Artificial intelligence (AI) is forcing India's IT services companies to rethink how they grow.
HCL Technologies Ltd (HCLTech) is entering the data-centre business, becoming only the second Indian IT services company after Tata Consultancy Services Ltd (TCS) to do so. But while TCS is taking an infrastructure-led approach, HCLTech is betting on a full-stack AI strategy spanning data-centre infrastructure, compute and applications.
Investors, however, remain more concerned about HCLTech's slowing growth than its longer-term AI strategy. On the BSE, HCLTech's shares fell 4.42% to close at ₹1,167 a piece on Tuesday, a day after the company unveiled the strategy alongside its weakest full-year growth guidance in four years.
The company plans to invest about ₹3,500 crore ($365 million) to build and operate data centres with up to 50MW of capacity. TCS last October committed $6.5 billion over six years to build 1GW of data-centre capacity.
Building the full AI stack
The similarities end there. For HCLTech, the investment is about controlling more of the AI value chain rather than merely hosting AI workloads.
"Our whole value is in delivering full-stack AI services, which means it's data centre, it's the GPUs, it's the models, it's the applications that we will deliver on top of it. So the overall value creation is significantly of a very different magnitude when you really look at this as a full stack, and that's really what we want to play," said C Vijayakumar, chief executive of HCLTech, during the company's post-earnings press conference.
“It's a significant difference between the two. TCS offering is aimed at large hyperscalers and they offer infrastructure services such as land, power, cooling facilities, compute and network services to their clients to help them run their AI workloads. These hyperscalers include likes of Google, Microsoft, AWS, and AI-native firms including OpenAI and Anthropic,” said Ashutosh Sharma, vice-president at Forrester Research.
“On the other hand, HCL is focusing on offering a full stack of sovereign AI services up from the underlying datacenter. The company aims to complete this compute layer with foundational AI capabilities through their partnership with Sarvam.ai. Further HCL will bring their AI orchestration solution and talent layer to compete the entire stack,” added Sharma.
Last month, HCLTech became the country's first IT services company to acquire a stake in an AI startup after buying 10% of Bengaluru-based Sarvam AI for about $150 million. Sarvam develops AI models in Indian languages.
The companies are also targeting different AI workloads.
“TCS is working with large AI companies and is looking to sell AI compute power through the likes of OpenAI, which would be more capex and compute intensive,” said Sushovon Nayak, lead IT analyst at Anand Rathi Institutional Equities.
“On the other hand, HCLTech is primarily focussed on handling customer work through small language models, which require lesser compute power and more cost effective as they are built around specific client needs,” said Nayak.
Broadly, data-centre providers follow one of two models. One rents out physical infrastructure—land, power and cooling—while customers deploy their own servers or use cloud services from hyperscalers. The other combines infrastructure with AI compute capabilities, the model HCLTech is pursuing.
India's data-centre market is also expanding rapidly. The country is expected to have about 7GW of capacity by 2030. Conglomerates including Reliance Industries Ltd, Adani Enterprises Ltd and Hiranandani Group, along with companies such as Airtel and Uber, have entered the sector.
Betting on AI, battling slower growth
HCLTech's move comes as AI begins reshaping the economics of traditional IT services. The company has outgrown TCS over the past three years and was the fastest-growing among India's top five IT firms last fiscal year, even as management warned AI is already causing pricing deflation of up to 3% in parts of its business.
Brokerages said HCLTech's strategy positions it to benefit from the shift.
“We believe HCLT (HCLTech) is investing ahead of the market to build the next-generation AI stack,” Motilal Oswal Financial Services analysts Abhishek Pathak and Keval Bhagat said in a note dated 13 July.
“We believe a clearer playbook is now emerging through HCLT's five pillar AI strategy—transforming services, building differentiated IP, expanding AI-led services, strengthening AI partnerships, and scaling AI talent. This should help the company defend against, and eventually benefit from, AI disruption, and HCLT seems ahead of the curve here,” the analysts added.
Like TCS, HCLTech does not expect to fund the investment on its own. Management said on Monday it was exploring financing through silicon companies, OEM vendors, and a mix of debt and equity.
Bank of Baroda Capital Markets said the strategy could also help HCLTech control AI costs as clients increasingly seek outcome-based pricing.
"The AI data centre and AI model related investment indicates a desire to control the cost of compute and of 'intelligence' as clients become more price sensitive and want to sign output-based agreements with their vendors," said analysts Girish Pai and Lopa Notaria in a note dated 14 July.
Markets want proof
The market, however, remains focused on HCLTech's near-term outlook.
The company reported revenue of $3.65 billion for the June quarter, down 0.9% sequentially, and guided for constant-currency revenue growth of 1-4% for the full year, its weakest July guidance in four years. Management also flagged persistent macroeconomic uncertainty as a continuing headwind.
"HCLT has in the past called out revenue deflation with an overall 2-3% impact on its portfolio due to AI related deflation. This sets a negative outlook for the next few years and not increasing the FY27 revenue growth guidance despite better than expected revenue growth and strong TCV numbers leads us to believe that revenue deflation will be seen sooner than expected," said Nirmal Bang Institutional Equities analysts Shubham Dalia and Suket Kothari in a note dated 14 July.