Can you share a brief introduction to HCLTech, covering its core services and regions?
HCLTech has been in business for more than 25 years, though the HCL Group dates back to 1976. Around the same time Microsoft and Apple launched, we began with digital word-processing machines and later built server-class systems for banks and research centers in India. By 1998–99, we recognized the growing demand for services and spun off what is now our global business, making us one of the fastest-growing players in the space.
Today, financial services, telecom, media, and technology account for about 51% of our business, with retail, healthcare, manufacturing, energy, and public services also significant. We started in engineering and still build products such as implants, medical devices, aviation software, and automotive technology, partnering with leaders like Microsoft, Google, Apple, NVIDIA, and Intel. Alongside engineering, we run a $4.5 billion applications business, a global infrastructure arm, BPO services, and a $1.2 billion software portfolio.
How is HCLTech implementing AI internally?
We have worked with AI for eight to nine years, using it to improve efficiency in software engineering and IT operations—but generative AI has been a game-changer in the past three. It helps us deliver services faster: products reach the market quicker, applications run more reliably, and networks are automatically managed. That reduces labor, increases predictability, and raises quality. We have also built AI-based platforms we now sell through our software arm.
We were early in designing AI chips. For example, when Apple’s privacy policy limited Meta’s access to ad-targeting data, Meta lost market share. To regain competitiveness, they built training and inference chips to run video models. We helped them design and deploy those chips in data centers, and today we run some of those centers. That gave us a critical early advantage by working with AI leaders at scale.
In the past two years, we have integrated generative AI into our platforms. Our AI Force platform helps customers adopt GenAI, and we use it internally to accelerate delivery. Code generation improves productivity by about 30% in coding, but that covers only a fraction of the software lifecycle. AI Force automates the full cycle from requirements to deployment, modernizing legacy systems—some 30 to 40 years old—at scale. We are embedding generative AI into IT operations and support, and later this year our next products will be agentic, coordinating across multiple agents and protocols.
How is AI impacting the way that HCLTech delivers its services to clients?
The real value comes through what we call “value stream innovation”—generating new client value with AI. We do this in three ways. The first is AI Foundry, which builds infrastructure—chips, data centers, and data management—so AI applications can run effectively. In healthcare, for instance, we automated parts of diagnostics. Doctors receive AI-driven recommendations, saving three minutes from a typical 20-minute consultation. With 1,000 doctors on the platform, scaling to 20,000, we estimate over $50 million in value for the customer, alongside better patient care.
Second, we run six labs worldwide where customers identify high-impact use cases, create proofs of value, and move them into production. Third, we focus on engineering and “physical AI”—using video models in ports, yards, and utilities to improve safety and efficiency. In New South Wales, we analyze video feeds to assess water quality remotely, avoiding constant manual checks, for example. Altogether, our GenAI portfolio spans four offerings: AI Force, AI Foundry, AI Labs, and AI Engineering.
In the past, you have advocated for Indian IT to build its own LLMs, not rely on US models. Why is that?
Large language models are expensive—GPT-4 reportedly cost around $200 million—and without a massive business case, it does not make sense. Instead, we are focused on building small language models (SLMs) for specific purposes. For example, our infrastructure business has the data to train an SLM that can predict and prevent IT downtime. In telecom, where networks are still manually managed, an SLM could monitor towers, detect failures, and help diagnose issues remotely.
That said, LLMs still have their place. IIT Bombay is building “Vashini” for real-time translation across India’s languages, something large LLMs do not fully support. But for global companies like us, given the pace of innovation and capital intensity, a more efficient approach is to partner on LLMs while building SLMs for targeted use cases.
How do you evaluate India’s sovereign AI ecosystem compared with other markets?
There are two sides to AI capability. The first is research—the deep work required to build models. That requires depth, not scale. OpenAI, for instance, has maybe 200 researchers. On this front, the U.S. and China lead. India is probably in the top 10, but not the top three.
The second side is the developer ecosystem, and here India is very strong. With its education system feeding a large developer base, I would place India in the top three alongside the US and China. Internally, we are retraining and hiring aggressively. By year’s end, 90% of our 180,000–200,000 employees will be able to use AI systems, and we expect to have 8,000–10,000 AI developers within the company. India may not yet lead in research, but it is absolutely among the leaders in AI development talent.
Today the big conversation is around agentic AI. How are you seeing these technologies evolve?
Agentic AI is everywhere—you see it on every billboard along Highway 101 in the Bay Area.
Generative AI is about generating, but agentic AI goes further: it makes decisions, learns over time, and rethinks workflows.
It can be human-in-the-loop or not, and it automates across complex systems like SAP or Oracle.
The challenge is orchestration and standards. We are working with MIT on NANDA, a protocol that goes beyond Google’s A2A or Anthropic’s MCP by enabling multiple protocols to interoperate. It is open source and still early, but the goal is seamless ecosystem communication. For us, being in the agent orchestration space positions both HCLTech and the wider community to help shape how agentic AI evolves.
What other major technology trends are you watching this year?
A big trend is AI factories. Customers are increasingly investing in private AI infrastructure with companies like NVIDIA. These “factories” integrate GPUs, data centers, models, and applications specific to each business. We estimate there are about 100 AI factories today, but that number could triple or quintuple in three years.
