Weeks after DeepSeek shook the AI market, Red Hat answered calmly. It named the problem first: the licence does not say how the model was made. Then it drew its conclusion. The future of AI is open, built on smaller, optimised models that a company tunes to its own data and runs wherever it wants. Soon after, at Red Hat One in Las Vegas, DeepSeek was a recurring topic.
I have seen this film before. Ten years ago it was Kubernetes, an open project that changed how everyone runs software. Now it is models. You no longer need the biggest one, or an army of data scientists. You need a good small model, your own data and a platform you control. The second wave follows the script of the first, only faster. We stand on the shoulders of giants, and the giants are open source.