# Dodo Digest: Everybody's Tokenmaxxing. Are You? > AI adoption is accelerating so fast that companies are burning through budgets faster than expected, with one firm reportedly spending over $500M on Claude tokens alone. The real edge isn't consuming the most tokens, it's building sustainable systems around them. Plus, we shipped Dodo Payments v1.101.0. - **Author**: Rishabh Goel - **Published**: 2026-06-05 - **Category**: Newsletter - **URL**: https://dodopayments.com/blogs/newsletter-june5 --- ## Everybody's Tokenmaxxing. Are You? ### TL;DR - AI adoption is accelerating so quickly that companies are burning through AI budgets faster than expected. - One mysterious company reportedly spent over $500 million on Claude tokens alone. - Uber had to cap employee AI spending after exhausting its budget in just a few months. - Startups are shipping products faster than ever, but infrastructure decisions matter more than ever too. - We shipped Dodo Payments v1.101.0 with several improvements focused on making payments and operations even smoother. ## Hello everyone, I came across a story this week about a mysterious company that reportedly spent over $500 million on Claude tokens. At first, that number sounded absurd. Half a billion dollars on AI usage alone feels like the kind of statistic you'd expect a few years from now, not today. But the more I looked into it, the less surprising it became. Because everywhere you look, companies are running into the same problem. Not whether AI works. But how much it costs once it actually does. ## The Signal For the last two years, most of the conversation around AI has been about capability. Can it write code? Can it replace workflows? Can it automate repetitive tasks? We've largely moved past those questions. The answer, in many cases, is yes. The new challenge is economics. Uber recently had to cap employee AI spending after teams blew through budgets much faster than expected. At the same time, companies like Lovable are signing larger infrastructure agreements because demand continues to outpace what they originally planned for. That says something important. AI adoption is no longer experimental. It's operational. And once something becomes operational, cost becomes one of the most important variables in the equation. A $500M bill: the new reality of AI usage at scale ## Where Things Start to Change What makes this particularly interesting is that we're seeing an entirely new generation of companies being built around AI. Every week there seems to be another startup launching a website builder, coding assistant, automation platform, workflow tool, or AI-native SaaS product. And many of them are growing surprisingly fast. We're probably in one of the most active startup-building periods we've seen in years. The barrier to building has dropped dramatically. A small team can now launch products that previously would've required dozens of engineers. But lower barriers create a different challenge. When everyone can build quickly, the quality of your infrastructure decisions starts to matter much more. Because growth compounds complexity. And complexity compounds costs. A new generation of AI-native products competing for the same workflows ## How To Think About This Here's how to approach this if you're building today: 1. **Optimize for economics, not just capability** -- A feature that works isn't enough. It has to work sustainably as usage grows. 2. **Assume success will stress your infrastructure** -- Many products aren't prepared for what happens when usage suddenly increases 10x. 3. **Pick systems that reduce operational overhead** -- The less time your team spends managing infrastructure, the more time they spend building. 4. **Think globally from the start** -- The internet doesn't scale region by region anymore. Products often find users everywhere from day one. 5. **Avoid rebuilding critical systems later** -- Payment infrastructure, billing, compliance, taxes, and subscriptions become harder to change once growth begins. ## What This Means for Builders This is one of the reasons we've always believed infrastructure should remove complexity, not add to it. Most of the AI startups being built today are relatively small teams. They're focused on product, distribution, growth, and customer acquisition. The last thing they should be worrying about is tax registration in multiple countries, compliance obligations, or maintaining payment operations across different markets. That's where the Merchant of Record model becomes valuable. Instead of spending months building and managing those layers internally, founders can focus on the thing that actually differentiates them: their product. Because if AI is reducing the cost of building software, founders should be looking for ways to reduce the operational cost of selling it too. ## What We Shipped We spend a lot of time thinking about product quality internally. As usage grows, expectations grow with it. Small improvements that save a few clicks, reduce confusion, or improve visibility can end up making a big difference over time. That's the thinking behind many of the updates we shipped in v1.101.0. This release focused on improving day-to-day workflows across billing, subscriptions, and customer management, while making the platform more predictable and easier to operate at scale. A few highlights: - **Subscription Payment Retries** to automatically recover failed renewal revenue without additional integration work. - **Business Proration Settings**, allowing teams to define upgrade and downgrade behavior once and apply it across products. - **Business Name Collection for B2B Invoices**, making invoices cleaner and more accurate for business customers. - Various bug fixes and stability improvements across the platform. None of these changes are particularly flashy on their own. But infrastructure products are rarely judged by individual features. They're judged by how they feel after hundreds of interactions. And that's where we continue to focus our effort: making payments, subscriptions, and billing feel simpler than they did yesterday. Subscription Payment Retries: recovering failed renewal revenue automatically ## One Last Thought A few years ago, the biggest question in AI was whether the technology would work. Today, the question is whether it can scale economically. The companies winning right now aren't just building smarter products. They're building sustainable systems around them. And as AI becomes more deeply integrated into software, that distinction will only become more important. Because eventually, everyone will be tokenmaxxing. The winners won't be the companies consuming the most tokens. They'll be the ones generating the most outcomes from them. Also, join our loving [Discord community](https://discord.gg/dodo-payments-1305511580854779984)! Best, **Rishabh Goel** **Co-Founder, Dodo Payments** --- - [More Newsletter articles](https://dodopayments.com/blogs/category/newsletter) - [All articles](https://dodopayments.com/blogs)