A partial Tech in Asia article on “tokenmaxxing” makes a narrower point than simple AI enthusiasm or skepticism: raw AI usage, and especially token burn, is not the same thing as business value. In the available text, the piece argues that corporate teams can treat token burn as a vanity metric and should focus on measures that reflect whether work is actually improving instead. The partial article says teams can “track vanity metrics like token burn if you must, but make sure you’re measuring what really matters.”
The example it opens with is deliberately showy. According to the supplied article text, a software developer at an AI-themed meetup in Hong Kong said last month that he usually hits the token limit of his two Claude Max accounts every day with coding work. The article presents that as part of a broader culture of “tokenmaxxing,” which it describes as maximizing AI usage by running as many prompts, agents, and automated workflows as possible.
That framing matters because, in the visible portion, the piece is not treating heavy AI use as inherently suspect. Instead, it argues that token burn and prompt volume can be weak stand-ins for productivity when organizations are still trying to decide what successful AI adoption should look like. The article’s author writes that for an AI consultancy, “token burn is meaningless if there are no visible outcomes,” because it measures “number go up,” not whether work is better or newly possible.
Why the metric problem matters
The partial text ties that argument to a broader adoption gap inside companies. It cites a 2025 McKinsey survey saying 62% of companies are either experimenting with AI or just starting their AI programs. In the article’s telling, that helps explain why many teams still have not settled on meaningful ways to evaluate AI use.
The visible examples in the article point in that direction. Tech in Asia’s text says one content writer at an insurance company had never had AI-related KPIs or benchmarks raised with him, while an international investment bank’s most advanced metric was a manager dashboard showing prompt counts per employee. The author describes that as a useful first signal, but not much more than that.
Because the full premium article is not supplied here, the specific alternative metrics it recommends are not available in the evidence. What is supported is the narrower takeaway from the accessible text: the article frames “tokenmaxxing” as a productivity flex that can obscure the harder question of whether AI is improving outcomes in a way a business can actually measure or use.
That limitation is worth keeping in view. The supplied material supports the article’s critique of token burn as a headline metric, plus its examples and McKinsey citation, but not the full list of what companies “should measure instead.”
