HP urges businesses to run AI locally on premium PCs
HP says businesses concerned about rising cloud AI token costs should run more AI workloads locally on higher-specification PCs. The company argues this can reduce operating costs, improve response times, lessen reliance on internet connections and keep sensitive data closer to where it is created, though the shift also benefits HP by encouraging sales of more expensive machines and related services.
AI-capable PCs represented 46 per cent of HP’s PC sales mix in the quarter, and the company expects this to reach 60–70 per cent in fiscal 2027 and more than 70 per cent in 2028. Personal Systems revenue rose 18 per cent year on year to $11.8 billion despite a 16 per cent fall in PC volumes, driven by higher prices and a greater share of premium products; HP also expects further price rises as component costs increase.
- HP promotes costly AI PCs as an alternative to cloud token spending.
- Premium machines helped revenue rise despite lower PC sales volumes.
- HP expects AI PCs and customer prices to keep increasing.
New here? Start with this
Businesses can use AI through online services, which process requests in remote data centres and may charge according to usage. Running some AI tasks directly on a computer instead requires a machine with enough processing power and memory, and can reduce the need to send data over the internet.
HP makes computers and related services, and is encouraging companies to choose higher-specification PCs for this work. The approach may appeal to businesses seeking to control service costs or keep sensitive information closer to its source, while also creating a market for more expensive machines.
The shift is part of a wider change in how companies decide where AI tasks should run: on local devices, in data centres, or across both. The choice depends on factors such as cost, speed, internet access and how an organisation handles its data.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Cloud AI costs are escalating rapidly and genuinely concern many businesses. Local processing offers real technical advantages: reduced latency for time-sensitive applications, stronger data sovereignty for regulated industries, and freedom from cloud provider dependency and internet connectivity requirements. For organisations with consistent, predictable AI workloads, shifting processing locally can deliver better total cost of ownership despite higher initial hardware investment whilst improving operational resilience.
The case against
Cloud solutions remain superior for most organisations because they achieve economies of scale, enable access to the most capable models without massive capital expenditure, and allow businesses to pay only for what they use. Compelling organisations to purchase premium hardware substitutes substantial upfront costs for ongoing expenses, and most lack the technical expertise to manage distributed AI infrastructure, maintain security patches across numerous devices, or optimise local models effectively. HP's pronounced financial incentive to sell expensive machines—evident in the company's shifting sales mix and rising prices—raises legitimate questions about whether this advice genuinely serves business interests, particularly smaller enterprises with limited IT resources.
Read the full article at the source →
Originally published by The Register as “HP has a solution to expensive AI tokens: Buy a more expensive PC”.