Palo Alto Networks CEO Says AI Token Prices Must Fall 90%

Palo Alto Networks CEO Nikesh Arora told CNBC that AI token prices need to fall as much as 90% before enterprise adoption can scale. He said token costs should reach roughly one-fifth of current levels within the next 12 months and drop 90% by the following year.

Arora's remarks followed OpenAI's release of its GPT-5.6 family, which the company described as 54% more token-efficient on agentic coding. Arora called the gain a good start but said another round of improvements would be needed.

Arora joins other executives who have flagged rising token costs as a barrier. Companies have shifted toward cheaper open-weight models, including some from Chinese providers, according to usage data cited in the coverage. Meta launched Muse Spark 1.1 on the same day, pricing input tokens at $1.25 per million and output tokens at $4.25 per million.

The company also began offering new accounts $20 in free credits.

Amazon raised $25 billion in debt this week to fund AI infrastructure. SK Hynix completed a $26.5 billion U.S. listing, the largest by a foreign company, with shares opening 14% above the offer price. Analyses from JPMorgan, Gartner, and Apollo have noted that falling token prices may not reduce total enterprise spending if usage grows faster than costs decline.