HCLTech bets ₹3,500 crore on owning AI data centres
HCL Technologies — better known as HCLTech — is a giant Indian tech services firm that helps the world’s big companies run their computer systems, write software, and adopt artificial intelligence. On 13 July 2026, its board surprised the market by approving an investment of up to ₹3,500 crore to build and own AI data centres in India, through a new wholly-owned subsidiary. The company’s stated ambition is to become a “full-stack AI” provider, combining the physical infrastructure of AI computing with the design, operations, and software services it already sells. The move is a sharp strategic turn: for at least two years, HCLTech’s management had repeatedly and publicly insisted it would stay asset-light, building IP and services for other people’s data centres rather than owning the data centres itself.
The strategy U-turn
As recently as October 2025, CEO C Vijayakumar told analysts on the Q2 FY26 earnings call: “We strongly believe in an asset light business model. … Our direction is actually going slightly in the reverse. We are going to build more IPs and monetize IPs.” He added that the AI Factory opportunity was “all about servicing large, mega, and giga data centers that are being set up by large tech players … without having to invest in big assets and data centers and real estate.” A year earlier, in July 2025, he had put it even more plainly: “We don’t want to play in the assets kind of game. We are totally focused on services.”
The July 2026 announcement therefore does not extend an existing strategy — it reverses it. HCLTech is now choosing to own the very infrastructure it previously said it would only manage for others. The press release points to a market opportunity: India’s data centre capacity is projected to grow from 1.8 GW today to 5–7 GW by 2030, driven by AI training and inference workloads. Management’s own framework, outlined in the April 2026 call, categorises the IT services market into three segments: “AI disrupted” (shrinking), “AI amplified” (growing ~10%), and “AI native” (growing 30%, including AI Factory and custom silicon engineering). The data centre investment is a bet on the third, fastest-growing bucket.
What the company already does in AI data centres
HCLTech is not starting from zero. From the earnings calls, it is clear the company has been building a services business around AI data centres for several quarters. In the April 2026 call, Vijayakumar disclosed a $100 million-plus AI Factory deal with a global technology major, covering “design, implementation and support of a next-gen AI data center.” The company already had two such clients and hoped to add three or four more in FY27. It has trained and retrained its infrastructure teams and hired lateral talent to work on the new networking and operational tools that hyperscaler-grade AI data centres require. The software portfolio — the recently launched Kinetic AI.DC Ops product, the AI Force platform, and the Actian data intelligence division — is intended to give HCLTech an integrated stack that spans the physical data centre and the software that runs on it.
What was missing until now was ownership of the physical assets. The ₹3,500 crore investment fills that gap, aiming for up to 50 MW of capacity.
The numbers that frame the pivot
An investment of ₹3,500 crore is a leap for a company that has been a model of capital discipline. The table below traces the key financial thread over six fiscal years — the revenue base, the cash it generates, the capital it spends, and the margin it keeps.
The FY26 capex of ₹1,422 crore is the largest in at least two years, yet the new commitment is roughly 2.5 times that figure. The company can afford it — net cash stood at ₹8,106 crore, and free cash flow was ₹18,553 crore in FY26 — so it does not need to borrow and will not strain its balance sheet. But the investment will inflate the capital base just as returns on that capital have begun to soften: ROCE slipped from 30.8% in FY25 to 27.2% in FY26, and the EBITDA margin has been gliding down from a 27.8% peak in FY21 to 21.0% in FY26. The most recent quarter, Q1 FY27, printed an EBITDA margin of 20.9%, the lowest quarterly reading in the entire series stretching back to at least Q2 FY24. Adding a large fixed-asset base with its own depreciation load — annual depreciation was already ₹4,355 crore in FY26 — will test whether the services engine can keep margins from drifting further.
What changes for the investor
The shift from services-only to asset-owning changes the financial profile. In the asset-light model, HCLTech’s services business delivered a return on invested capital of 45.3% as of October 2025, with free cash flow conversion at 125% of net income. A data centre business, by contrast, is capital-intensive: it requires large upfront spending, takes years to fill with customers, and carries depreciation charges that will drag on reported profit even if the underlying cash flows are healthy. The company’s own EBITDA margin has been compressing for years, and the capex step-up adds a new layer of fixed cost that will not immediately produce revenue.
HCLTech is betting that owning the pipes, not just the software that runs through them, will position it as a full-stack provider in a market where digital sovereignty and AI supply constraints are driving demand for local, secure AI compute. The board has approved the bet; the execution timeline and the returns it will generate are not yet in the public domain.
---