Some companies are beginning to confront the operational economics behind Anthropic’s Claude platform as enterprise usage scales deeper into many CFOs’ finance functions.
Recent reports involving Microsoft and Uber have highlighted growing concerns around enterprise AI consumption costs tied to agentic coding and workflow tools. Microsoft’s issue began when it reportedly began winding down portions of its internal Claude Code licenses after token-based usage costs climbed sharply, and Uber said its newest technology challenge was that it exhausted its 2026 AI budget by April after deploying Claude Code across thousands of engineers.
The developments are drawing attention to a broader finance challenge emerging around enterprise artificial intelligence. Unlike traditional software-as-a-service contracts built around fixed licensing models, Claude’s pricing structure increasingly resembles infrastructure consumption pricing that is similar to that of a utility. It’s based on usage, runtime activity and tool execution.
Earlier this year, OpenAI CEO Sam Altman said he believes AI will eventually be treated more like electricity or water, with companies paying for intelligence based on usage demand rather than flat subscription access. In its enterprise offerings, OpenAI has also increasingly expanded token-based and credit-based enterprise pricing tied to usage consumption and model access.
Anthropic’s pricing structure reflects a similar shift toward consumption-based enterprise AI pricing. The company’s pricing documentation outlines separate charges for input tokens, output tokens, cache writes, cache reads, regional inference routing and agent runtime sessions, alongside additional usage costs tied to tools like web search and code execution. The company also notes that its newer Opus 4.7 tokenizer may consume up to 35% more tokens for the same text compared to earlier models.
The pricing complexity is beginning to raise broader questions around how companies measure the value generated from rising AI usage. Uber COO Andrew Macdonald recently questioned how directly companies could connect rising AI coding-tool usage to measurable business output. “That link is not there yet,” Macdonald said during an appearance on the Rapid Response podcast while discussing whether increased Claude Code usage was translating into materially higher levels of customer-facing product development.
Macdonald added that organizations may increasingly need to evaluate “token consumption and the associated cost versus headcount” as enterprise AI usage scales, a strategy that is known among techies as “tokenmaxxing.”
“If you’re not actually able to draw a direct line to how much useful features and functionality you’re shipping to your users, that trade becomes harder to justify [because AI is] not free,” he said.
Finance teams build controls around AI spending
Anthropic CFO Krishna Rao argued during a podcast episode uploaded May 13 that companies continue increasing AI spending because newer Claude models are driving broader operational usage across enterprises. “Customers see that, and then they invest really heavily in more tokens with the newer models,” Rao said, adding that “the returns to frontier intelligence are extremely high,” particularly in enterprise environments where AI systems are increasingly embedded into operational workflows.
Rao also described how Anthropic’s own finance organization has integrated Claude into accounting and reporting work internally. According to Rao, Anthropic now has “a library of skills for Claude that are specific to finance,” while the company also uses Claude to help produce statutory financial statements across its legal entities.
“What used to take hours to produce … now comes down to 30 minutes,” Rao said. Rao added that Anthropic’s net dollar retention rate now exceeds 500% on an annualized basis, reflecting how existing enterprise customers continue increasing spending after adopting the platform.
The usage patterns reflect how quickly Claude has expanded into operational finance work. Top accounting firms and financial institutions have increasingly integrated Claude into forecasting, reporting, audit preparation, reconciliations and finance workflow automation in recent months.
Pressure to accelerate enterprise AI adoption is also intensifying across professional services. A recent Financial Times report said KPMG executives were meeting regularly with Silicon Valley AI startups as the firm looks to avoid disruption tied to artificial intelligence. KPMG US CEO Tim Walsh told FT, “Regardless of how fast we’re moving, I’m confident it’s not fast enough.”
KPMG leadership has also emphasized a growing willingness to partner externally on AI infrastructure rather than building every system internally. In an exclusive interview with CFO.com last year after his appointment, KPMG Deputy Chair and U.S. Managing Principal Atif Zaim said the firm uses a mix of internal development and technology partnerships as AI adoption accelerates. “We’ve never felt we must invent everything to feel like we’re innovating,” Zaim said.
The growing complexity is beginning to reshape how finance teams govern AI usage internally as Claude adoption expands across operational finance work. A recent Reddit discussion among unverified FP&A professionals described organizations implementing spending caps, monitoring dashboards and internal usage reviews tied to Claude Enterprise deployments. One commenter said their organization checks Claude costs daily because the platform has become “one of our largest variable costs in the business right now.”
Other finance users described employees individually consuming between $100 and $300 per month in Claude usage, while engineering-heavy users reached several thousand dollars monthly through token-intensive coding and data workflows. The discussion reflected growing concern over how finance leaders will control Claude spending as usage expands across organizations.
Some CFOs have taken to LinkedIn to publicly raise concerns around enterprise implementation, billing structures and customer support experiences as organizations scale adoption internally as well.
AI investment rises as finance teams seek measurable returns
The growing scrutiny around Claude pricing arrives as finance organizations continue accelerating AI spending across operations. A recent Bain & Co. survey found that more than half of CFOs expect enterprise AI spending to rise by at least 15% over the next year, while three-quarters of finance department leaders said they expect AI budgets inside finance itself to increase.
Finance leaders are also embedding AI more deeply into pricing strategy and operational decision-making. A March report from Gartner, covered by CFO Dive, found companies increasingly using AI-driven pricing models to respond to affordability concerns and margin pressure as AI tools move deeper into finance operations.
At the same time, finance leaders are still trying to determine how AI value should actually be measured. A recent CFO.com opinion piece by Alexander D. Hilton, a digital strategy and AI consultant at Agile New England, an ACM Chapter, argued that many enterprise AI business cases still rely on “the wrong accounting logic,” prioritizing local productivity gains and time savings over broader financial throughput.
Hilton also cited research showing more than 80% of firms reported no measurable productivity or employment impact from AI over the past three years despite rapidly growing investment levels.
The growing attention around Claude pricing reflects how quickly AI adoption is moving. As usage expands across finance functions, CFOs may increasingly face questions from CEOs, COOs, CTOs, boards and stakeholders around monitoring AI consumption and determining whether rising enterprise usage is generating real, measurable value.
