpenClaw.
This is causing AI costs for enterprises to skyrocket as their systems start consuming more and more tokens. “Every chat generates a token, and as you expand the use of AI across your organization, costs start to add up,” said Prakash Arunkundrum, chief strategic transformation officer (CSTO) at HP. luck. “What all data center people are trying to solve is how to lower the cost of tokens.”
HP’s solution to the token problem is to avoid the token problem altogether by moving AI from the cloud to people’s devices, or “the edge.” HP is betting on “AI PCs,” personal devices that can run AI models locally, and has also debuted its own local-first AI model, HP IQ.
“The core of our strategy is building meeting rooms with AI-powered devices, thinking AI PCs, Copilot-enabled printers, and smart intelligence peripherals that understand the user,” Arunkundrum says. “Then we can stitch them together with software to deliver experiences across devices in the office.”
HP reported better-than-expected revenue in the first quarter of 2026. The US company posted revenue of $14.4 billion, an increase of nearly 7%. Personal systems sales increased 11% to $10.3 billion. Interim CEO Bruce Broussard attributed the jump to “the continued momentum of AI PCs,” which currently account for 35% of devices sold. The company is scheduled to announce its second quarter financial results on May 27th.
“Edge AI is local, private, and secure, reducing total cost of ownership by reducing the use of tokens,” Arunkundrum said. “Imagine doing quality inspection on a production line. You don’t want to move it elsewhere.”
Running AI locally also helps ensure data sovereignty. This is important in Asia, Arunkandram argues. In Asia, governments increasingly view data protection as essential for privacy, national security, and technological self-sufficiency.
“In Singapore and other parts of Asia, the need for a sovereign cloud with its own environment prohibits the use of AI in some cases,” says Arunkandram. “With edge AI, you can run many of these models directly from your device, whether open source or proprietary, so all your data stays in-country.”
Arunkundrum’s Technology Journey
Aruncandram spent his early career working at management consulting firms in the San Francisco Bay Area, including PwC and AT Kearney. He then spent 10 years at Logitech, eventually becoming president of the business division.
Arunkandram said his 10 years at Logitech shaped the way he thinks about technology. “I’m a product guy, so I consider the customer and user experience to be the most important thing. For example, when we make a computer mouse at Logitech, we study how it interacts with the user’s hand and how it can become an extension of the user,” explains Arunkandram. “From a leadership perspective, you have to start with the customer and think about their perspective, what are their frictions, what are their pain points, and what can help them be great.”
Arunkundrum joined HP last October as CSTO, a new role created specifically for him. He affectionately refers to HP as an “OG startup” and says he was drawn to its deep heritage. “Every technology company you know is influenced by HP, a business that founders Bill and Dave started in their garage,” Arunkandram mused.
His first year at the company was hit by a global memory chip shortage. AI has caused a huge surge in demand for these chips, and the most sophisticated chips are being produced by a small number of manufacturers. This has led to soaring prices due to supply shortages, putting pressure on device manufacturers around the world.
HP reported an operating margin of 5.3% in the first quarter of this year, which was lower than 6.3% in the same period last year. The company also warned in February that “increased memory costs” could cause this quarter’s profits to fall below the lower end of expectations.
Arunkandram declined to comment on how the memory chip shortage is specifically impacting HP, citing the company’s future earnings. Still, he explained, “Memory pressure is affecting everyone…and as a large company with millions of devices that people need, we certainly have overcome this problem.”
HP’s AI Playbook
As a new generation of AI labs scrambles to stay ahead of the curve in Silicon Valley, HP is betting on its advantage in putting AI to work on devices and in the office.
An AI-driven software layer called Workforce Experience Platform (WXP) sits on top of HP’s business hardware innovations. First announced in 2024, WXP serves as a “command center” for a company’s internal IT, providing chief information officers with a complete view of a company’s devices and applications through telemetry and analytics.
Under Arunkundrum’s leadership, the company is also working on its AI revolution from within. In the APAC region, 78% of staff are now using AI in their daily workflows, higher than the global average of 72% (as calculated by Boston Consulting Group).
Last October, the company launched Garage 2.0 in Singapore. Garage 2.0 is an innovation incubator that gives startups access to HP’s engineers, infrastructure and customers, and is, in Arunkandram’s words, a deliberate recreation of the “very small garage” where the company itself got its start.
“The idea behind Garage 2.0 was to connect startups with other people in the ecosystem that we work with, such as academics and people who can bring products to market,” says Arunkandram.
Above all, Arunkundrum wants HP’s products to benefit society more broadly. “Our hope is that we can influence humanity and better expand human potential,” Arunkandram concludes. “That’s my goal in 10 years.”
