Source: Meta shutters Claudeonomics, an internal, employee-built leaderboard tracking each employee's token usage, due to the data “being shared externally”
The InformationJyoti Mann
Context & Ripple Effects
Meta’s internal AI-use culture had turned token consumption into a visible status competition: Claudeonomics rewarded employees for high token use, part of the broader workplace “tokenmaxxing” pattern documented across companies. Its closure shows that a grassroots adoption mechanism can become a governance issue when its underlying usage data leave the company.
The episode also foreshadows a shift from encouraging consumption to managing it. Later coverage reported planned limits on employee token use and a push toward MetaCode as internal AI-spending forecasts climbed, tying employee experimentation to infrastructure-cost control.
First-order effects
Meta loses a public internal incentive and measurement layer for employee AI-token usage; employees who used the leaderboard for recognition no longer have that channel.
The reported external sharing of usage data makes internal telemetry, rather than AI adoption alone, the immediate control problem for Meta.
Second-order effects
Teams building internal AI tools are likely to face tighter access, data-handling, and publication rules, reducing the ease with which employee-created dashboards can spread usage metrics.
Token-use incentives become harder to separate from spend management: the later move toward limits and MetaCode suggests Meta will steer demand toward tools it can more directly govern.
Third-order effects
If similar controls become standard, enterprise AI adoption will move away from raw-token consumption as a productivity signal and toward centrally managed, auditable workflows.
The pattern strengthens the case that AI compute is becoming an internal budgeted utility: organizations will need to balance experimentation incentives against data exposure and variable inference costs.
The trend: Enterprise AI programs are shifting from employee-led usage gamification toward governed, cost-aware consumption of centrally controlled AI tools.
Plenty of my Meta friends told me folks have been building bots that just run in a loop burning tokens as fast as they can due to this policy. It's an absolutely stupid policy and is similar to how Meta uses LoC to measure eng output. Managers are supposed to use it as a proxy …
FB leadership thinks the problem here is that the leaderboard got shared externally, not that “spend as many tokens as possible” is not a good target to give people
New: Meta has taken down its internal AI token leaderboard. It now displays a message that says: “It was meant to be a fun way for people to look at tokens, but due to data from the dashboard being shared externally, we've made the decision to shutter Claudeonomics for now.”
.@johncoogan says the recent reporting on Meta's ‘tokenmaxxing’ is less of a sign of bad incentives at the company, and more of a tell about its potential strategy for more vertical integration: “I think it makes clearer the strategy with MSL. Because it's clear that they're [vid…
Hearing the same from Meta friends. As bad as LOCs were, they were still somewhat auditable. If you committed 100k lines of bloated nonsense, your peers saw this, and thought less of you. It's also part of your commit history and reputation forever. OTOH token usage is an
@kevinroose ...annnnd the “claudenomics” dashboard inside meta has been taken down, im told unclear to me if this was a homespun one by employees or an official one. (employee projects come and go frequently) conspicuous timing though...
Inside Microsoft it's the same silly games and incredible token waste for nothing. There's just not been as extensive reporting on it like at Meta (I talked with devs at both companies to confirm all of these as usual)
Devs game everything and anything seen as a target for more bonus or promos. This was no different. Talked w devs at Meta and this was SO MUCH waste - just to get on the leaderboard!! Meta removed the leaderboard now. Source: me (Next up I expect to do the same: Microsoft)