IBM India puts sovereign AI and internal productivity claims on the table

IBM India is pitching locally controlled infrastructure as part of India’s AI opportunity, while its managing director says an internal coding agent is used by about 80,000 engineers. Sandip Patel also claims IBM’s automation, process redesign and AI work delivered $4.5 billion in operational savings by the end of last year, more than twice its original target.

Watch Desk analysis

What happened

Patel told The Times of India that IBM’s Sovereign Core can be deployed in a client- or partner-owned data centre, with control-plane operations and key management kept within the customer’s jurisdiction. Built on that software stack, IBM Bharat Core offers reusable Indic AI services, models and connectors, and is designed to guide users towards building blocks and infrastructure for Indian requirements. IBM is also developing India-specific implementation tools and a reference architecture for sovereign AI in government environments.

Patel said IBM’s coding agent, Bob, is being used by around 80,000 engineers and has produced productivity gains of 45%, including among client teams. Those are his account of IBM’s internal results, not an independently measured industry benchmark.

Why it matters

The pitch joins two consequential strands: who controls the infrastructure behind AI, and how much productivity companies can actually extract from using it. For organisations in regulated sectors, keeping data, encryption keys and operational control within a chosen jurisdiction could matter as much as the model itself. IBM’s figures also offer a striking, if company-reported, measure of AI’s promised workplace impact.

Our read

This is more than a sovereignty slogan: IBM describes a deployable software stack and an India-focused set of AI services. But the productivity claim needs the detail behind it. A 45% gain sounds impressive; the task measured, baseline and method decide what it means outside IBM’s own account.

What to watch

• Which Indian organisations deploy Sovereign Core or Bharat Core, and under what terms.
• Whether IBM publishes more detail on how it measures Bob’s claimed productivity gains.
• How the India-specific tools and government reference architecture develop. Activity teaser: IBM India says its sovereign infrastructure software can keep control-plane operations and key management within a customer’s jurisdiction, while its coding agent Bob is used by about 80,000 engineers. The company also claims 45% productivity gains, a figure that deserves the measurement details as well as the headline. Its India-focused Bharat Core combines reusable Indic AI services, models and connectors. Which should matter more when assessing sovereign AI: where the infrastructure is controlled, or evidence that it delivers measurable benefits?

Discussion spark: Should sovereign AI be judged mainly by who controls its data and infrastructure, or by independently measurable results?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.