Indian IT services giant HCL has confirmed it is building its own AI datacenter capacity, funded from record quarterly earnings. For UK infrastructure buyers wary of calculate GPU requirements for AI workloads only to be handed a hyperscaler bill, it's a signal worth watching closely.
View the data behind this chart
| Revenue | Net income | Advanced AI | |
|---|---|---|---|
| YoY growth | %3 | %20 | %62 |
What HCL actually announced
HCL revealed its datacenter ambitions alongside Q1 results that showed revenue up 3 percent year-over-year to $3.65 billion and net income up 20 percent to $488 million. The standout figure was HCL's "Advanced AI" segment, covering its own AI platform-building work, which grew 62 percent year-over-year. CEO C. Vijayakumar has committed ₹3,500 crore ($36.5 million) toward facilities with the "potential to scale to 50MW of capacity", and said HCL is already in advanced talks with clients to secure committed consumption from day one.
Why a services firm building its own datacenters matters here
For years, enterprise AI buyers have faced a binary choice: rent capacity from a hyperscaler or take on the capital risk of building it themselves. HCL's move suggests a third model is emerging — global systems integrators owning the infrastructure layer and wrapping it in their own software and managed services. Vijayakumar's own framing is blunt: "The biggest opportunity is not to rent AI, but to own the full stack." UK buyers evaluating vendor strategy should note that this full-stack pitch — combining datacenter design, DevOps and cloud operations with an existing software portfolio — is precisely the pitch increasingly used to justify moving workloads off public cloud altogether. It's worth reading alongside the case to explore the business case for AI workload repatriation before assuming rented hyperscaler capacity remains the default.
The scale reality check
A planned 50MW ceiling sounds modest set against Meta's own facilities, some of which are being built to gigawatt scale. That gap matters for UK procurement teams: HCL's own datacenters won't compete on raw scale with the largest cloud providers any time soon, and the company hasn't disclosed where the facilities will be built, when they'll come online, or how energy supply will be secured. Vijayakumar has been explicit that the initial focus is India's "sovereign AI ecosystem", positioning the investment around secure AI and managed AI infrastructure for that market rather than international expansion. UK buyers shouldn't expect HCL capacity on British soil in the near term, but the model — services firms owning infrastructure rather than purely reselling hyperscaler slots — is one that could reshape how similar vendors pitch to UK enterprises.
What this means for vendor selection today
The more immediate UK-relevant signal is HCL's booking momentum: $2.4 billion of new business in the quarter, a record, including a deal with an unnamed Fortune 250 semiconductor equipment OEM to build an AI-led digital supply chain backbone using SAP integration. Separately, reports point to HCL winning a Europe-headquartered Fortune Global 50 client's digital workplace and network transformation business, reportedly switching from Infosys. That kind of enterprise-scale AI transformation work is exactly the territory UK buyers are assessing when they weigh services partners against direct infrastructure ownership. Before committing to any vendor's full-stack story, it's worth running the numbers independently — teams can compare the TCO of cloud vs. on-premise AI infrastructure to see where the economics genuinely favour owned capacity versus rented.
Building the alternative to hyperscaler lock-in
HCL's entry adds to a growing list of non-hyperscaler players positioning themselves as full-stack AI infrastructure providers, even if this particular build targets India first. UK organisations that have been quietly assembling their own on-prem AI capability, partly to avoid unpredictable cloud egress and reservation pricing, will recognise the logic. Those starting that journey should first build your first UK on-prem AI cluster with a realistic view of costs, and understand the cost of AI servers before benchmarking any vendor's full-stack claims against what's achievable in-house.
Practical next steps for UK buyers
None of HCL's disclosed plans change UK procurement immediately — no facility location, timeline or energy strategy has been confirmed. But the direction of travel, from a major global IT services player with $3.65 billion in quarterly revenue, reinforces that owning infrastructure alongside software is now seen as the higher-margin play. Buyers should use this moment to stress-test their own hyperscaler dependency, price out the alternatives, and where hardware decisions are imminent, source GPU accelerators for AI deployments through channels that keep procurement options open rather than locking into a single vendor's roadmap.