5 Key Takeaways
- Infosys says AI services crossed 8% of revenue and could make up nearly half of its work within two to three years if current growth continues.
- AI services at Infosys include building AI agents, coding work, and modernizing older technology systems, driven by strong client demand.
- Infosys is backing the shift with hiring—20,000 college graduates last year and another 20,000 planned this financial year—while about 80,000 employees already work on coding tools.
- Enterprise clients are cost-conscious and exploring lower-cost tokens and open-weight models, while sovereign data requirements and data center constraints shape capacity build-outs.
- Parekh warned of longer-term risks including expensive chips, uncertain infrastructure usefulness, cheaper inference from new foundation models, and social backlash that may become visible in five to ten years.
Infosys CEO Says AI Services Could Account for Nearly Half of Company’s Work Within 2-3 Years
Artificial intelligence is no longer a peripheral experiment at Infosys. The company’s AI services crossed more than 8 percent of revenue in the last quarter and are growing at a double-digit rate sequentially. Salil Parekh, Infosys Chief Executive Officer and Managing Director, said on Tuesday, October 13, 2026, that if the current pace continues, AI work could make up almost half of all Infosys activity within two to three years. That would be a remarkable shift for a company built on traditional software services and outsourcing.
Parekh shared the outlook at the Bloomberg New Economy Forum 2026. He described an AI services business that is still small relative to Infosys’s overall size but expanding quickly. The category includes building AI agents, coding work, and modernizing older technology systems for clients. In simple terms, AI agents are software programs that can perform tasks, make decisions, or interact with other systems on behalf of a user.
The CEO put the growth in perspective by looking at the likely trajectory.
“It (AI services) is growing at a fantastic pace. If you sort of extrapolate that by a few quarters, you already start to see that’s 20-30 per cent of our work in a few quarters, and maybe in two or three years, almost half of our work,” Parekh said.
The sequence would imply a rapid compounding effect as more client work shifts from traditional maintenance and development toward AI-enabled projects. It would also move AI from a single-digit share to roughly one-fifth or one-third of Infosys’s work within a short period.
Parekh said the expansion reflects “a huge demand” for AI services among clients. For years, analysts and investors have debated whether artificial intelligence would threaten India’s IT services industry, which has historically relied on large teams of engineers managing software, maintenance, and back-office processes. Parekh rejected the pessimistic view. “My own sense is it is good news for Infosys, and by sort of that same view, good news for the tech services in India,” he said.
The company is backing that belief with hiring. Infosys hired 20,000 college graduates last year and plans to hire another 20,000 this financial year. The new graduates are trained in AI and then work with clients to deploy the technology and manage organisational change. Overall, Infosys has around 3 lakh, or 300,000, employees. About 80,000 of them are already working on coding tools for clients and on some internal projects.
Parekh said enterprise clients are already good users of AI services. Many large businesses are not waiting to adopt the technology; they are actively using it in areas such as software development and system updates. At the same time, clients are looking for ways to control costs as they scale up. With strong demand and capacity still being built, some are exploring lower-cost tokens, which are the basic units of data that AI models process. They are also considering cheaper or open-weight models instead of relying only on the most recent commercial systems.
Open-weight models make their underlying parameters available for developers to use and adapt, which can reduce dependence on a small number of closed, proprietary platforms. Parekh said this cost-consciousness is part of the maturing of enterprise AI. It suggests that customers see AI as a core tool rather than a novel experiment. If cheaper models can deliver similar results, businesses will naturally move toward them.
The CEO also addressed whether physical infrastructure can keep pace with the surge in AI and cloud computing. He said data has become critical to the AI ecosystem. Stricter sovereign data and cloud requirements, which often require information to be stored and processed inside a specific country, are further restricting some of the capacity build-outs. This means the location and design of data centres are becoming as important as the software running inside them.
Data centre expansion carries its own risks, Parekh warned. Chips are “massively expensive” today and facing tremendous competition. It is unclear how long the capacity being built would remain useful, especially if hardware improves quickly or if demand shifts. The economics of building large-scale computing facilities can change fast, and expensive infrastructure may not always align with future needs.
New foundation models could also reshape the market. Foundation models are large, general-purpose AI systems that can be adapted to many different tasks. Parekh said new foundation models taking a different approach from the two large closed-model companies could make inference costs several times cheaper. Inference is the cost of running a trained AI model to produce answers or complete tasks, as opposed to the cost of training it in the first place. Cheaper inference would affect overall capacity planning.
Despite the enthusiasm, Parekh cautioned that the AI build-out faces longer-term risks. “Today the demand is fantastic, but these are some of the risks that could play out,” he said. He added that the impact of those risks could become visible five to ten years later. The industry is currently in a heavy investment phase, and the full consequences of that spending may not be clear for years.
Another risk is social backlash, especially in Western markets. Parekh said the social response to AI build-out needs to be balanced with the economic drivers, which “may be out of sync right now.” This points to a tension between the rapid commercial push into AI and public concerns about job displacement, privacy, energy use, and inequality. The comment suggests that companies and policymakers cannot focus only on growth without addressing public concerns.
If Infosys’s AI services do reach nearly half of its work within two or three years, it would mark a structural shift for the company and for India’s broader technology services sector. The industry has long been defined by large-scale software delivery and maintenance work. AI could turn it into a provider of intelligent automation, continuous modernization, and custom AI tools. The transition is already visible in the company’s workforce strategy and client conversations.
The next few quarters will be a crucial test. Infosys expects AI services to move from just over 8 percent of revenue to a much larger share if current growth holds. Hiring remains strong, and clients are actively adopting AI services. But infrastructure constraints, chip costs, model pricing, and social acceptance will all shape whether the forecast becomes reality. For now, the message from Infosys is clear: AI is not a threat to its business, but the engine of its next phase.
No comments:
Post a Comment