According to media reports,
Strategic "Liberation": From Strong External Partners to Core Autonomy
This move is seen as a major turning point in Microsoft's AI strategy. Since Microsoft adjusted its cooperation agreement with OpenAI last year,
Autonomy Timeline: Microsoft has set an internal goal to achieve full autonomy in AI capabilities within the next three to five years.
Ecosystem Compatibility: Although shifting toward self-research, Microsoft stated that it will continue to support a diverse ecosystem including external open-source models.
Computing Arms Race: Investing Heavily in NVIDIA GB200 Clusters
To support its "self-research dream," Microsoft's investment in underlying hardware has been nothing short of extraordinary.
Top-End Deployment:
has already begun deploying NVIDIA's latest GB200 chip clusters.Microsoft Computing Power Doubling: According to the plan, its total computing power will increase to the world's most advanced scale within the next 12 to 18 months, providing an endless "fuel" for model training.
First Victory: Lightweight Specialized Models Perform Exceptionally Well
While working on general large models, Microsoft has already made a small test in vertical fields.
Release of Voice Model: On April 2 local time,
released a new voice transcription model.Microsoft Testing Dominance: In tests of 25 major languages, the model outperformed all current competitors in 11 languages.
Application Scenarios: As a lightweight tool, this model will be first applied to office suites such as Teams, significantly enhancing productivity experiences.
Industry Insight: Has the "Honeymoon Period" Between Giants Ended?
Microsoft's move sends a clear signal to the outside world: even the strongest global allies will eventually face each other when it comes to core technology sovereignty. As

