Token pricing is a near-term operational barrier for enterprises running agentic and large-context AI workflows, because per-call costs compound rapidly as usage scales. CNBC reports that in an interview on July 9 Nikesh Arora, chairman and CEO of Palo Alto Networks, said token prices should fall to roughly 20% of current levels within 12 months and to about 10% of today's levels (a 90% reduction) the following year, according to CNBC. Media coverage notes Arora welcomed OpenAI's efficiency gain for GPT-5.6, reported as 54% more token-efficient on agentic coding, but called it only "a good start," using his words reported in PYMNTS. Reporting across CNBC, TheNextWeb and Yahoo Finance emphasizes that rising agentic usage and enterprise experiments are driving bills higher even as headline per-token prices fall.
OpenAI Unleashes ChatGPT Work: AI Replaces You, Elites Profit