'We need to see the pricing for AI come down': Palo Alto Networks CEO Nikesh Arora says AI is too expensive — and needs to fall 90% to become affordable

• Palo Alto Networks CEO Nikesh Arora says 90% reduction in AI pricing is needed to drive widespread adoption, 20% reduction by next year
• OpenAI's latest GPT-5.6 Sol model is 54% more efficient across agentic coding tasks
• Shift to consumption-based models is making high costs even more pronounced

Palo Alto Networks CEO Nikesh Arora has warned that AI pricing is still too high for widespread enterprise adoption, arguing that it should be drastically cheaper.

Although model efficiency and performance have come a long way in recent years, he believes cost remains one of the biggest barriers for widespread adoption.

In a recent interview with CNBC, Arora laid out how he believes token prices need to fall by as much as 90%, however he acknowledges that the change is unlikely to happen immediately, instead welcoming a 20% price reduction over the next year or so.

Is AI pricing holding enterprises back from widespread adoption?

Arora's comments come in response to OpenAI's latest announcement, revealing that its latest GPT-5.6 Sol model is now 54% mode token-efficient across agentic coding tasks – a major improvement.

Though Palo Alto Networks' CEO admitted that this in itself is a good start, it's not enough of an improvement to drastically change pricing for major enterprises at the moment.

"It's important to understand the demand continues to be infinite," he added, implying that cost remains prohibitive. The issue of cost is also becoming more of an issue as AI companies evolve their pricing strategies, with customers moving toward consumption-based models rather than flat monthly fees per user.

Importantly, Arora's vision for cheaper AI isn't unfounded. Virtually all previous technological revolutions have already followed a similar pattern. In the case of AI, compute would become cheaper, models would become more efficient and competition could keep prices low, all working in the favor of enterprise customers.