Palo Alto Networks CEO: We need to see the pricing for AI come down

Summary

Palo Alto Networks CEO Nikesh Arora discusses the need for AI token pricing to fall significantly to enable enterprise adoption. He highlights current friction from model companies, expects token efficiency improvements to reduce costs to one-tenth within 12 months, and notes infinite AI demand with insufficient compute. The conversation touches on frontier AI models versus task-specific models.

• Palo Alto Networks reports over 1,000 customer meeting requests for its Frontier AI Defense offering.
• CEO Arora says AI token pricing must decline to about one-tenth for broad enterprise adoption.
• He sees demand for AI consumption as infinite, constrained only by compute availability.
• Frontier AI models will dominate high-stakes research, while task-specific models offer low-cost solutions.
• The discussion follows an interview with OpenAI's Sam Altman on a 54% token efficiency improvement.