Mphasis CEO Nitin Rakesh warns against ‘race to the bottom’ pricing in AI era

Nitin Rakesh, CEO of Mphasis

• Mphasis CEO warns IT firms against aggressive discounting amid AI adoption
• Clients want AI-enabled services focused on productivity and measurable outcomes
• AI to disrupt IT operations more than software engineering jobs

Did our AI summary help?

Mphasis CEO Nitin Rakesh, on January 23, cautioned IT services companies against chasing aggressive discounting as enterprises accelerate adoption of AI-led delivery, arguing that the real shift is not about cutting costs alone but rebuilding how services are delivered.

"The race to the bottom is a very risky race," Rakesh said at the company’s post-earnings press conference, held after nearly half a decade. "Let’s cut the price is a very discreet conversation. You have to fundamentally go after redoing how you deliver the service itself."

His comments come even as Mphasis posted steady quarterly growth in Q3 FY26. The company’s revenue rose 2.6 percent sequentially and 12.4 percent year-on-year to a little over Rs 4,000, while operating margin came in at 15.2 percent for the quarter.

Cutting price vs cutting effort

Rakesh said customers are no longer satisfied with vendors merely optimising the cost of people. Instead, they are pushing for vendors to reduce the effort required to deliver outcomes, especially in "run" operations, where automation is increasingly becoming the lever.

"Run" operations or "run" spending refers to day-to-day IT operations work required to keep systems running, such as monitoring, maintenance, support, and incident management.

"There is only that much you can cut costs on when you're focusing on optimizing the cost. But not the effort," he said. "You're trying to optimize the people cost. Without saying OK, can I eliminate these 30-40% of the people required to do this?"

He also pushed back against fears that AI will stop enterprises from writing software, flagging that modernisation demand is large and long-lived.

Modernisation refers to updating older enterprise applications and infrastructure by rebuilding, re-platforming, or re-architecting them to work better with cloud, data, and AI-ready systems.

"If I just look at the amount of legacy that sits in, all of that has to be rewritten. That itself is a hundred billion dollars-worth of opportunity," he said. "That's where all the money will go that is coming out of the run."

Legacy refers to older, mission-critical systems and applications that companies still depend on but which are expensive and complex to maintain or upgrade.

Clients want AI-enabled services, not people-enabled

Rakesh said enterprises are starting to measure service providers on productivity and measurable outcomes, even when the commercial model still resembles an hourly rate structure.

According to the latest numbers, Mphasis’s billing remains largely traditional, with time-and-material (T&M) contracts accounting for 46.8 percent of revenue in Q3 FY26, fixed-price projects making up 44.8 percent, and transaction-based work contributing 8.4 percent.

T&M means billing is based on the number of hours or days spent and the resources deployed, fixed-price means a committed fee for a defined scope, and transaction-based means pricing linked to volumes such as claims processed or tickets resolved.

Citing an example from a high-tech client, he said a new CIO moved to bring in productivity measures by setting GitHub metrics for contractors and service providers.

GitHub metrics are software engineering activity signals tracked on GitHub, such as code commits, pull requests, reviews, and issue closure, used to measure delivery velocity and throughput.

“They will still pay by the hour, but they're expecting a certain outcome powered by AI,” he said, adding that the client is using tools including Gemini’s Copilot and Anthropic’s Claude.

In another engagement with the same customer, he said the client pushed to separate services from SaaS and demanded a shift in how operations work is delivered.

He argued that the company should move towards AI-enabled versus people-enabled, or risk losing relevance. “I want to get to 100 percent (AI-infused) of my revenue, I will lose that revenue,” he said. “If I'm delivering with people… somebody else is going to come in and eat that up.”

Run spend squeeze is the mega trend

Rakesh also warned that the biggest AI-led disruption will play out in operations and “run” spending, rather than in software engineering headcount reduction.

He said he has not seen enterprises planning to cut engineers because they are using agent tools, but is seeing pressure on run operations.

“I have not seen any enterprise telling me I'm going to reduce the number of engineers,” he said. “What they are telling us is I have a backlog that goes out two years. Can I eat into the backlog faster.”

At the same time, he said customers are coming to service providers asking them to take on responsibility for driving productivity-led savings.