Nikesh Arora, the CEO of Palo Alto Networks has weighed in on OpenAI’s latest announcement that its GPT-5.6 Sol model is 54% more token-efficient for agentic coding. Speaking in an interview with CNBC’s Seema Mody on Squawk on the Street, Arora said that the improvement was ‘a good start’ but warned that far deeper cost reductions are needed to make AI adoption viable at scale. Arora is of the point of view that token efficiency must drop to 20% within the next 12 months and as much as 90% by the following year. Rising token costs, he said, are straining enterprise budgets and making AI tools increasingly difficult to implement. “We need to see the pricing for AI come down,” he emphasized, noting that high costs are a major barrier to widespread adoption.

The growing executive pressure

Arora has joined the growing chorus of industry leaders calling for reform. Last week, Palantir CEO Alex Karp criticised the token model used by OpenAI and Anthropic, suggesting open-weight models could be a solution. Karp told CNBC: “Something has gone completely wrong. The basic view among enterprises is I’m going to chillax and waste my time with tokens.” This frustration has led many businesses to experiment with cheaper open?weight tools, including models from China that are rapidly closing the performance gap with US labs.

Rising AI spending

At the same time, AI infrastructure spending continues to surge. SpaceX raised $25 billion last month in a bond sale, while Amazon raised another $25 billion in debt this week to fund AI expansion. Arora suggested that while demand for AI remains “infinite,” budgets will eventually rationalise as technology becomes more efficient.

Nikesh Arora told CEOs that layoffs is not the solution

Recently, Nikesh Arora issued a stark warning about the future of work in the AI era after telling CEOs that layoffs is not the solution. Recently, Arora said that massive job cuts are not the solution to the industry’s current talent shortage, and has advised opting for a gradual, specialised hiring strategy to build a future-proof workforce. Now speaking on the 20VC podcast (via Fortune), Arora said that 90% of employees at large companies lack AI fluency, a gap he believes could determine career survival. “I think we’re back to a Darwinian moment where everybody has to figure out who’s really good,” he explained, stressing that workers must learn independently as no formal university courses exist to prepare them.The comments made by Arora come amid widespread layoffs tied to AI adoption. A 2025 Orgvue study found that 39% of employers have already made redundancies due to AI, while companies like Coinbase, Block, and Cloudflare have slashed thousands of jobs. Arora contrasted his approach with CEOs such as Brian Armstrong and Jack Dorsey, who cut 30–40% of staff, saying: “They’ve figured out there’s no redemption. ‘I can’t train these people, I’m going to just find the people who are going to come in and help me do this stuff.’”At Palo Alto Networks, Arora is letting natural attrition—about 2% of employees leaving each month—reshape the workforce. He replaces departing staff with AI-fluent talent recruited through hackathons, aiming to transform 20–25% of his team within a year and achieve majority AI literacy in three years.