Amazon just shut down its AI leaderboard tracking internal token usage. The gamification was driving more AI-powered tasks but fewer useful results. “Please don’t use AI just for the sake of using AI,” the Amazon SVP instructed his staff.
Amazon is not alone. Uber blew past its 2026 artificial intelligence coding budget in just four months. Google’s CEO, Sundar Pichai, revealed that the company’s token usage has grown sevenfold in a year. Many companies including Meta, Microsoft and Salesforce are reportedly pushing to limit token usage.
It’s unsurprising what happens when you set the wrong incentives: You get the wrong results. You get what you measure.
Hype, these days, is invariably accompanied by jargon. In an attempt to demonstrate corporate progressiveness, boardrooms and C-suites across America scramble to keep up with modern phraseology. They’re throwing around terms like tokens per query, cost per inference, GPU hours, and even model utilization. “Tokens are the new oil for the enterprise” is the latest slogan; tokens are apparently the new measure for AI adoption and productivity, the proof of AI discipline, the real unit for driving AI ROI. “Tokenmaxxing” has topped the charts for weeks.
It all sounds smart, but it’s the wrong conversation.
While companies get better at measuring AI spend, many still have no idea whether their AI investments are driving increased revenue, creating faster decisions, reducing friction, or creating any tangible and measurable advantage at all. With all the token math, they know what the intelligence costs, but not whether the intelligence is useful. And, very quickly, the AI ecosystem is running into unsustainable economics.
The dashboards are getting better; returns are not. Word counts and lines of code are proliferating, but none of that trickles down to the bottom line.
Cost Obsession is Short-sighted
After a decade of cloud overspending, finance leaders are now homing in on everything AI. They’re tracking, to the cent, cost per workload, cost per transaction and utilization rates.
The truth? AI is expensive. Deloitte reports that AI is one of the fastest-growing investments in technology budgets. And, AI inference costs alone are consuming more and more of organizational budgets. But, you cannot confuse costs of managing AI with the value it brings. That is where you lose sight of the potential of the technology.
McKinsey found 64% of organizational leaders believe AI is enabling their innovation, but only 39% can show any impact on operational earnings at the enterprise level. Other studies show basically the same results: heavy investment, limited return, massive experimentation, little in the way of scale.
Basically, companies are becoming increasingly sophisticated about spending, but are still uncertain about what they are getting in return.
Cloud Spending is Not a Comparable Model
With cloud economics, leaders basically believed that spend and return move in tandem. More business activity meant more compute, more IT spend. Not always a perfect map, but the logic was intuitive. AI is a different beast.
A small number of tokens can create an insight that helps close a major account, resolve a high-risk customer issue, or speed up an impactful decision. In contrast, there will be times when a much larger token bill produces mediocre output of no use—‘AI slop’ loves tokens. Same category of spend, but a totally different result.
The question for leaders becomes: What are you optimizing around?
Tokenomics determines efficiency of spend, but are you addressing the willingness to spend in the first place?
Are you looking at how many tokens you can afford, when you should be looking at how this new system is going to help you make consequential decisions?
Are you focused on how your tokens are billed, when you should be looking at how you can execute better, faster actions at scale?
When AI leaders overemphasize token consumption, they end up optimizing around the path of least resistance, not the path of greatest value. That’s at the center of the AI scaling challenge we see today.
Measure AI Like This
The companies that show the most maturity in AI are those that separate the cost of intelligence from the value of intelligence. The first category is tokenomics. This is where financial discipline belongs. Companies should absolutely manage model mix, caching, batching, routing, infrastructure choices, and vendor economics to know exactly what AI costs and where waste is coming from. It’s the fine-tuning of the cost of adoption.
But, the second category is where ROI exists, and is missing in too many organizations. This is where leading organizations ask a series of questions around their AI investment:
- Did AI reduce cost per resolved customer issue?
- Did it speed up lead qualification or shorten sales cycles?
- Did it improve throughput in a high-friction workflow?
- Did it reduce human hours?
- Did it expand margins in a specific process or function?
- Did it help the company move faster than competitors in meaningful ways?
These are the questions that determine whether AI is actually working. The question needing to be asked is not, “How do we reduce token spend?” but, “For every dollar we spend on intelligence, how much business value are we creating?” It is holistic business value, business case fundamentals, and it represents the heart of adoption. Executives pursuing AI ROI need to begin with questions of value, results, and impact, not cost.
What business outcomes can we directly attribute to AI spend? Which deployments are becoming more valuable over time, not simply cheaper? Where are we overmanaging costs in ways that suppress performance? What proprietary advantage is this investment building in data, workflow, or execution? If token prices rise or fall sharply, what changes for us strategically and what doesn’t?
This line of questioning shifts leaders’ thinking to look at AI as a business system, not simply a technology tool.
What really matters
Tokenomics matters, but don’t let it distract you from the bigger question of whether AI is worth the investment of revenue and time for your use cases. The companies that win will not be the ones who save every penny possible from their AI bills. They will be the ones who use AI to enable faster decisions and lower friction, and create better customer outcomes, all which will compound over time.
You don’t want to be an organization with AI leader charts and magnificent dashboards, and perfectly optimized spending on AI tools that have driven a net zero return. Be the organization that uses AI to its fullest potential, as an overarching business system, delivering tangible stakeholder value.
