Qualcomm has opened up access to its Mojo programming language, releasing it to the open-source community. The language was originally created by Modular before the chipmaker acquired the startup.
The Journey From Modular to Qualcomm
Modular first introduced Mojo in 2023 as a superset of Python engineered for high-performance computing on modern hardware. The company raised significant venture funding and attracted attention from leading AI researchers. Qualcomm acquired Modular in 2024, folding the team into its broader AI and machine learning division. Now Qualcomm has made the full source code of Mojo available under an open-source license.
Why This Matters
The open-sourcing of Mojo lowers barriers for developers building custom AI pipelines. Instead of relying on proprietary tools, teams can now compile, optimize and deploy models using a language that directly targets GPUs and NPUs. For Qualcomm, the strategy strengthens its ecosystem for edge AI by giving developers more control over performance tuning. Competitors such as Nvidia and AMD face pressure to offer similar openness in their software stacks.
Smaller startups and academic groups stand to benefit most. They gain access to a production-grade language backed by a major hardware vendor without paying licensing fees. The shift also accelerates experimentation in AI compiler design and runtime optimization.
What Developers Can Expect
The open-source release includes the Mojo compiler, standard library and documentation. Contributors can submit patches, propose features and port Mojo to new hardware targets. Early benchmarks suggest Mojo achieves up to 68,000 times speedup over pure Python on certain tensor operations when compiled correctly.
Developers already working with Python will find a familiar transition path. Mojo preserves most Python syntax while adding type system enhancements, ownership semantics and explicit parallelism controls. The community has already begun packaging Mojo for Linux, macOS and Windows through package managers.



