Tokenomics Is the Budget. Humanomics Is the Bound.

About the Author

Rajeev Ronanki

Chief Executive Officer

Show bio ↓

Tokenomics is what you can compute, and it scales on its own. Humanomics is deciding what is worth computing, and deciding what tasks remain actionable by humans.

By April 2026, Uber had already blown through its entire AI coding budget for the year. The company pushed heavy adoption through the engineering org, from a third of engineers in February to 84% a month later; driven in part by company-created leaderboards that ranked people on how much AI they used. When the token bill came, so did the realization. By June the company decided to cap spend at $1,500 a month per employee, per tool, just to slow the accelerated burn.

In June, the Linux Foundation gave this growing challenge a name, the Tokenomics Foundation. A standards body meant to bring the same cost discipline to AI tokens that FinOps once brought to cloud spend. One of its organizers, J.R. Storment, told TechCrunch that companies had begun reporting that they were already as much as three times over their entire 2026 token budget, with eight months still on the calendar.

This is tokenomics. It measures what you can compute, and quickly scales the moment you let it. Compute doesn’t wait for a business case, nor does it ask for permission. It just runs and the bill arrives later, often well after the original approval was given for it.

Of note is figure below, which points in two directions at once. Per-token prices continue to drop, which reads like relief on a pricing page. However, the volume tells a different story. Those numbers, from EY, put the rise at roughly 30x, and the chart below shows why. In the previously cited TechCrunch article, Jellyfish's head of research, Nicholas Arcolano, shared that spend per developer climbed a whopping 18.6 times in nine months.

The price of thinking falls while the cost of thinking rises, and a finance team reading only the rate card never sees the turn coming.

On the left is the sticker price of raw compute. When GPT-4 launched in March 2023, OpenAI listed it at $30 per million input tokens; by April 2026 Ramp's benchmark data put the blended average businesses actually pay across models near $0.72. On the right, is the true cost of work. EY's analysis of agentic AI costs found a customer interaction that ran 4 cents in 2023 now hits at $1.20, inclusive of a 2026 agentic workflow—once tools, reasoning, and self-checking loops enter the bill. The price of AI tokens dropped 97% while the cost of task became 30 times more expensive. To often, a finance team reading only the rate card sees half the picture.

This is where most of the conversation stops, and the challenge rises. Tokenomics is an accounting question that has an owner, a dashboard, and a foundation with its name on the door. Humanomics asks what is worth computing at all, and where the line falls between the work a machine should carry, and the judgment that stays with a person.

A budget is a number you are cleared to spend. A bound is clarification on what your company stands for, and which of its decisions should never go to a system tuned for the wrong objective. It’s not a contest between the cost of people and machines. Rather, the bound is the discipline of aiming both at the same outcome, so the compute you can afford goes to the results you actually want.

You can hold a huge budget and no bound at all. That is more or less how a company spends $500 million in a month on AI, without noticing.

The companies selling the tokens have started saying this out loud. In June, Sam Altman told CNBC that whether AI spending ever pays for itself is the most fair criticism of AI going right now; and that customer worry about cost is the second most common thing he focuses on. PwC’s April 2026 AI Performance Study put numbers on the split: roughly three-quarters of the measured economic return from AI had pooled into about a fifth of companies, and everyone divided what remained.

That leaderboard logic, of ranking on tokenization of AI, or tokenmaxxing, did not stay at Uber. Plenty of companies made that same meter the goal, believing that more AI usage meant more output and benefit. It appears to be easy to measure, which is a great deal of the appeal, yet it tells you almost nothing about whether the work was worth doing. Arcolano put it plainly: whether heavy [AI] spend pays off comes down to the business value of what actually ships, and that’s something that companies are not measuring.

Jellyfish's own data shows why it matters. Its heaviest token users were roughly twice as productive as lighter ones, yet they spent ten times the amount of tokens to get there. The leaderboard measured effort, while noone measured whether the effort added up to anything.

The bound rests on the one input that does not get cheaper as compute does: human judgment that’s spent at the moment a decision gets made. And that judgment is under pressure from the very tools meant to support it. A World Economic Forum essay in June described “human in the loop” as a phrase we repeat for reassurance, without asking whether the person in that loop can actually overrule a confident, wrong machine. One example cited is a bank manager in Kuala Lumpur named Diana. Her name being on every lending call the model makes, and who interrogates its output rather than initialing it. While her employer can buy a sharper model next quarter, Diana’s experience and knowledge is not for sale, much less on those same terms.

The danger of putting machines in charge in measurable, as studies find that bad machine advice nudges people toward the wrong decision about a quarter of the time, and human experience offers no immunity. In a Spring 2026 survey of business leaders, most now lean on AI for the majority of their decisions. Many also reporting that their teams argue less than they once did. Debate is friction, and friction is frequently where judgment lives.

None of this is an argument for spending less for computer. So long as it’s delivering measurable value, there is little reason not to spend heavy. The real need is in drawing the bound, and that includes a third driver that company budgets fail to track: the signals that inform leaders to make more effective and defensible decisions. In most organizations, the majority of the information that should drive a decision, by some counts as much as 95%, never reaches the decision maker who could taken action before the window closes. Instead, that valuable signal is buried in a disparate system or siloed data. Compute will not solve that; add more agents and the room only gets louder. If you believe NVIDIAs Jensen Huang, we’ll soon be seeing up to hundred agents beside every employee, across thousands of them. This is where the right signals, including those derived from agents, will need to be surfaced and properly delivered for informing leader decisions.

This comes through the emergence of a new coordinating layer known as Systems of Attention: the fourth layer of enterprise software following Systems of Record, Engagement, and Insight. Signal Labs built it. SignalOS™ serves as the industry-agnostic operating system that allocates attention against the objective, lifting the few signals that matter out of the noise and putting them in front of the person who owns the decision, inside the window where the decision still counts. While a dashboard reports what already broke; Systems of Attention convene the enterprise on what is breaking at present, in time to act.

Tokenomics is the budget, and the budget will keep growing on its own, because that is what compute does. Humanomics is the bound, and the bound will only ever be a set of human decisions about what the institution is actually for. One of them has to answer to the other. The companies that last will be the ones that spent on compute the way you spend on anything serious, against an outcome someone owns, before the meter decided the outcome for them.