I'm 35 years old, which means I'm safely out of the loop on what younger people are saying these days. I don't know what rizz means. I'm fuzzy on the manosphere. I have no idea what looksmaxxing is supposed to help with. But there's a new word I keep hearing called tokenmaxxing, and even though it might be one of the dumbest things ever to enter the corporate vocabulary, I fear we're all going to be affected by it. I recently read a piece titled "Amazon employees are 'tokenmaxxing' due to pressure to use AI tools, and it's worse than you think." Amazon, the company famous for creating a balanced, employee-friendly work environment where warehouse staff allegedly urinate into bottles to meet hourly quotas, is now extending the same operating philosophy into its software division. The vehicle is an internal AI coding tool called MeshClaw. Every major tech company is rerouting capital toward AI agents, so it isn't surprising that Amazon, which is spending around $200 billion on AI infrastructure this year alone, would also push some kind of agent on every employee it has. But for some reason, simply having the tools isn't enough in the race to that mythical 10x AI productivity. So Amazon set a target: more than 80 percent of developers must be using AI tools every single week. To make sure that's actually happening, the company started tracking something called token consumption on internal leaderboards. A token, as most people now know, is a chunk of data an AI model processes. The more you use the AI, the more tokens you burn. The more tokens you burn, the higher you climb on the board. The problem with this approach is straightforward: a token, like a line of code, holds no value on its own. A developer who writes ten thousand lines of garbage is not ten times more productive than a developer who deletes a hundred lines and makes the system simpler. In most companies, the second developer is the more valuable one. I believe we'll eventually draw the same conclusions about token consumption, but for now, we're collectively pretending size matters when it comes to AI usage. So Amazon's software engineers are doing exactly what the incentives demand. They build little MeshClaw agents whose entire purpose is to do nothing, or to perform tasks that don't need performing. In other words, they're deploying artificial intelligence to do fake work so that a dashboard somewhere will show they are using artificial intelligence. This is it. It's May 2026, and we've reached peak stupidity. What's more concerning is that Amazon isn't alone. A few weeks ago, news broke that Uber had blown through its entire 2026 AI R&D budget in just four months because the company was rewarding engineers for how much AI they used, not for what they actually built. Uber aggressively rolled out Claude Code, and engineers were incentivized to delegate everything to the agent. As a result monthly token bills grew up to $2,000 per engineer. And what’s really funny is that once the budget was gone, internal audits suggest that the massive token burn actually led to an increase in "code bloat," requiring even more AI tokens to later "refactor" the junk the AI initially wrote. Meta took a different route. It built an internal leaderboard called Claudeonomics, ranking its 85,000 employees by token consumption, publishing the top 250 on the intranet, and handing out tier titles like Token Legend, Session Immortal, and Cache Wizard. Keep in mind this is a company burning through billions of dollars and immense compute resources to fund what amounts to an in-house arcade game. Over thirty days, Meta employees collectively burned through more than 60 trillion tokens, which is equivalent to roughly $9 billion in public API compute. Meta presumably gets bulk pricing, but even at the most generous discount the real cost lands somewhere above $100 million. Then Dario Amodei takes that number, adds it to Anthropic's balance sheet, and tells the world that enterprise AI adoption is exploding. The same companies spending a hundred million dollars a month so engineers can win badges named after wizards are simultaneously letting 8,000 people go because the AI bill has to be paid somehow. George Carlin put it best. "It's a big club, and you ain't in it."