AI, too expensive for Apple: the giant reportedly cut its employees’ budgets

Is generative artificial intelligence too expensive, even for a company as gigantic and powerful as Apple? Journalist Mark Gurman got wind of Apple's internal policy, which recently changed to reduce the budget given to its employees.They will have to spend less money on AI tools.

Apple wants to spend less and less on AI, at least for its employees

According to Bloomberg journalist Mark Gurman, Apple has “quietly” lowered the daily spending cap for some of its engineers, with funds allocated for using AI models. This cap would have fallen from $300 to $200 per day. Apple could even go further: in some teams, the budget would drop to $100 per day. This is already the case in some non-engineering departments, and for these, Apple would like to reduce it to just $25. The only way for employees to get around these restrictions is to “request an exemption from management,” writes the journalist.

These budgets concern the use of generative AI: one can imagine tools such as Claude Code or Cursor for software development. This covers tokens, or tokens, but also all other data entrusted to AI models for processing. Providers such as Anthropic, OpenAI, Google, and others charge their clients based on the number of tokens consumed. And the more powerful a model is, the more expensive it is and the more tokens it consumes.Facing these changes, internal reactions would be mixed: some employees would claim that few of them used the full amount of these credits, “which made the previous limits unnecessarily generous.” Others fear that this cost-cutting policy could extend to layoffs.

Mark Gurman therefore describes a dilemma for Apple to resolve: “keeping costs under control while encouraging employees to incorporate more AI into their daily work.” This is precisely one of the challenges facing John Ternus, who has just taken the helm of Apple: “running the shop” more efficiently. Apple still wants to be at the forefront of technology and productivity: it does indeed need generative AI tools for this. But this productivity comes at a price,and Apple does not seem willing to pay too much for it. Fortunately, it is far from being in the red: its revenue is still growing. When asked, the company declined to comment.

Tech companies are tightening their belts

It must cost quite a lot, according to the creator of OpenClaw: in one month, he spent more than $1.3 million on tokens. Apple is therefore far from being the only company limiting the costs associated with the internal use of generative AI.The most notable case is undoubtedly Microsoft’s, which reportedly ended its Claude Code licenses after consuming a substantial number of tokens without achieving conclusive results. The same applies at Tesla: a weekly limit of $200 was imposed; at Uber, the limit is $1,500 per month for AI coding tools. As for Meta, internally developed AI models are now favored there.

In response, OpenAI is trying to attract these large companies with a change in its billing model last August. For some major clients, OpenAI charges based on performance rather than the volume of tokens consumed. Unsuccessful attempts are no longer billed: only work actually completed counts. Companies have no choice: an IBM study revealed last year that only 25% of AI-related projects delivered the expected returns. OpenAI, Anthropic, and Google therefore have to adapt to these changes in spending policies.

👉🏻 Follow tech news in real time: add 01net to your Google sources and subscribe to our WhatsApp channel.