Sam Altman's enthusiasm for "tokenmaxxing" in May has given way to a corporate backlash, as companies discover that maximizing artificial intelligence token usage creates massive bills without proportional productivity gains. What began as a Silicon Valley status symbol has become a cautionary tale about the gap between AI hype and real-world economics.

What Exactly Is Tokenmaxxing and Why Did It Fail?

"Tokenmaxxing" refers to maximizing usage of tokens, the building blocks of generative AI systems that correspond to small pieces of text an AI system reads or writes. Each token represents roughly three-quarters of a word. Just a few months ago, tech executives promoted high token consumption as a badge of honor, signaling high-performing employees and innovative companies.

OpenAI CEO Sam Altman said in May he was "excited to see what will happen with tokenmaxxing startups, both for how they work internally and the products they can build". Nvidia CEO Jensen Huang went further, declaring that "if your $500K engineer isn't burning $250K in tokens, something is wrong". Meta even ran an internal competition rewarding token usage. The trend initially boosted revenue for leading AI developers like Anthropic and OpenAI, but the economics quickly unraveled.

Sam Altman

Vincent Gusdorf, head of AI analytics at Moody's Ratings, explained the shift: "It's very easy to create something you don't need with AI. As bills started to pile in, people realized that those new tools are quite expensive and you need to use them wisely".

How Are Companies Now Rethinking Their AI Spending?

Enterprise leaders are adopting a more disciplined approach to AI deployment. Rather than treating token consumption as a virtue, companies are now implementing strategic changes to control costs while maintaining productivity:

• Model Routing: Automatically directing simpler queries to cheaper, efficient AI systems while reserving powerful models like Anthropic's Claude Opus 4.6 for complex tasks requiring deep reasoning.
• Right-Sizing Tool Selection: Avoiding the use of premium AI models for routine tasks like email generation, which can be handled by less expensive alternatives.
• Employee Empowerment Over Enforcement: Shifting from rewarding high usage to teaching employees when and how to use AI effectively, allowing them to make informed decisions about tool selection.

Jue Wang, a management consultant at Bain & Company, described the financial shock facing large enterprises: "The token cost for them has been doubling, almost every other month. Let's say $200 per developer per month. Multiply that by 20,000 developers, which is often what we're dealing with at these companies, and that quickly gets you to a number that is not a line item that any general manager has planned for".

Palantir CEO Alex Karp was blunt about enterprise frustration, telling CNBC earlier this month that something had gone "completely wrong." He said he was channeling the voice of American businesses privately "livid" about paying so much for tokens that create no value.

Are Cheaper Alternatives Changing the Game?

The tokenmaxxing backlash is being complicated by the emergence of lower-cost alternatives. Chinese startups like Moonshot's Kimi and Zhipu's GLM now offer AI capabilities that nearly match top U.S. models at a fraction of the price, potentially extending the tokenmaxxing trend in unexpected ways.

However, industry observers believe the broader pattern is clear. Raffi Krikorian, chief technology officer at Mozilla, compared tokenmaxxing to an earlier software industry metric that fell out of favor: "It's similar to how software companies once considered how many lines of code a programmer wrote to be a good metric of productivity. That later fell out of favor. I think tokenmaxxing is moving through the exact same pattern. I think this is going to be an interesting blip that we're all going to look back to laugh at in a year".

What Does This Mean for OpenAI and Sam Altman?

The tokenmaxxing collapse presents a strategic challenge for OpenAI and Altman. While the trend initially boosted revenue, its fading signals that enterprise customers are becoming more cost-conscious and skeptical of AI spending. Microsoft CEO Satya Nadella has already raised public doubts about the data protection practices of leading AI providers, warning that customers are "paying twice for AI, first in spending on tokens and second by feeding all their proprietary data to them".

Separately, Altman is pursuing new growth opportunities. He recently told South Korea's President Lee Jae Myung that OpenAI hopes to use Korea as a testbed for its first hardware product, a screenless smart speaker with a rechargeable battery designed to function as a household AI companion. The device is expected to launch in the first quarter of 2027, though the company has kept most hardware details under wraps. This pivot toward physical devices suggests Altman is diversifying OpenAI's revenue streams beyond software tokens as enterprise spending discipline tightens.

The tokenmaxxing era reveals a fundamental tension in AI adoption: the gap between what's technically possible and what's economically sensible. As companies mature in their AI deployment, the focus is shifting from maximizing usage to maximizing value, a transition that will likely reshape how AI vendors price their services and how enterprises budget for artificial intelligence in the years ahead.