After decades of layering new software on top of ageing enterprise systems, companies are running into the limits of legacy technology, believes Mphasis CEO Nitin Rakesh.
In a detailed post-earnings interaction, Rakesh said the IT services firm is positioning its NeoIP platform—enabling agentic AI-led IT operations and observability—to help enterprises rethink how they modernise and operate long-standing core systems, using AI to extract intelligence and drive change, rather than simply automate around legacy constraints.
Rakesh traced the challenge back to the evolution of enterprise computing. Early software automated manual record-keeping in sectors such as banking and insurance. While the digital and cloud era improved reach and experience, core systems, many built in the 1970s and 1980s and still running on mainframes, largely remained untouched.
Enterprises, he added, have continued to add layers without fundamentally transforming systems of record. “We keep putting lipstick on the same thing,” he said in the Q3 earnings call press meet.
That model, Rakesh argued, has increasingly become unsustainable as real-time payments, instant onboarding and always-on digital experiences put pressure on systems never designed for such speed or scale.
Modernising By Shrinking the Core
Rather than advocating wholesale replacement of core systems, Mphasis focuses on extracting business intelligence from legacy platforms using data and AI to drive customer and operational experiences. This approach progressively reduces the complexity of the core until it becomes feasible to modernise.
Enterprises, Rakesh noted, often lack up-to-date documentation of how their systems actually work. Over decades of changes, the most accurate representation of business logic is not on paper anymore; it’s the code itself.
AI, he argued, makes it possible to relearn that logic far faster than traditional manual methods.
Rakesh cited the case of a large modernisation programme involving tens of millions of lines of legacy code. Where conventional approaches would have taken several years just to reverse-engineer business rules, Mphasis’ AI-assisted techniques compressed that phase by roughly 80%, completing it in a matter of months through iterative analysis.
Beyond “Break-Fix”
Rakesh said the same approach is extending into IT operations. Traditional “break-fix” models, where systems are repaired only after failures occur, are increasingly misaligned with AI-era expectations.
“Even your dumb car has a yellow light that tells you something is going wrong,” he said. “But our smart IT operations don’t do that even today.”
The goal, he said, is to move operations toward more predictive and preventive models, where incidents are anticipated and addressed earlier, reducing downtime and manual intervention over time.
Resetting Pricing And Delivery Models
These changes are also reshaping how Mphasis structures commercial engagements.
Rakesh said traditional IT services pricing, based on headcount, effort and long timelines, is gradually giving way to outcome-oriented models, particularly in large modernisation and operations deals. Historically, legacy modernisation programmes were priced over six to seven years based on estimated effort, often costing several dollars per line of code.
By re-engineering delivery using AI-assisted approaches, Mphasis is now committing to outcomes over shorter timeframes and at lower unit economics, Rakesh asserted.
“What the client cares about is the outcome,” Rakesh said, adding that customers are increasingly indifferent to how many people or tools are deployed at different stages, as long as delivery risk, timelines and results are clear.
He described this as “savings-led transformation”, where efficiencies in existing systems free up the budget for reinvestment rather than simply cutting spending.
Talent, Hiring And the Changing Pyramid
The shift toward AI-assisted delivery is also influencing workforce strategy. Rakesh said Mphasis has moved away from traditional campus hiring over the past two years, instead engaging candidates through internal programmes and hackathons focused on live projects and emerging technologies.
The most important attributes the company looks for, he said, are “learnability” and “technical skills”.
Hiring continues across experience levels based on project needs, including a higher intake of junior talent in the US. In the December quarter, the company reported a headcount of 31,272—an increase of 463 over the previous quarter.
Over time, Rakesh said the workforce structure could evolve toward a more fluid, diamond-shaped model rather than a classic pyramid, moving from junior-heavy structures to agile teams dominated by mid-level specialists. However, he emphasised that such changes will take several years and require alignment with customers.
Deal Momentum And AI Infusion
During Q3, Mphasis secured $428 million worth of new deal wins, with 62% of them being AI-led.
Responding to questions on deal wins, Rakesh said most recent engagements include elements of AI-led delivery, spanning modernisation, operations and software lifecycle transformation. Some deals require a three-to-six-month ramp-up before revenues flow, while others convert more quickly.
Responding to a question from AIM on disclosing AI revenue, Rakesh said the more relevant measure for him is how many customer engagements are being impacted by what the company is building, rather than carving out AI as a separate revenue line.
“What’s important to me is, are we bundling services and software together in how many engagements?” he said. “The metric I gave was that clients representing almost 50% of our revenue today are on some form of a NeoIP engagement. That does not mean 50% of revenues are on that engagement alone.”
The CEO, however, emphasised that the company cannot continue to deliver without an AI-led approach. Otherwise, clients, he added, will either figure out how to do it with the same vendor, with another provider, or on their own.
“That’s why the right approach should be whether I’m able to infuse this in every engagement, every customer,” he said.
Giving an AI revenue breakdown, he argued, risks confusion, where strong growth in a reported AI bucket alongside slower overall growth could be misread as revenue deflation.
Partners And Investors
Rakesh said that the NeoIP platform has been built largely through internal development and ecosystem partnerships rather than large acquisitions, with hyperscalers and model providers acting as technology enablers.
He also addressed recent stake sales by Blackstone, which sold around a 9.5% stake in Mphasis through a ₹4,600-plus crore block deal during the quarter. While Blackstone is no longer a majority shareholder, Rakesh said the focus remains on execution and customer relevance rather than changes in the shareholding structure.
“This is not the beginning of the end,” he said. “This is the end of the beginning.”
Looking ahead, Rakesh said Mphasis is entering a new phase, with investments made over the past several years beginning to show tangible opportunity. “Talk is cheap,” he said. “Executing this is really hard.”
Mphasis reported a 2.6% quarter-on-quarter and 12.4% year-on-year increase in revenue in the third quarter of FY26 to ₹4,002.6 crore.
Net profit after the exceptional item related to the labour law change declined 5.7% quarter-on-quarter to ₹442.2 crore.
Its shares declined 2.32% on Friday, ending the day at Rs 2,745 apiece.
The post Why Mphasis CEO Believes AI Can No Longer Be ‘Lipstick’ on Legacy IT appeared first on Analytics India Magazine.