Partnering with you to
either build a new
team or extend your
capabilities.

From ideation and prototyping to your final build and testing, we offer a range of technical services.

Connect to clusters of
like-minded technical professionals
from around the world.

Distance shouldn’t stop
you from accessing
the most exciting
projects.

Articles

Will Mojo Programming Language be a game changer

We explore Mojo in greater detail and assess it’s potential to unseat Python as the language of choice.

image
  • Copy link

Background

Modular a company founded by Chris Modular (creator of Swift) and Tim Davis is focused on unifying ML/AI infrastructure, initially had no intention of creating a new programming language. However, as they built their platform, they realized that programming across the entire stack was too complex and that existing languages did not support the features they needed for AI development.

They wanted a scalable programming model with powerful compile-time metaprogramming and integration of adaptive compilation techniques. They also recognized the importance of the host CPU in AI systems, along with specialized accelerators.

What is Mojo 🔥?

Mojo is a language designed for next-gen compiler technology, specifically MLIR. It aims to unify the AI ecosystem by providing full compatibility with Python while offering improved performance and control over code deployment.

Python's limitations, such as low-level performance and the two-world problem (systems programming vs. glue layer), motivate Mojo's development. Other approaches, like Python subsets and embedded DSLs, fall short in solving these challenges.

Mojo aims to simplify the AI development process, eliminate the need for C/C++ within Python libraries, and provide a unified language for heterogeneous compute. It leverages MLIR and offers incremental migration from Python to Mojo.

In summary, Mojo addresses the complexity of AI programming, provides Python compatibility, and improves performance and control over deployment, offering a unified language for AI development.

Overall, Mojo's design and incremental progress toward Python compatibility aim to solve the challenges faced in AI compute and provide a unified language for AI development.

So, will it take off?

We see so many “next best things” in technology it is hard to know what become mainstream and what will fall away  or remain niche. 

Mojo certainly has some great people behind it. It also seems to be addressing a real world problem that is likely to only get bigger as AI infiltrates more of ecosystem. Furthermore they are committed to open source and already getting a sizeable community behind it.

Python is widely adopted and loved by many, getting people to switch to a new language will be tough. However, given familiarity Mojo will offer Python developers making the switch will be relatively low pain for quite a lot of up side. 

I guess we will have to wait and see but at Remobi we are taking a serious look at it and putting it through its paces.