Which companies reportedly overshot their annual AI budgets?

Uber is one of the companies that reportedly blew through its annual AI budget in a matter of months.

What did Meta do with its internal AI usage leaderboard?

Meta quietly killed its internal AI usage leaderboard, indicating a shift in approach towards AI.

Why is there a new challenge for enterprise leaders regarding AI spending?

The challenge arises from the need to prove the return on investment of AI spending, as not every experiment delivers business value.

Enterprises Struggle to Justify AI Spending After Tokenmaxxing Trend

Q: What was the unofficial motto of Silicon Valley earlier this year regarding AI usage?

The unofficial motto was 'tokenmaxxing', where companies encouraged employees to push generative AI usage as far as it would go without worrying about cost.

The unofficial motto of Silicon Valley earlier this year led to excessive AI usage, but now enterprises are realizing not every experiment delivers business value. Measuring ROI is the new challenge.

The recent trend of tokenmaxxing, where companies encouraged employees to push generative AI usage as far as it would go without worrying about cost, has come to an end. The bill for this enthusiasm is now due, and many enterprises are struggling to measure and justify their AI spending.

What Happened

Tokenmaxxing was the unofficial motto of Silicon Valley earlier this year, with CEOs from startups to Fortune 500 companies encouraging employees to experiment aggressively with generative AI. The message was simple: push AI usage as far as it would go, learn fast, and don't worry about the cost. However, this enthusiasm has now given way to a more cautious approach.

Uber reportedly blew through its annual AI budget in a matter of months, while several companies cut back on Claude licenses for parts of their organization. Meta quietly killed its internal AI usage leaderboard, indicating that the tension between enthusiasm for AI and the reality of its cost has created a new challenge for enterprise leaders: proving the return on investment.

Background and Context

New Enterprise Associates (NEA) partner Tiffany Luck recently addressed this shift in her conversation with industry analysts. Luck noted that many enterprises are still in the early stages of understanding how to measure and justify their AI spending. "We went through a phase where companies were throwing tokens at every possible use case," Luck said. "Now they're realizing that not every experiment delivers business value. The next phase is about disciplined deployment and measurement."

The trend is not isolated to a few companies. Industry reports from late 2024 and early 2025 indicate that enterprise AI spending grew by over 200% year-over-year, but only a fraction of those investments have translated into measurable productivity gains or revenue growth.

Why It Matters to the Industry

The measurement gap is a key problem for many enterprises. Traditional metrics like cost per transaction or time saved are often inadequate for evaluating AI tools that affect workflows, decision-making, and customer experience in less linear ways. "You can't just look at the cost of API calls and decide if it's worth it," Luck explained. "You need to understand how the tool changes the work itself. That's harder to measure, but it's essential."

This shift in approach has significant implications for companies currently evaluating or expanding their AI investments. The message is clear: budget without strategy leads to waste. The tokenmaxxing era may have accelerated adoption, but it also created a hangover that many enterprises are now struggling to overcome.

What Comes Next

The next phase of AI adoption will require a more disciplined approach to deployment and measurement. Enterprises must develop internal frameworks to track AI ROI effectively, moving beyond traditional metrics to evaluate the impact of AI on workflows, decision-making, and customer experience.

This shift will also require companies to invest in tools and technologies that can help them measure and justify their AI spending. The market for these solutions is growing rapidly, with startups stepping in to help enterprises track return on AI spend.

Key Facts

Tokenmaxxing was the unofficial motto of Silicon Valley earlier this year, with CEOs encouraging employees to push generative AI usage as far as it would go without worrying about cost.
Uber reportedly blew through its annual AI budget in a matter of months, while several companies cut back on Claude licenses for parts of their organization.
Meta quietly killed its internal AI usage leaderboard, indicating that the tension between enthusiasm for AI and the reality of its cost has created a new challenge for enterprise leaders: proving the return on investment.
New Enterprise Associates (NEA) partner Tiffany Luck noted that many enterprises are still in the early stages of understanding how to measure and justify their AI spending.
Industry reports indicate that enterprise AI spending grew by over 200% year-over-year, but only a fraction of those investments have translated into measurable productivity gains or revenue growth.

The shift towards a more disciplined approach to AI adoption will require companies to invest in tools and technologies that can help them measure and justify their AI spending. As the market for these solutions continues to grow, one thing is clear: the era of tokenmaxxing is behind us, and a new reality has set in.