HP Inc. emphasized the growing shift toward local AI inference on devices at The Six Five Summit’s AI Unleashed 2026 event, citing significant cost savings and operational efficiencies for enterprises. Speaking on Wednesday, Ketan Patel, president of Personal Systems at HP, highlighted how organizations are increasingly deploying AI agents to reduce reliance on cloud-based models, which have driven expenses sharply higher.
One HP customer’s cloud AI spending rose from $20 million in the prior year to between $200 million and $300 million in the current year, Patel noted, framing the shift as a critical financial consideration for chief financial officers. Local inference on basic models can save an estimated $650,000 per 1,000 employees, while advanced local deployments may deliver up to $2 million in savings per 1,000 employees, according to HP’s calculations.
The company outlined technical capabilities enabling this transition, including current laptops supporting open-source models with up to 20 billion parameters and next-generation systems capable of hosting up to 120 billion parameters locally. On-premises infrastructure can manage models with up to 1 trillion parameters. Patel described PCs as evolving into systems that understand context and support AI agents, with nearly 72% of organizations either deploying or piloting such agents.
Analysts from Signal65, Ryan and Nick, underscored the trend, noting that enterprise requests for proposals increasingly prioritize local AI execution for both security and cost reasons. The agent-based AI market is projected to reach a $6 trillion total addressable market by 2028, reflecting broader industry momentum toward hybrid AI architectures.
HP’s stock closed at $28.58 on August 26, down 3.80%, though it has delivered a 36% gain year-to-date and a 61% return over the past six months. Revenue growth for the period was reported at 5.74%.
