Is Mojo the Next Big Leap for AI Programming? Performance Claims Up to 35,000× Faster

Python is the most popular language in artificial intelligence and machine learning, and its popularity is expected to grow as AI expands across industries. Julia is another language some developers choose, and a new competitor, Mojo, has recently been announced by AI infrastructure company Modular AI.

Mojo combines Python's syntax with C's portability and performance, aiming to become the ideal choice for AI research and production.

Do we want the power of AI to be in the hands of a few locked behind an API or do we want it to be accessible to all of the world? At Modular we take the stance that the world needs fair and equitable access to AI.

Mojo merges Python's ease of use with C's performance, unlocking unparalleled programmability of AI hardware and scalability of AI models, and it is claimed to be 35,000 times faster than Python.

Mojo combines Python's usability and C's performance, unlocking AI hardware programmability and model scalability. It is also said to be 35,000× faster than Python.

The Mojo development team explains that they did not initially intend to create a new language, but while building a unified ML/AI infrastructure platform they realized the existing stack was too complex. Their goal is an innovative, scalable programming model that supports accelerators and heterogeneous systems common in machine learning, featuring strong compile‑time metaprogramming, adaptive compilation, caching throughout the compilation process, and capabilities not found in existing languages.

Jeremy Howard, after testing Mojo, wrote in a blog post that it might be the biggest programming breakthrough in decades, recalling the excitement he felt when first using Visual Basic v1.0.

Mojo leverages MLIR, allowing developers to exploit vectors, threads, and AI hardware units, with tests showing a 35,000‑fold speed increase over Python.

Modular AI’s founder is Chris Lattner, the renowned compiler expert and creator of Swift, who announced the company last year and now leads it as CEO.

Chris Lattner’s career includes leading major projects such as LLVM, Clang, MLIR, and contributing to Swift, as well as heading AI compiler and runtime teams at Google Brain and TensorFlow.

2011‑2017: Senior Director & Architect, Apple Development

Jan‑Jun 2017: Vice President, Tesla (Autonomous Driving)

Aug 2017‑Jan 2020: Senior Director, Google Brain

Jan 2020: Joined SiFive, Platform Engineering

In January 2022, Lattner co‑founded Modular AI with Tim Davis to rebuild global ML infrastructure from the ground up, serving as CEO.

Industry observers are already calling Mojo “the biggest programming advancement in decades.”

Trial address: https://docs.modular.com/mojo/get-started.html?continueFlag=407d506a7e76ce1b185a9a2faef09e43