Palo Alto Networks CEO Says AI Token Costs Must Fall Up to 90%

Palo Alto Networks CEO Nikesh Arora said on CNBC on Thursday that AI token costs need to fall as much as 90% to support large-scale enterprise adoption. He called OpenAI CEO Sam Altman’s claim that the lab’s newest model was 54% more token-efficient on agentic coding tasks “a good start,” and said prices need to keep dropping, by as much as 20% over the next year and 90% the year after that. The remarks put Arora among a growing group of executives pressing frontier labs on price, after Palantir CEO Alex Karp said the previous week that “something has gone completely wrong” with the token model.

On the network’s “Squawk Box,” Karp criticized the token model used by Anthropic and OpenAI and described open-weight models as a possible solution. “The basic view among enterprises in this country is I’m going to chillax and waste my time with tokens,” Karp said. CNBC reported that high costs were leading businesses toward cheaper open-weight tools, including Chinese models.

After announcing its intent on June 3, the Linux Foundation launched the Tokenomics Foundation at the FinOps X 2026 conference, where it said the new group would work with the FinOps Foundation on open standards for managing enterprise AI costs. J.R. Storment, executive director of the FinOps Foundation, called it “a vendor-neutral home” for questions about token economics. “This is an urgent need for these giant consumers,” he said during the conference keynote. The Tokenomics Foundation plans to bring enterprises, hyperscalers and frontier model developers into the standards effort, and says tokens are the most easily metered part of broader AI spending. CIO Dive reported that Oracle started a limited rollout of token bundles intended to make AI spending more predictable and that AWS opened a public preview of its AWS FinOps Agent to provide visibility into enterprise AI cost anomalies. Storment said companies want clear financial controls for AI use.

TechCrunch reported that Sequoia partner David Cahn estimates 2026 AI infrastructure spending at about $1.5 trillion and calculates that the industry needs roughly $3 trillion in revenue to justify the chips and data-center expenditures. The same report said Anthropic is thought to have reached about $60 billion in annual recurring revenue, while OpenAI reportedly earned about $13 billion over the 2025 calendar year.

That account also cited a note from Apollo chief economist Torsten Slok, who focused on Amazon, Google, Meta and Microsoft and their projected increases in free cash flow in 2028. If that payoff arrives more slowly, Slok warned, it “would risk tipping the economy into recession and the S&P 500 into a correction.”

Benedict Evans wrote Thursday that the market forces he could see pointed toward frontier models becoming low-margin commodity infrastructure as the supply shortage eases. He cited reports putting current inference gross margins at roughly 40% to 50%, including server depreciation but excluding the cost of training new models. RBC Wealth Management’s Tyler Frawley wrote that companies using AI may capture more value than model and compute sellers. Frawley cited an X post in which Coinbase CEO Brian Armstrong said routing suitable work to cheaper models, while reserving frontier systems for harder tasks, had kept Coinbase’s AI spending roughly flat as token use increased.

But Arora also told the network that demand “continues to be infinite” and predicted that “all these things will rationalize over time” as the technology becomes more efficient and budgets adjust. Altman, meanwhile, linked the claimed 54% token-efficiency gain to enterprise concerns about spending and value.

Storment will be responsible for hiring the Tokenomics Foundation team focused on managing AI costs.

SpaceX Buys Cursor and Puts AI Coding Inside the xAI Stack

SpaceX's June 16 merger filing says X67 Inc., a wholly owned subsidiary, will merge into Anysphere, with Cursor surviving as a SpaceX unit. Reuters reported that the all-stock deal values the AI codin

The Implicator

OpenAI Weighs Drastic Token Price Cuts to Blunt Anthropic's Enterprise Run

OpenAI is weighing significant cuts to what it charges for tokens, the metered unit AI companies use to bill for their products, in anticipation of similar cuts it expects at Anthropic, the Wall Stree

The Implicator

KPMG Survey Shows Only 26% of Companies Can Track AI Costs Fully

A new KPMG survey reported by The Wall Street Journal found that only 26% of companies fully track AI costs. It put 50% at partial visibility and 22% at little or none until billing, as vendors meter

The Implicator

Marcus Schuler

San Francisco

Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: editor@implicator.ai