As AI programming gradually becomes the core tool for programmers around the world, the operational costs of underlying large models and market competition are undergoing a subtle transformation. According to insiders, due to the high costs of purchasing and operating third-party large models, Microsoft has planned to break the status quo and officially launch its self-developed AI programming large model at the upcoming annual Build developer conference.
This strategic shift had already shown signs earlier. Microsoft CEO Nadella had previously asked internal employees to stop using Anthropic's Claude model entirely by June and switch to its own GitHub Copilot. Although the official reason given at the time was that the cost of calling Claude was too high and no longer cost-effective, considering the latest developments, this move clearly aimed to pave the way for Microsoft's upcoming self-developed programming model.
As the highlight of this year's Build conference, this self-developed programming large model will be directly used to upgrade GitHub Copilot, facing off against strong competitors such as Cursor and Claude Code in the developer community. In addition to focusing on the high-value programming niche, Microsoft also plans to simultaneously launch different parameter specification self-developed large models covering multiple vertical fields, including transcription, reasoning, speech, and image, to comprehensively deepen its full-stack AI capabilities.
Looking back at Microsoft's AI strategy, the risks of over-reliance on a single technology supplier once constrained it significantly. As the largest investor of OpenAI, Microsoft had signed an exclusive non-disclosure agreement that restricted its ability to develop human-level AI before April 2026. It wasn't until April this year, when the cooperation terms were restructured, that Microsoft regained full research freedom. During this period, to support the AI operations of core businesses such as GitHub, Azure, and Office365, Microsoft had to purchase large amounts of Claude series models. However, the high computing costs were even too much for a trillion-dollar tech giant like Microsoft to bear.
For global developers, Microsoft's release of a self-developed programming model is undoubtedly a major positive development. Currently, when GitHub Copilot calls third-party large models such as Claude and GPT, it generally faces billing premiums of 3 to 5 times, or even up to 10 times. Once Microsoft introduces a highly competitive self-developed alternative, it not only can greatly reduce its own operational costs but also directly lower users' cost barriers.
Currently, the arms race in the AI programming sector has entered a feverish stage. In addition to Microsoft's full-scale counterattack, Elon Musk's xAI is also continuously investing, with a new large model containing at least 1.5 trillion parameters set to be integrated into Grok soon. As technology giants continue to reveal their core cards, the global AI programming market is about to experience a fierce reshuffling driven by both technology and cost.
