Tokenmaxxing is a failed experiment

I want you guys to read this tweet by Mark carefully. A mysterious company burned through 500 million dollars in Claude LLM tokens in just a single month, all because nobody bothered to set usage limits. Now, I tried to find out which company this was, but I couldn't find the name. All I could find was this news article on Yahoo Finance. So I guess the one thing this company is good at is hiding their identity. But they literally spent 500 million dollars in a single month just on LLM tokens. And if you do the math, 500 million in a month is basically like spending 16 million in a day. Like, that's insane. I don't know what company this is, but they must be really fucking lousy if they can't even see 16 million dollars leaving their bank accounts every fucking day and nobody even notices that. And that's not all. Uber ran company leaderboards ranking engineers by how much AI they used, not what they actually ship, and they ended up burning through their entire 2026 budget by April. As in, they've still got like eight months to go for the year, and they've burned through the entire year's budget in the first four months. And the real funny part over here is that they can't even connect it to any consumer features. So where did all that money go? I don't know. Summarizing emails, summarizing long text, I don't even know. Like, just complete bullshit, right? They literally didn't build or ship or improve anything about the actual app or product or the user experience at all. And it just gets even more crazy over here. There are employees who are literally using enterprise AI models to check the weather. Like, that where the Uber budget went, basically. Something that takes two taps on your phone is now costing your company inexpensive LLM tokens. Microsoft, dear Microsoft, they've started canceling Claude Code licenses to their employees. Why? Because the token bill has just spiraled out of control. And other companies are now starting to lay people off just to pay the AI bill to OpenAI and Anthropic. And this isn't because AI has somehow automated to replace the work. This is because they literally have bills to pay to these LLM companies, and they have to lay people off to pay those bills. They're literally firing people just so they have money to pay for all of the tokens they're wasting. Now, I recently made a video titled "Token Maxing Needs to Stop" right here, and I talk about this exact behavior, right? CEOs and non-technical managers are spending hundreds of billions of dollars on LLM tokens, and they have no clear consumer benefit to show for it. Like, nobody's winning here. Not the company, not the customer, not the employees. Nobody, except I guess OpenAI and Anthropic, who are just printing money from all these LLM tokens being spent. So now let's look at this other tweet by Charlie here. It turns out that Amazon had an AI leaderboard, which they have now removed, right? So they had a leaderboard where they were tracking how many tokens each employee was using, and to pretty much nobody's shock, they've now removed it because the costs were skyrocketing with no clear payoff or benefit to end users. Like, I just don't understand what these CEOs are even thinking at this point. Like, you incentivize your employees to use as much tokens as possible, right? And then you're surprised that your employees are using as much tokens as possible. Like, I wish I had a surprise Pikachu face to put over here, but you literally incentivized your employees to use as much tokens as possible. And now you're surprised that they're using as much tokens as possible. What the fuck? And Starbucks, of all companies, is getting into AI. And they have an AI agent that can't even count coffee cups, right? Like, imagine having an AI agent in your coffee shop that can't even count your coffee cups. Like, that's just insane. And let's not forget Uber, who burned through $3.4 billion in just four months. So that's the yearly budget that they set for 2026 that we just read in the previous tweet, right? They burned through such a huge amount in just four months with nothing to show for it. And it's such a catastrophic failure that I don't even know where to begin with this. Like, are you seeing the problem here or is it just me? Companies just rewarding the meaningless consumption of LLM tokens instead of actual productive output. Like, this shit has to stop. I've already made a video on this topic, and now I have to make another because these managers don't fucking understand what they're doing until they burn through billions of dollars. And quick side note about Uber here. I don't understand how anyone could end up burning through 3.4 billion dollars in LLM tokens and nobody fucking notices that. Like, if you do the math, this is even worse than this mysterious $500 million company over here. 3.4 billion in four months. That literally [costs] 28 million every fucking day. Are you seriously telling me that nobody in your entire company noticed million[s] leaving the company budget every fucking day? Like, the sheer stupidity of some of these companies is just mind-boggling to me. Like, how do you not notice $28 million leaving your bank account every fucking day? Like, you've literally burned through like $200 million in just a week. How do you not even notice that? What the fuck is going on in these companies? Like, this is just insane. And the whole notion of trying to track productive AI usage by just looking at the raw LLM tokens is just stupid, right? It's a fool's way of tracking AI usage. Instead of trying to measure AI usage like this, what we should really be doing is look at stuff like what real problems have been solved, what features have been built, and what actual productive output has been achieved by AI, right? And if you notice, none of these things are things that can be tracked just by looking at how many raw LLM tokens you're consuming, right? So the whole tracking raw LLM tokens was just a dumb idea right from the start, right? So we're already starting to see the consequences of this whole token-maxing charade, and any company that's still trying the token-maxing is just asking to go bankrupt. That's all I wanted to say in this video. Also, quick self-promo before I end the video. If you're a business owner or a manager and you want to integrate AI into your company without all of the massive cost problems that we discussed in this video, then top link in the description. You can book a call with me right now, and I'll personally help you to become an AI-first company. That's all. Thank you for watching. Like, comment, subscribe, all that good stuff.