HP has a solution to expensive AI tokens: Buy a more expensive PC

HP executives discussed the rising costs of AI tokens during their Q3 earnings call, suggesting that enterprises facing this issue might consider purchasing more expensive PCs to run AI models locally. This approach could help customers optimize costs, reduce dependence on network connectivity, and provide faster response times.

HP's Personal Systems president, Ketan Patel, emphasized that AI PCs are now offering more value compared to previous years, particularly in terms of token economics. Patel also highlighted the benefits of keeping sensitive data close to where it is generated, as well as the reduced dependence on network connectivity and improved response times when running AI models locally.

The company is already seeing customers using AI PCs for various purposes, including employee productivity, engineering and design, customer service, and predictive maintenance. HP expects the percentage of AI PCs in its PC mix to increase to between 60 and 70 percent by fiscal 2027 and over 70 percent by 2028. The shift is anticipated to be margin accretive, as AI PCs come with a premium price tag and offer opportunities for additional services and security products.

HP's Personal Systems revenue grew 18% year-over-year to $11.8 billion, despite a 16% decline in PC volumes, primarily attributed to higher prices, a richer mix of premium machines, and sales of attached services and peripherals. CFO Karen Parkhill stated that HP expects to continue increasing prices as component costs rise, and Patel noted that the effect of these price hikes may take anywhere from immediate to a few months to reach customers.