AI's Next Power Shift: Who Wins When the "Rent" Stops?
Sovereign AI and open weights are reshaping enterprise AI. Discover who wins when companies stop paying model rent and take control.
“AI Disruption” Publication 10,400 Subscriptions 20% Discount Offer Link.
Bigger models are better, more compute is better—enterprises can simply connect to an API and enjoy the most advanced intelligence.
Enterprises don’t need to train models themselves, nor build their own AI infrastructure. As long as they pay by the token, they can directly use the world’s strongest large models.
This model has indeed driven the fastest wave of generative AI adoption.
But by 2026, I’m increasingly paying attention to another emerging shift:
Enterprises are no longer satisfied with merely “renting AI”—they want to take control of AI back into their own hands.
The key term behind this is one that has been appearing more and more frequently: Sovereign AI.
And if this trend truly takes hold, it will change not only how enterprises use AI, but may also reshape the profit distribution across the entire AI industry chain.
Previously, using AI was like renting a house. Now enterprises are starting to want to “buy a house.”
At first, the way enterprises used AI was very simple.
They sent data to the remote end, called APIs provided by vendors like OpenAI or Anthropic, and paid according to usage.
In the beginning, the economics were easy to calculate.
Employees occasionally asked a few questions or generated some text; each call consumed only a small number of tokens. Compared with training models themselves, the API cost was almost negligible.
