TCS is making the boldest bet in its tier. It wants to grow a $30 billion business while shrinking its dependence on the 584,000 people who deliver it, by putting AI agents alongside its associates and selling the result as outcomes instead of hours. The clearest sign of conviction is the June 2026 deal with Anthropic, the makers of Claude. TCS became a Global Premier Partner, stood up a dedicated Claude business unit, and is equipping 50,000 of its own people with Claude first. That is the largest single commitment to the Claude channel anywhere in the corpus.

That means TCS uses itself as the test client. Engineers, finance, legal, marketing and sales staff run their daily work on Claude, and the lessons learned get packaged into offerings for clients in regulated industries like banking, healthcare, aviation and telecom. Around it sits a wider toolkit: the WisdomNext platform that picks which AI model to use, HyperVault data centers for clients who need their AI kept on sovereign infrastructure, and parallel programs with Microsoft, Google and NVIDIA so no single vendor owns the stack.

The same scale that makes TCS formidable also makes the bet risky. It has the largest labor base in the sector, which means it has the most revenue to cannibalize if agents really do replace billable effort. Growth is already negative in constant currency, renewals are coming under price pressure, and the firm still bills the overwhelming majority of its work by time and materials.

The verdict: TCS has the deepest AI proof points in its peer group and a margin at a four-year high, yet investors stayed skeptical. The stock fell to a six-year low and lost the premium it had held over peers for more than a decade, the price of asking investors to believe a labor-heavy giant can rewire its own economics before the deflation reaches its renewals.