Software king of the world, Microsoft’s boss Satya Nadella, has warned that the AI boom risks becoming too concentrated in the hands of a few dominant model builders.
According to The Wall Street Journal the Microsoft chief executive, who helped trigger the current AI race through the company’s early partnership with OpenAI, has now become one of the loudest voices calling for a reset.
His message is aimed at the companies building the most powerful proprietary AI systems, including OpenAI, Anthropic and Google, even though he did not name them directly.
Nadella said the public would not accept a future in which a handful of AI companies gathered most of the value while warning that the same technology could wipe out white-collar jobs or become a weapon.
He questioned the idea that those companies should be allowed to keep demanding vast amounts of money, energy and data centre capacity while asking everyone else to trust them with the future.
Vole has rolled out a new set of lower-cost AI models intended to reduce costs for customers facing rapidly rising AI bills. The company has launched Copilot Cowork, an autonomous AI agent that lets users choose between different models, including cheaper ones, while it carries out longer-running tasks.
Microsoft is considering hosting a version of DeepSeek, the low-cost Chinese AI provider that has unsettled US rivals by offering powerful models at cheaper prices.
Such a move would almost certainly increase DeepSeek’s usage and could add more pressure on OpenAI and Anthropic, which are already facing the prospect of a prolonged price war.
Meanwhile, Vole remains one of OpenAI’s most important backers and has invested billions of dollars in the company. It also reached a multibillion-dollar agreement with Anthropic last year, meaning Nadella’s criticism is less a clean break than a carefully aimed shove from inside the tent.
A Microsoft spokesman said the company would continue to support its partnerships with OpenAI and Anthropic, and that Nadella’s argument was not intended as a zero-sum fight.
The deeper issue is that Vole has fallen behind in developing its own frontier AI systems. According to Recon Analytics, some Copilot subscribers increasingly preferred rival tools such as Google’s Gemini in the second half of 2025.
Unable to rely on a world-beating frontier model of its own, Microsoft is using its scale and money to push AI models towards commodity pricing and behave like normal software tools.
Nadella said companies should not think of AI mainly as a way to cut jobs. Instead, he argued that businesses need to reorganise work around human capital and what he called “token capital.”
“No, how about we think about reorganising the jobs? AI companies must provide a recipe for how that can be done. Yes, it’s a lot of change management, it’s a lot of displacement, but there is a path,” he said.
His vision is of companies using a range of models at different prices, while keeping control of their own data, knowledge and intellectual property. Future companies would become continuous learning systems, combining human judgment with AI tools that remain under their control.
That matters because firms that hand too much of their learning to outside AI platforms risk being commoditised themselves. In other words, the machine does the learning, the model owner gets the value and the original company gets to clap politely while being hollowed out.
Nadella said the AI industry now had to earn “social permission” by showing that people still have agency and economic opportunity. A nicer slogan will not be enough, which is unfortunate for the technology industry, where slogans are often treated as a substitute for civilisation.
