Yesterday, the programming language that powers fast AI inference on consumer hardware went fully open source. If you have been waiting for local AI to be worth the hassle, that wait just got shorter.
What's happening
Mojo is a programming language built specifically to make AI models run fast on the hardware you already own: your Mac, your Windows laptop, your server in the back room. It was created by Chris Lattner, the same person who built LLVM and Swift, and the company behind it (Modular) released the full compiler and toolchain under an Apache 2.0 open source license on August 18, 2026.
You are not going to write Mojo yourself. That is not the point. The point is that the tools you already use to run AI locally, things like Ollama, are built on this kind of infrastructure. When that infrastructure becomes free, open, and community-improvable, the whole stack below your AI assistant gets faster and more reliable over time, at no cost to you.
What this means practically: local AI (running a model on your own machine instead of sending your data to a cloud) keeps getting more capable without getting more expensive. The gap between what a Fortune 500 company can run and what a two-person shop can run on a spare laptop is closing.
Try this this week
If you have never run a local AI model, now is a great time to start. The setup is about five minutes:
- Download Ollama and install it (Mac, Windows, or Linux).
- Open your terminal and run:
ollama pull llama3.2:3b - Start a session:
ollama run llama3.2:3b - Try something real: paste in a customer email and ask it to draft a reply, or dump your meeting notes and ask for a three-bullet summary.
Your data stays on your machine. No subscription. No API key. No usage limits.
If you want something with a friendlier interface than a terminal, look at Open WebUI, which wraps Ollama in a chat interface that works in any browser.
The bigger picture
The build9 approach is always: smallest working thing, running on what you have, owned by you. Open source AI infrastructure is what makes that possible at the small-team level. Every time a foundational piece like Mojo goes open source, it lowers the floor for what a nonprofit, a local clinic, or a cafe can do without a vendor contract. The tools keep getting better. The cost keeps going down. That is the trend worth watching.