Satya Nadella urges Microsoft staff to not only make the most of tokens, but also use AI smarter – Firstpost

On a recent episode of the New York Times podcast “Hard Fork,” Microsoft CEO Satya Nadella gave a candid review of the company’s AI habits. Asked how much “token maximization” is happening at Microsoft, Nadella answered “quite a lot” before the question was even finished.

Calling himself a “token maxer,” Nadella joked that constantly interacting with AI could become a habit. However, he stressed the importance of looking beyond the initial excitement and focusing on meaningful outcomes, according to a Business Insider report.

“It’s easy to get caught up in the novelty,” he said, adding that users ultimately need to ask themselves what they are actually trying to build or accomplish.

Over the past year, leading technology companies have been actively encouraging their employees to incorporate AI into their daily workflows. Some companies have introduced internal rankings that measure AI usage through token consumption (units of data processed by AI models). However, as the costs associated with large-scale AI deployments become clearer, organizations are increasingly prioritizing efficiency over outright usage.

Nadella did not explicitly say that Microsoft is restricting employees’ access to AI tools, but he emphasized the importance of choosing the right model for each task.

“There’s no need to deploy frontier models for everyday problems,” he explained, citing Microsoft Copilot’s automatic mode that automatically selects the best model based on the task at hand. The goal, Nadella said, is to balance performance and cost-effectiveness, rather than pursuing the use of AI that has little practical value.

As an example, Nadella recently revealed that he used AI-assisted coding to build a tool that can monitor workplace discussions related to software projects. The system can identify relevant conversations, generate implementation plans, apply updates, and ensure that the software continues to function properly without having to be directly involved in every meeting or message thread.

Microsoft’s top executives have spent the past few years restructuring the company for an AI-driven future, aiming to make the 220,000-employee organization more agile and more competitive against smaller, faster-moving rivals.

Part of that effort included changes in leadership. In October, Nadella appointed a new head of Microsoft’s commercial business, giving him more time to focus on technical and product-related efforts. The following month, he brought on new AI advisors to help Microsoft rethink its long-term business strategy in an era increasingly defined by artificial intelligence.

Palantir’s Alex Karp also complained about similar issues, recently stating:

The company said it has built internal tools aimed at curbing what he calls “token maximization,” or the excessive use of AI tools that give the appearance of productivity without creating meaningful value.

Karp argued that many companies rely on AI for low-impact tasks, risking mistaking activity for outcome. He likened the behavior to a habit that is difficult to break, and said organizations need to focus on results instead of endlessly producing dashboards, reports and other AI-driven output with little practical benefit.

Although his comments were delivered with humor, Karp’s broader point was that companies should ensure that the use of AI is tied to real business value, rather than an obsession.