A corporate fad of \u201ctokenmaxxing\u201d on artificial intelligence technology is hitting its limits as workplaces throwing AI at everything are seeing the costs rise without a similar spike in productivity.<\/p>\n
What started as tech industry-fueled springtime hype over squeezing as much AI-generated work as possible out of products like OpenAI\u2019s ChatGPT and Anthropic\u2019s Claude has shifted to a summertime backlash.<\/p>\n
\u201cIt\u2019s very easy to create something you don\u2019t need with AI,\u201d said Vincent Gusdorf, head of AI analytics at Moody\u2019s Ratings and author of a new report that recommends a more disciplined approach.<\/p>\n
\u201cTokenmaxxing\u201d refers to maximizing usage of tokens \u2014 the building blocks of generative AI that correspond to small pieces of text that an AI system reads or writes. Each token is about three quarters of a word. And there\u2019s typically a limit to how many you can use, with pricier versions of AI products offering higher caps.<\/p>\n
\u201cAs bills started to pile in, people realized that those new tools are quite expensive and you need to use them wisely,\u201d Gusdorf said.<\/p>\n
Tech executives cast high AI usage as a badge of honor<\/p>\n
Just a few months ago, Silicon Valley executives were promoting high token consumption as a signal of high-performing employees. The stereotypical tokenmaxxer was staying up late \u2014 perhaps ignoring their significant other \u2014 while orchestrating an army of 24-hour AI agents performing work on their behalf. <\/p>\n
OpenAI CEO Sam Altman said in May he was \u201cexcited to see what will happen with tokenmaxxing startups, both for how they work internally and the products they can build.\u201d <\/p>\n
Nvidia CEO Jensen Huang said \u201cif your $500K engineer isn\u2019t burning $250K in tokens, something is wrong.\u201d Facebook parent Meta had an internal competition rewarding token usage. <\/p>\n
The trend boosted revenue for leading AI large language model developers like Anthropic and OpenAI, but it fizzled as it became apparent it wasn\u2019t necessarily the best strategy for everyone else. <\/p>\n
Microsoft CEO Satya Nadella has admitted that tokenmaxxing can be addictive but warned in a recent blog post that customers of those models are paying twice for AI, first in spending on tokens and second by feeding all their proprietary data to them. While promoting Microsoft\u2019s own approach, Nadella\u2019s comments were unusual in the way he raised doubts about the data protection assurances of leading AI providers.<\/p>\n
Palantir CEO Alex Karp went further, telling CNBC earlier this month that something had gone \u201ccompletely wrong.\u201d He said he was channeling the voice of American businesses privately \u201clivid\u201d about paying so much for tokens that create no value.<\/p>\n
\u201cThe basic view among enterprises in this country is, \u2018I\u2019m going to chillax and waste my time with tokens. I\u2019m going to get no value and they\u2019re going to get my IP,\u201d Karp said.<\/p>\n
Workplaces look more for better \u2018routing\u2019 of their AI work<\/p>\n
Bain & Company management consultant Jue Wang said many of the big businesses her firm advises have been taking a closer look at returns on their AI investments.<\/p>\n
\u201cThe token cost for them has been doubling, almost every other month,\u201d she said. \u201cLet\u2019s say $200 per developer per month. Multiply that by 20,000 developers, which is often what we\u2019re 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.\u201d<\/p>\n
Sometimes that just means not using the AI equivalent of a sledgehammer to crack a nut. <\/p>\n
\u201cNot everything needs a Claude Opus 4.6,\u201d she said of one of Anthropic\u2019s more capable models suited to software engineering or deep research. \u201cAnd yet you see so many companies, so many users, default to using Opus for everything, including generating emails.\u201d<\/p>\n
That\u2019s led to a search for tools that do AI \u201cmodel routing\u201d \u2014 in which easier queries get automatically sent to cheaper and more efficient AI systems and more complex tasks go to more powerful models.<\/p>\n
Open-source AI models built in China offer less costly alternatives<\/p>\n
Software developer Hassan El Mghari said companies\u2019 sticker shock over the \u201cridiculous amount of money\u201d spent on subscriptions to AI products from leading U.S. companies has led many away from rewarding high usage. <\/p>\n
\u201cIt\u2019s better to kind of just empower employees on how to use this stuff and let them use AI when and however much they need to,\u201d said El Mghari, who leads developer experience at the startup Together AI, which supplies developers with a variety of \u201copen-source\u201d AI models. <\/p>\n
At the same time, those who favor racking up as many tokens as possible are having a field day with new open-source models from Chinese startups like Moonshot\u2019s Kimi or Zhipu\u2019s GLM, which nearly match the capabilities of top U.S. models at a fraction of the price.<\/p>\n
\u201cThere is some validity to the theory that this could push tokenmaxxing a little bit further,\u201d said Raffi Krikorian, the chief technology officer at Mozilla. \u201cBut if we look at the industry overall, I think it\u2019s realizing that tokenmaxxing is a dumb thing.\u201d<\/p>\n
It\u2019s similar, Krikorian said, 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.<\/p>\n
\u201cI think tokenmaxxing is moving through the exact same pattern,\u201d he said. \u201cI think this is going to be an interesting blip that we\u2019re all going to look back to laugh at in a year.\u201d<\/p>\n
O\u2019Brien writes for the Associated Press. <\/p>\n","protected":false},"excerpt":{"rendered":"A corporate fad of \u201ctokenmaxxing\u201d on artificial intelligence technology is hitting its limits as workplaces throwing AI at…\n","protected":false},"author":2,"featured_media":123023,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[8],"tags":[24,7738,53,182,532,62077,11466,58558,157,748,2315,2518,12450,61291,1955],"class_list":["post-123022","post","type-post","status-publish","format-standard","has-post-thumbnail","category-anthropic","tag-ai","tag-ai-system","tag-anthropic","tag-claude","tag-company","tag-corporate-fad","tag-developer","tag-few-month","tag-openai","tag-product","tag-startup","tag-token","tag-usage","tag-vincent-gusdorf","tag-workplace"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/123022","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=123022"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/123022\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/123023"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=123022"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=123022"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=123022"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}