Sovereign AI
Also called: Indigenous AI · Topic: International Trade Policy, WTO and Intellectual Property · NCERT: Beyond NCERT
Meaning
Sovereign AI means a country builds and controls its own artificial intelligence. It owns the AI models, the data used to train them, and the compute (the computer power, mostly GPU chips, that trains and runs the models). The aim is technological and strategic autonomy. This means India should not depend on a foreign company that can change the licence, cut off access or see sensitive data. It matters because AI is becoming basic infrastructure, much like power or telecom. It also has a trade angle, because the chips usually still come from abroad. Export controls on advanced chips can therefore limit how far a country can go alone.
Example
Under the IndiaAI Mission (run by MeitY, the Ministry of Electronics and Information Technology), the government pays for part of the cost of compute. Private Indian firms then build the models. Soket's EKA-1 covers all 22 scheduled languages. ZenteiQ's Dhīr models run offline on mobile phones, so defence data never leaves the device. But these models are still trained on imported GPUs. So India now controls the model and the data, but not yet the chip.
Don't confuse with
- Data localisation: a rule that data must be stored or processed inside the country, as in RBI's April 2018 rule on payment data. It covers only the data, while sovereign AI also covers the models and the compute.
- Self-reliance: the general goal of cutting dependence on imports across the economy. Sovereign AI is that same goal applied to one technology.