June 3, 2026

The tokenmaxxing comedown: companies grapple with the AI FOMO premium

Rebecca Moretti
Editorial Lead

From 0 to 100… to 100 to uncertain. The tokenmaxxing era may be ending shortly after it started. For the past year or so, many companies have been in a “move fast and spend tokens” phase. Ramp data shows that businesses' average monthly token spend 13x’d from January 2025 to January 2026, while the average AI contract hit $1M this year. Now, some are realizing they may have moved too fast.

  • After Uber burned through its entire 2026 AI budget in four months, its president and COO Andrew Macdonald questioned whether it was worth it. On the Rapid Response podcast, Macdonald said it’s hard to draw a link between rising Claude spending and rising customer value (think: the shipping of more useful features).
  • Uber is far from the only one: ServiceNow has also reportedly blown through its full-year AI budget, and The Verge reported that Microsoft started canceling direct Claude Code licenses. An AI consultant told Axios that one of their clients spent half a billion dollars in one month after failing to put usage limits on employee Claude licenses. Duolingo said it has backtracked on evaluating AI use in performance reviews.
  • Ramp data shows AI costs spike 50% or more about one in every four months for the biggest spenders. The fastest adopters may be the ones who need to recalibrate fastest.

AI’s trillion-dollar question… Are companies putting the capex cart before the ROI horse? In a rare occurrence, analysts expect Amazon, Meta, and Microsoft will announce negative cashflows in at least one quarter this year because of their hundreds of billions in AI spending. Meanwhile, mass layoffs are partly subsidizing that spend. For the second straight month, AI was the leading stated reason for U.S. job cuts across sectors in April, when tech companies announced 33,361 cuts. Meta alone slashed 8K staff in May. Whether mass layoffs are a result of AI advancements or merely a convenient excuse, they likely won’t come close to offsetting AI expenses. So the question is… will it all pay off?

  • No one really knows. The playing field is new, the data is sparse, and the research is inconclusive. There isn’t yet a proven, standardized way for companies to measure AI ROI today.
  • Complicating this measurement, there are two sides to AI spending: one is internal (e.g. boosting staff productivity) and the other is external (improving a product with AI).

The bottom line:

The risk of being behind outweighs the ROI risk… At least, that appears to be the general consensus. It’s still too early to understand what the results of this massive spend wave and layoffs will be. But companies have been going full-steam ahead because the perceived risk of being left behind in the AI revolution outweighs the risk of being wrong. Another tension is the pressure to be seen as AI-forward in public markets. Still, we’re seeing the beginning of a pullback period in which companies will be much more focused on cost efficiency.

For more stories like this, subscribe to the Leading Indicators newsletter.

Stay ahead of the curve with Leading Indicators
Unsubscribe anytime.
Rebecca Moretti•Editorial Lead
As the Editorial Lead at Ramp, Rebecca focuses on melding original data with financial news to craft engaging stories that highlight leading indicators. Previously, Rebecca was the Senior Editor of Robinhood’s Snacks newsletter, which she helmed for five years, and was Senior Editor of SoFi’s On the Money newsletter.
Ramp is dedicated to helping businesses of all sizes make informed decisions. We adhere to strict editorial guidelines to ensure that our content meets and maintains our high standards.

Don't miss these

Will Big Tech AI splurges pay off? New Ramp data shows paid adoption is up

Big Tech is spending hundreds of billions on AI — but will it pay off? New Ramp data shows paid AI adoption is rising, led by spend on OpenAI and Anthropic.

As AI adoption crosses 50%, the tokenmaxxing economy splits off and up

Here's what the midpoint of the "tokenmaxxing" economy signals.

Salesforce reports earnings as companies layer AI workflows over SaaS systems of record

While incumbents are holding the system of record, several AI-natives are seeing breakout growth in adjacent workflow layers and entirely new categories.