Open Source AI Is Booming. The Money Is in Everything Around the Model.
The model is free. The bill isn't — and that's the whole open-source AI business model. Everyone keeps announcing "open" models like the download is the product. It isn't. The weights are free; everything that turns a download into something a real company can actually run is where the money changes hands. If you're still arguing about which model is "most open," you're missing the only question…
Open source artificial intelligence is experiencing a boom, but the real money lies not in the models themselves, but in the surrounding ecosystem. While companies often tout their "open" models, the truth is that the model weights are free, while the real value comes from the additional services required to put the model into production.
These services include hosted inference, managed cloud deployment, enterprise governance, fine-tuning, commercial licenses, and consulting. The top 200 open models, representing just 0.01% of the 2 million models on Hugging Face, account for nearly half of all downloads. Developers are increasingly turning to Chinese companies like DeepSeek and Alibaba's Qwen for their open models, rather than those offered by American labs, whose releases are more limited and voluntary.
While these open models may seem like a "free lunch," they come with their own set of challenges and costs, such as the need for specialized hardware, development resources, and regulatory compliance. As the market becomes more concentrated, profits are increasingly flowing to those who control the infrastructure, distribution, and trust around these models.
This shift in business model highlights the need for companies to carefully evaluate their specific requirements and validate the performance of open models against their own data before making a commitment.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
