Accenture scrambles to curb AI spending as employees burn tokens on PDF conversions

Companies are scrambling to stop employees from maxing out AI budgets with small tasks

The era of tokenmaxxing is over. After the AI industry encouraged companies to max out their AI budgets earlier this year, and some companies even built employee leaderboards to encourage internal AI usage -- they are now realizing just how easy it is to spend huge sums of money on AI and get little in return. We now appear to be entering the era of token rationing. Recent news has been rife with stories about AI cutbacks and now 404 Media reports that consulting firm Accenture has been attempting to stop its employees from depleting its token reserves by using AI to do basic tasks -- like converting PDFs into presentation slides. The cutbacks take place not long after Accenture threatened that employees would "risk losing out on promotions" if they didn't use AI, 404 writes. 404's reporting is based on leaked audio from a recent internal meeting involving Accenture's agentic AI strategy lead, Justice Kwak. "We're hitting this inflection point where AI is becoming material to the cost structure," Kwak says. "Spend is becoming very unpredictable; and leadership, especially at the CFO, COO, and CIO level, are still asking the question of whether they're getting value from what we're spending on in the context of AI." The cost of tokens has thrown into doubt the AI business model -- as evidenced by what's being called the "AI selloff" which has battered some AI-dependent businesses the last few days, especially memory chip makers. The AI industry has reached the stage where it can't just be exciting and new anymore. It has to prove its worth.

The Token Belt-Tightening Is Coming for PDFs

The tokenpocalypse is upon us, and it's coming for nontechnical staff. Many large companies have pushed employees to use AI for nearly everything, partly to justify the enormous sums they poured into the technology without any clear integration strategy. But a new report from 404 Media shows that some of those same firms are now tightening access after discovering that employees are burning tokens on tasks that never required AI in the first place. The publication reportedly obtained audio of a meeting at consulting giant Accenture, where staffers have been told they're seeing "soaring token spend" driven by pointless tasks like converting PDFs into presentation slides. Those uses are actually starting to burn more tokens than the work being done by technical staff. "We're seeing from some of the data internally, at least, that it's actually not our engineers that are driving the token consumption," an Accenture employee said in the audio, per 404 Media. "It's a lot of the non-engineers that are doing some of those behaviors." While it may seem a bit absurd to think that workers who focus on BS-ing their way through presentations would be using more tokens than their colleagues over in engineering, PDFs are an extraordinarily inefficient way to feed information into AI systems. Depending on the tool and the document, a model may have to extract and interpret not just the text but also every page's layout, images, charts, and other visual elements. The file format may be a boss's best friend, but that could change once the token bills start piling up. It's not exactly hard to figure out how Accenture found itself in this situation. According to a report published by the Financial Times earlier this year, the Fortune 500 consulting company went out of its way to push people to use AI tools, going so far as to track logins made by staff and tying promotions to their chatbot usage. In a memo to the staff, the company reportedly said that moving up in the company would require "regular adoption" of AI, signaling that using the technology would not be optional. Accenture, of course, is far from the only company to use some shockingly uncritical metrics to push AI usage. Big Tech firms like Meta and Amazon started leaderboards to track which employees in the company were burning through the most tokens and incentivizing them to push themselves onto the list. This, of course, just leads to people doing menial tasks with AI that they could otherwise do themselves, just for the sake of burning tokens. That approach is dumb but fine in the "free money" era of AI, where costs weren't linked to token usage. But as the major AI labs head toward IPOs, they've shifted to a usage-based model for pricing. The result has been some absurdly high bills. That is less sustainable, and now the companies are starting to ask their employees to cut back on AI usage instead of ramping it up. Weird how a company that tells other businesses how to operate more efficiently and effectively failed to see the very obvious and predictable token crash coming. Probably nothing to read into there.

The AI About Turn: Step Aside Token-maxxing; Tokenpocalypse is Here

In February Accenture said it would promote only those who use AI, now they are seeking ways to curtail AI usage Exactly a week ago, we had reported the new trend in AI usage where enterprise CEOs had shifted from being AI evangelists to AI misers. Now it appears that these head honchos have turned cost-conscious with a vengeance and want token rationing to become part and parcel of their operations. An article published by 404 Media notes that of all the companies, Accenture, which had at one time announced that future promotions would be based on token use, has reversed its position. The report claims access to a leaked audio where Accenture is trying to understand how to stop non-technical workers from blowing through AI token budgets. The report claims that the consulting giant has witnessed a trend where employees used AI tokens to do basic stuff like converting PDFs into presentation slides, which is worrisome as Accenture is seeing a "soaring token spend" across industries. This turnaround in Accenture's position is indeed fascinating and possibly explanatory of the hype that AI has brought in its wake. In February, CEO Julie Sweet had revealed that she had asked employees to get friendly with AI if they wanted future promotions. Now, it looks like the staffers got really friendly and ended up blowing all their tokens and more. Looks like the era of tokenmaxxing is coming to an end as companies are casting aside employee leaderboards built to encourage AI usage and moving back to a more practical position of using the new technology for tasks that delivered better returns. Looks like we are now in the age of token rationing where employees get to use them only for relevant tasks, not merely as shortcuts to reduce time and effort. The move seems to be quite an overarching one as reports said even Meta has rolled in to reduce AI usage in order to keep down costs, which a report said had crossed billions of dollars this year. According to the 404 media article, the leaked audio from an internal meeting had Accenture's agentic AI strategy head Justice Kwak. "We're hitting this inflection point where AI is becoming material to the cost structure. Spend is becoming very unpredictable; and leadership, especially at the CFO, COO, and CIO level, are still asking the question of whether they're getting value from what we're spending on in the context of AI," the report said. This trend exposes a further chink in the armour of the ongoing AI goldrush, especially since the cost of tokens has created concerns over the existing business model that companies are built on. In fact, even the stock markets appear to be concerned about this trend as evidenced by a report from CNN which called it an "AI selloff." The report quoted James Reilly, senior markets economist at Capital Economics, to suggest that the single-day drop of Nasdaq (2.21%) and S&P 500 (1.44%) caused largely by selling in both semiconductor chip stocks and other AI-related shares, may be a result of a growing trend of rising volatility, which is "evidence of excessive froth" and calls into question the sustainability of this rally. "Whatever the cause, with AI companies' sky-high valuations and incredible growth trajectories, it doesn't take much to set off investors. The Kospi is up more than 90% this year, so when the wind blows in an unexpected direction, it can lead traders - and, often more consequentially, trading algorithms - to head for the exits. They fear the top of the Jenga tower could tip over," the report said. Of course, it is not just the markets that are concerned. Microsoft CEO Satya Nadella called out the general frenzy around AI in an interview with The Wall Street Journal earlier this week. You can't say, hey, all white-collar jobs are gone and this could even be a weapon and we will use all the power to build datacentres. The public, he predicted, wouldn't tolerate just a few models and companies "doing all of the learning for the world," he told the WSJ. Meanwhile, his company also began exploring alternatives to the AI giants to power up Microsoft's own AI offerings. And the first stop they made is with China's DeepSeek as a partner for Copilot, albeit with some caveats such as having its servers in the US. This move also happens to run parallel to Microsoft's efforts to shift its Copilot Cowork to a usage-based pricing model across the entire enterprise AI space. The report says the Redmond-based tech giant might be considering a locally hosted version of DeepSeek as a cheaper option for its users. And barely days before these reports, we saw the company ask its employees to reduce Anthropic's Claude licenses. Then came reports that they are making money in the Chinese markets by selling OpenAI's ChatGPT as part of the Microsoft Azure offerings. Then came a report that quoted Suleyman as saying that he wants Microsoft to be the fourth AI lab after DeepMind, Anthropic and OpenAI. Whatever be the actual case, the fact remains that all these incidents coming one after the other could slow down the frenzy over AI as the next best thing after sliced bread. Maybe, it is for the global good that the hype is being replaced with a practicality that went up in smoke when the new age whiz kids announced AI as the magic wand for everything and anything.