Microsoft CEO Satya Nadella recently acknowledged the advancements of DeepSeek, a Chinese AI startup that made waves earlier this week with its open-source model, R1. The model has drawn attention for its claims of matching or surpassing Western AI technology at a significantly lower cost.
“DeepSeek has introduced some real innovations,” Nadella remarked during an investor call following Microsoft’s latest earnings report. “Of course, as technology evolves, it becomes more widely accessible and commoditized.”
DeepSeek’s breakthrough has sparked discussions among investors regarding Microsoft’s AI spending. Some have questioned whether the company and its partner OpenAI could achieve similar efficiencies in AI training and inference—key processes in developing and deploying AI—at a lower cost.
According to Nadella, Microsoft has already been focused on optimizing these efficiencies. “We’ve been seeing significant gains in both training and inference for years,” he explained, emphasizing how Microsoft’s software has continually enhanced AI performance and reduced costs across different model generations and hardware.
Much of this progress, Nadella noted, has been achieved in collaboration with OpenAI. He also stressed the importance of cost-effectiveness in AI deployment: “Releasing the best model isn’t enough. If it’s too expensive to operate, it’s not useful.”
Despite ongoing efforts to optimize costs, Microsoft remains committed to AI infrastructure expansion. The company plans to invest $80 billion in data centers this fiscal year to meet growing demand, although it anticipates a slowdown in spending starting in fiscal 2026.
Meanwhile, reports have surfaced regarding Microsoft and OpenAI investigating whether a group linked to DeepSeek may have accessed OpenAI’s data outputs without authorization. However, this has not deterred Microsoft from making DeepSeek’s R1 model available to customers. On Wednesday, the company added R1 to its Azure AI Foundry, a platform featuring over 1,800 AI models designed to support businesses in developing and managing AI applications.