Token Limit Exceeded

In April 2026, Uber’s CTO admitted the company had burned through its entire annual artificial intelligence (AI) budget in just four months. Employees had been encouraged to “use AI as much as possible” with internal leaderboards ranking colleagues by usage. By June the decision was made to cap employee’s spending on AI tools such as Claude Code and Cursor at $1,500 per month. Uber’s COO expressed how it is “very hard to draw a line” between token consumption and developing new features consumers care about.

Uber’s U-turn reflects a wider narrative across Enterprise AI adoption from unbridled consumption to sudden restraint. This article explores the nature of the enterprise AI adoption story, the largest end market driving AI infrastructure construction today.

What are ‘tokens’ and why do they matter?

Tokens are the basic unit of AI usage. Every time an AI model reads a prompt, works through a problem, uses a tool or writes a response, it is processing tokens – and each one has a cost. For companies, token spending has become a genuine budget line in its own right, rather than a cost buried within the wider technology budget.

A useful comparison is a metered utility bill. Just as a household pays for each unit of electricity or water it uses, a company using AI pays for each token its systems consume. Use AI sparingly, and the bill is small; deploy it across thousands of employees and automated workflows, and it quickly becomes a meaningful line item on the income statement.

Abridged timeline of recent events:

Nov 2025: Anthropic releases Claude Opus 4.5, taking the lead as state-of-the-art large language model (LLM) for coding and agentic applications. Paired with Claude Code, this triggers a wave of adoption amongst developers.

Q1 2026: Throughout the first few months of the year, companies and users wore token consumption as a badge of pride. Internal token leaderboards ranking users on tokens consumed were popularised as “Tokenmaxxing” took hold.

Q2 2026: Beginning in April, a backlash grew as employees gamed leaderboards, AI spend/token budgets became large enough to warrant more serious scrutiny, and companies became more focused on the return on investment (ROI). Uber, Meta, Amazon and others made headlines for limiting employee spend on tokens. Chief Financial Officers start emphasising efficiency and ROI on AI spending encouraging cheaper open and closed weight models as alternatives to the more expensive ‘frontier’ models.

This framing of events is Anthropic centric, warranted as the greatest beneficiary of this initial wave, reportedly growing annual recurring revenues from $9bn in December 2025 to $47bn by this May and $65bn in July. OpenAI has pivoted focus towards enterprise and shipped GPT-5.5, 5.6 alongside upgrades to its own coding application Codex to catch up and reportedly regaining enterprise traction. Numerous other AI inference infrastructure providers stand to benefit too.

Why does this matter?

The emergent token economy – companies paying to use AI models - underpins a large part of the AI infrastructure build out today. How much revenue can be generated from renting out computing power and selling tokens is a critical component of the ROI of a data centre. In this wave a lot of value has been captured by the current frontier labs (Anthropic, OpenAI) but many companies are fighting to earn their share.

Enterprise AI adoption

Currently software engineering remains the largest user of tokens. But workflows are not limited to software engineers: AI agents and workflows are being trialled and used in productions across nearly all business departments, automating and expanding processes from legal, marketing, sales etc.

With most enterprises still in the early stages of adoption there is bound to be wastage and rationalisation. Initial programs are full of trial and error and overspending. For that spending to be justified over time though, companies need to find real benefits to specific business KPIs, such as cost reductions or revenue acceleration.

One useful data source tracking corporate spending on AI is the Ramp AI Index: https://ramp.com/data/ai-index

Ramp provides a corporate spend management platform for businesses, including corporate cards, expense management, bill payments and other services. It tracks how its roughly 70,000 customers spend on AI products and services.

The charts below shows the USD monthly spend per employee for the median, top 10% and top 1% of businesses:

The data shows the sharp increase this year, though it does not yet show a deterioration of spend in recent months. The top 1% of businesses spend above $7,000 per employee per month on AI, compared with just the $12 for the median spender. The other stark difference is in the 600x difference between the top spenders and the median. This highlights that today AI spending is heavily weighted to power users/enterprises and lacks breadth.

The dataset has its limitations. Ramp’s customers tend to skew towards smaller and mid-sizes businesses, so it is fair to ask how representative the sample is. Corroborating this data are a range of CIO surveys and other industry research showing similar trends at the largest enterprises.

On 5 August, Uber’s CTO posted on X “We’re seeing some very interesting trends on AI costs. I think it’s another signal that we’re coming to the end of the so-called ‘tokenmaxxing’ era”. He went on to describe how Uber has expanded AI adoption across engineers whilst prioritising efficiencies and optimisations in order to continue scaling AI internally - a positive signal that enterprises are progressing along the adoption curve.

Conclusion

The shift from ‘tokenmaxxing’ to token discipline does not, in itself, signal weaker demand for AI – if anything, it points to enterprise adoption maturing. As companies move from experimentation towards production use, spending should become more sustainable and better tied to measurable business outcomes, a healthier foundation for the AI infrastructure buildout than unchecked consumption.

For investors, the direction of enterprise AI spending – not just its headline growth rate – is likely to remain a key signal to watch across the AI investment chain, from the model providers themselves to the infrastructure that supports them.

Important Information

The value of investments and the income derived from them can fall as well as rise, and past performance is no indicator of future performance. Investment products may be subject to investment risks involving, but not limited to, possible loss of all or part of the principal invested.

This document does not constitute and shall not be construed as a prospectus, advertisement, public offering or placement of, nor a recommendation to buy, sell, hold or solicit, any investment, security, other financial instrument or other product or service. It is not intended to be a final representation of the terms and conditions of any investment, security, other financial instrument or other product or service. This document is for general information only and is not intended as investment advice or any other specific recommendation as to any particular course of action or inaction. The information in this document does not take into account the specific investment objectives, financial situation or particular needs of the recipient. You should seek your own professional advice suitable to your particular circumstances prior to making any investment or if you are in doubt as to the information in this document.

Although information in this document has been obtained from sources believed to be reliable, no member of the EFG group represents or warrants its accuracy, and such information may be incomplete or condensed. Any opinions in this document are subject to change without notice. This document may contain personal opinions which do not necessarily reflect the position of any member of the EFG group. To the fullest extent permissible by law, no member of the EFG group shall be responsible for the consequences of any errors or omissions herein, or reliance upon any opinion or statement contained herein, and each member of the EFG group expressly disclaims any liability, including (without limitation) liability for incidental or consequential damages, arising from the same or resulting from any action or inaction on the part of the recipient in reliance on this document.
The availability of this document in any jurisdiction or country may be contrary to local law or regulation and persons who come into possession of this document should inform themselves of and observe any restrictions. This document may not be reproduced, disclosed or distributed (in whole or in part) to any other person without prior written permission from an authorised member of the EFG group.

This document has been produced by EFG Asset Management (UK) Limited for use by the EFG group and the worldwide subsidiaries and affiliates within the EFG group. EFG Asset Management (UK) Limited is authorised and regulated by the UK Financial Conduct Authority, registered no. 7389746. Registered address: EFG Asset Management (UK) Limited, Park House, 116 Park Street, London W1K 6AP, United Kingdom, telephone +44 (0)20 7491 9111.