What's happening
Enterprise customers are increasingly moving to constrain AI token consumption and prioritize cost efficiency, according to reporting from CNBC published June 26, 2026. The shift comes as both OpenAI and Anthropic have posted extraordinary revenue acceleration — Anthropic's annualized run rate reached $47 billion in May 2026, compared to roughly $10 billion in total revenue for 2025, while OpenAI was pacing closer to $25 billion in annualized revenue in 2026, up from $13.1 billion generated across all of 2025. Despite this growth, concrete examples of enterprise spending discipline are already emerging: Uber blew through its entire annual AI budget in just four months as of April 2026, subsequently implementing spending tiers on AI tools starting at a base level of $1,500 per month in June 2026. Separately, Lindy CEO Flo Crivello stated the company expects to save millions of dollars within months by switching to cheaper, non-frontier AI models.
The spending rationalization trend is unfolding even as OpenAI and Anthropic filed confidentially for IPOs in early June 2026, with both companies' valuations approaching $1 trillion. Gil Luria, an analyst at D.A. Davidson, noted that some of the largest enterprise customers may start limiting what he described as "out-of-control token spend." Luria added that current growth rates for both companies are "the fastest they will ever be, which is mostly a matter of basic math," and flagged the possibility of a period of spend rationalization ahead. Notably, 95% of enterprise AI usage is still running on frontier models as of June 2026, suggesting the migration to cheaper alternatives remains in early stages.
Why it matters for markets
The financial stakes for Microsoft and Amazon are substantial. Microsoft has invested more than $13 billion in OpenAI, and its Azure cloud platform is a primary distribution channel for OpenAI's models in enterprise settings. Amazon has committed as much as $5 billion to Anthropic, with AWS serving as Anthropic's primary cloud partner. Any deceleration in enterprise token consumption would flow directly into the AI-related revenue lines of both cloud businesses. Microsoft's total revenue stands at $318.27 billion, while Amazon's reaches $742.78 billion, with AWS representing the company's highest-margin segment — making AI workload volumes a meaningful variable in both companies' financial outlooks.
The IPO timing for OpenAI and Anthropic adds another dimension. Both companies filed confidentially for public offerings in early June 2026 with valuations approaching $1 trillion, meaning the trajectory of enterprise AI spending will likely influence how public markets price those offerings — and by extension, the mark-to-market value of Microsoft's and Amazon's stakes. Luria's observation that a period of spend rationalization "may be a blip ahead" introduces uncertainty into near-term revenue projections for both AI developers at a critical juncture in their capital markets activity.
Leadership at both Microsoft and Amazon have publicly acknowledged the cost dimension of AI scaling. Microsoft CEO Satya Nadella stated, "The last thing any of us want is a world where every company across every sector is ceding value to a few models that eat everything they see." Amazon AI executive Peter DeSantis was more direct: "AI has a cost problem. If we ultimately want AI to transform everything, the costs have to be different." These statements reflect an awareness at the highest levels of both organizations that the current cost structure of frontier AI consumption is not indefinitely sustainable at enterprise scale.
Sectors and assets to watch
Microsoft (MSFT) and Amazon (AMZN) are the most directly exposed public-market entities to the described dynamic. Microsoft's Azure platform hosts OpenAI's models for enterprise customers, meaning any broad-based reduction in token consumption would register in Azure's AI services revenue. Amazon Web Services is Anthropic's designated primary cloud provider, positioning AWS similarly as a pass-through for Anthropic's enterprise volume. Both companies have disclosed their investment positions — Microsoft at more than $13 billion into OpenAI and as much as $5 billion into Anthropic, Amazon at as much as $5 billion into Anthropic — creating equity exposure in addition to revenue exposure.
Beyond the two primary tickers, the broader enterprise software and cloud infrastructure sector warrants attention. The fact that 95% of enterprise AI usage remains on frontier models as of June 2026 indicates that any migration toward cheaper or smaller models would represent a structural shift in workload distribution across cloud providers and AI infrastructure vendors. Companies offering lower-cost model alternatives or inference optimization tools could see increased enterprise interest as procurement teams respond to budget pressures of the kind illustrated by Uber's experience of exhausting its annual AI budget within four months.
What to watch next
Key developments to monitor include the public filing and pricing of OpenAI's and Anthropic's IPOs following their confidential submissions in early June 2026, as the disclosed financials will provide the first comprehensive view of how enterprise spending trends are affecting revenue and margin trajectories. Quarterly earnings reports from Microsoft and Amazon will offer data points on whether Azure and AWS AI revenue growth rates are reflecting any deceleration in enterprise token consumption. The pace at which enterprises migrate workloads from frontier models — currently accounting for 95% of enterprise AI usage — to cheaper alternatives will serve as a leading indicator of pricing pressure on OpenAI and Anthropic, and by extension on the cloud platforms that distribute their services. Enterprise procurement policy changes, similar to Uber's introduction of tiered AI spending controls in June 2026, will also signal how broadly cost discipline is spreading across large corporate customers.