Part IThe Story
Guys as Old as Me
On February 17, 2026, the chairman of Infosys walked onto a stage in Bengaluru to explain to a room of investors why the industry he had helped invent was not about to die. The timing was not gentle. That month, roughly a trillion dollars had drained out of technology stocks worldwide, and more than $50 billion of it had come out of Indian IT companies, after new AI offerings from Anthropic and Palantir persuaded the market that the business of writing and maintaining other people's software might be ending. He was seventy. He had been in the industry, he said, for more than forty years, and he called that "the fortune or misfortune" of it. He opened with a joke about the safe-harbor clause. Then he told a story about old men.
Somebody had told him the other day, he said, about enterprise systems so old and so poorly documented that the companies running them keep retired engineers on contract. "Guys as old as me," he said. "75-year-old guys." Nobody else knows what is going on inside the machines. When something breaks, these men are "pulled in from Phoenix or Florida or wherever they are," and they solve problems that no one else can solve. Nandan Nilekani had rounded himself up five years for the line. He didn't say whether it was modesty or a setup.
The anecdote had a commercial purpose. Its subject was brownfield: the trillions of dollars sunk into mainframes from the 1960s, minicomputers from the 1980s and local networks from the 2000s, all still running, all in silos, all undocumented. Nilekani's argument was that AI's real productivity gains would come from modernizing that tangle, and that modernizing it was hard. Infosys executives told reporters afterward that productivity on legacy projects had moved about 1 percent, against 15 percent on new ones. "Accumulated debt over decades must be paid," he said. A man who once declared the world flat was now describing it as stratified, layer on layer of half-forgotten decisions, each one still load-bearing. Infosys shares closed that day up 1.8 percent, at ₹1,391.20.
The story worked on a second level, though he would never have said so. For two decades the Indian state and Indian capital have handled their crises the way those enterprises handle theirs. They find the man who knows how the system works and they pull him in. In August 2017, when Infosys's first non-founder CEO resigned in a public feud with the founders, the board called Nilekani back. In 2009 Prime Minister Manmohan Singh gave him a one-page brief and asked him to give every Indian a number. On Sunday, July 26, 2026, five months after the speech about old men, Prime Minister Narendra Modi announced that Nilekani would chair a high-powered task force to make India's examination system leak-proof. Its predecessors had leaked badly enough that the Supreme Court sent its own judges to inspect the National Testing Agency.
In his keynote that February, Nilekani said that technology had always been deterministic. "You said A plus B equals C. So no matter how many times you said A plus B, the answer was C." AI broke that promise, because the same prompt could produce a different answer every time. The question, he said, was how to build things with "the robustness, reliability and resilience of the deterministic world" in a world that no longer behaves that way.
The question is older than AI, and it describes most of his life. He has spent fifty years trying to lay deterministic pipe through a non-deterministic country.
By the Numbers
The Plumber's Ledger
$250Initial capital of Infosys, incorporated in Pune on July 2, 1981
6xGrowth in Infosys revenue during his CEO tenure (2002–2007), to $3 billion
600MAadhaar enrollments he promised the Prime Minister, and crossed by March 2014
138 croreAadhaar numbers generated, the world's largest biometric ID system
16.99BUPI transactions in January 2025 alone, worth ₹23.48 lakh crore
228,575Votes by which he lost Bangalore South in the 2014 general election
328,594Infosys employees as of March 2026
A Small Act of Rebellion
The family's first language was Konkani, a tongue that belongs to the western coast. Durga and Mohan Rao Nilekani came from Sirsi, in the forested uplands of northern Karnataka. Their son was born in Bangalore in June 1955. He grew up to speak Kannada fluently, along with English, Marathi and Hindi. That is five languages for a man who would later build a national identity system with only four mandatory fields.
Mohan Rao Nilekani was general manager of the Mysore and Minerva Mills, and by every account a believer in Fabian socialism. Under that doctrine the state was the instrument of progress, change came gradually through competent administration, and well-educated men in good offices could plan their way to a fairer country. His son would later call the India of those years "a very stultifying socialist regime." He would also say, in the middle of his own run for Parliament, that the Congress party was his "ideological home," and that he came "from a Nehruvian family." He kept a vivid childhood memory of his father driving him to a rally where Jawaharlal Nehru was speaking. Both views lasted his whole life. He never resolved the contradiction, and he never seemed to feel it as one.
Because his father's job moved, the boy was sent at twelve to live with an uncle in Dharwad. Dharwad is a university town with a reputation for Hindustani classical music and not much else of national importance. He left Bishop Cotton Boys' School in Bangalore for St. Joseph's High School in Dharwad, then went on to Karnatak College. One profile from his Aadhaar years says the move taught him independence. He has described himself differently. "I was from a small town and not at all sophisticated," he told an interviewer in 2006, remembering what came next.
What came next was IIT Bombay, in 1973. His father had wanted him to go to IIT Madras. Nilekani chose Bombay instead and electrical engineering, and later called the choice "a small act of rebellion." The phrase is modest. Most of his rebellions have been small and administrative, and most have succeeded. In Bombay he was surrounded by "boys from Cathedral and other up-market schools," the polished children of the city's English-medium elite, and he set about learning to survive among them. He joined the quiz circuit. He helped organize Mood Indigo, the college's cultural festival, twice. In his final year he ran for General Secretary of Student Affairs, which meant going from hostel to hostel giving speeches, though by his own later admission he was afraid of public speaking. At a quizzing event he met Rohini Soman. They married in 1981, the same year he co-founded the company that would make them both billionaires.
Then he got sick. On the day of the entrance examination for the Indian Institutes of Management, the usual next step for an ambitious IIT graduate who stayed in India, he was ill, and he missed it. Going to America for a master's, the other usual next step, was out of reach. "There was no money to go overseas," he said in 2023. "Serendipity has played a big role in my career."
It is a strange thing for a systems man to say, and he has said versions of it many times. Fifty years later, on July 26, 2023, IIT Bombay unveiled a plaque renaming its Main Building after him, in recognition of a gift of ₹315 crore. "Returned to IIT Bombay 50 years after I first entered the institution," he posted. "I am what I am because of IIT Bombay." The small-town boy who had been intimidated by Cathedral boys now had his name on the building their sons would walk past.
You Just Go When Murthy Calls
The job he took instead of business school was at Patni Computer Systems, then a modest firm. IBM had left India in the late 1970s, and Patni resold minicomputers made by Data General. "When I heard about this mini computer company," Nilekani recalled, "I said, 'Wow, this sounds exciting.'" The head of Patni's software division interviewed him.
N. R. Narayana Murthy was nine years older. He was born in 1946 in Sidlaghatta, a small town north of Bangalore, and was educated in Kannada medium until the tenth standard. He took an engineering degree in Mysore and a master's from IIT Kanpur. At IIM Ahmedabad he worked as chief systems programmer on what is described as India's first time-sharing computer system. Then he started a company called Softronics, which failed after about a year and a half. He liked to tell how a 1974 arrest near the Yugoslav-Bulgarian border, which he described as being detained and expelled for no good reason, turned him from a "confused leftist" into a "compassionate capitalist." He and Nilekani had both come out of the socialist inheritance. Murthy left it in a jail cell. Nilekani never quite left it.
Murthy didn't ask the young graduate about his transcript. "I gave him a set of puzzles to test his learnability," Murthy said, "and he solved them." Nilekani was hired as a software engineer trainee. His deference to the older man has been total and long-lasting. "If he asked me to jump off the cliff, I would have jumped off the cliff," he said decades later. In 1981, when Murthy proposed leaving Patni to start a company of their own, the decision took no deliberation. "You just go when Murthy calls."
Seven men left. Murthy, Nilekani, Kris Gopalakrishnan, S. D. Shibulal, K. Dinesh, N. S. Raghavan and Ashok Arora incorporated Infosys Consultants Private Limited in Pune on July 2, 1981. Its initial capital is usually given as $250. In the version told about Murthy, the ₹10,000 came from his wife, Sudha. The New York Times later described the company as starting "in a Bangalore garage" and called its founding "part of modern Indian folklore." Folklore has variants, and this one does too: one account puts the first office in a small Pune apartment, another sets the founding in Bangalore. The amount is the detail that matters. Seven engineers, one shared pool of savings, and an idea that in 1981 sounded faintly absurd: India's surplus of English-speaking programmers could do the world's software work from a distance, more cheaply than anyone in California or Cambridge.
The company moved its headquarters to Bangalore in 1983. Nilekani went to the United States to look after Infosys's interests there, and became, in one description, the firm's "marketing face." Its first outsourcing contract was with Data Basics Corporation of New York. In 1987 it opened its first overseas office, in Boston. In 1989 Arora left and sold his shares to the others. In 1990, Murthy later said, someone offered to buy Infosys for ₹2 crore. The founders declined.
The economic liberalization of 1991 changed the arithmetic. In February 1993 Infosys priced its IPO at ₹95 a share and found that India didn't want it. The issue was undersubscribed. Morgan Stanley stepped in to stabilize it, taking a 13 percent stake. By June 1993 the shares opened at ₹145. That same year the company introduced an employee stock option plan, the first of its kind in India, and over time an estimated 500 employees became millionaires through it. On March 11, 1999, Infosys listed American depositary receipts on Nasdaq, the first Indian company to list on an American exchange. Revenue that year reached $100 million.
The order of events matters. The founders' shareholding began falling the day the company listed, and it fell further with each instrument designed to share it out. By March 2024 the entire promoter group held 14.61 percent of Infosys. The Life Insurance Corporation of India held more than any single founder. The founders had built a company meant to outgrow its founders, and they did it on purpose.
The World Is Flat, and Then Flattened
In March 2002 Murthy handed the CEO title to Nilekani, after twenty-one years in it. The timing was ideal. In 2004 Thomas Friedman of The New York Times came to Bangalore, sat with Nilekani, and left with a phrase. Nilekani had told him the playing field was being leveled. Friedman turned the remark into the title of a 2005 bestseller, The World Is Flat, and the American press gave Nilekani a nickname, "the Bill Gates of Bangalore," that he never quite escaped.
The numbers supported the nickname. It had taken Infosys twenty-three years to reach its first billion dollars in annual revenue, in 2004. It reached the second billion in twenty-three months. Under Nilekani, revenue grew roughly sixfold, to $3 billion. The workforce passed 69,000. In 2002 the company created a business-process subsidiary called Progeon, with Citigroup paying $20 million for a minority stake; in 2006 Infosys bought Citi out for $115 million. In 2004 it launched Infosys Consulting, based in Fremont, California, to compete with the strategy houses whose work its engineers were usually hired to carry out. The same year, construction plans were announced for what would become the Infosys Global Education Centre, on a 337-acre campus in Mysore that could train 14,000 employees at once. It is often described as the largest corporate university in the world.
Nilekani was managing at a scale where intuition fails, so he built a structure for doubting himself. In late 2006, explaining his approach, he described three planning horizons: five-year scenario planning, three-year strategy and quarterly operations. The five-year exercise existed mainly to catch the company's own false beliefs. "Companies run on a set of explicit, tacit assumptions about the market, customers, about the future, world and technology," he said. "What happens to them often is that some of those assumptions are changing." By then, he noted, 40 percent of Infosys revenue came from services that had not existed five years earlier.
The people around him described a style both relentless and impatient. In the summer of 2008, Shoba Narayan of Mint watched him run a meeting on making Infosys environmentally sustainable, a pet project. He fired off questions. "Why are the rooms kept so cold? In Bangalore, we shouldn't even be using air conditioners, yaar." A carpooling proposal came up, built around a website. He wanted it on text messages instead. "Who wants to go online to check if a car is going to Electronic City?" He asked whether the water flushed from the urinals was being reused. It was, for landscaping. When the team explained that vendors would not make energy-efficient laptops for an order as small as Infosys's, he proposed pooling orders with rival firms. "I'll talk to Azim about it," he said, meaning Azim Premji of Wipro, the competitor across town.
The HR head at the time, Nandita Gurjar, called him "the closure man." "He is a man in a hurry," she told Narayan. Once she brought him news that Infosys had won a prestigious global award. He said "Oh great!" and moved on. She asked him what award would make him stop and take notice. He paused, she said, and answered: "The Nobel Prize."
— Nandan Nilekani, to Business Today, 2023There is always a gap between reality and potential and I have the urge to bridge that gap.
In April 2007 he handed the CEO job to Kris Gopalakrishnan and became co-chairman. That year Forbes ranked him 754th among the world's billionaires, with $1.3 billion, and noted that he sat on the board of Reuters. Infosys was a rotation, and the founders took turns at the top. Murthy had retired as an executive the previous year, on turning sixty, under a rule he had imposed on himself. In 2009 Nilekani sat with Dan Roberts of The Guardian in a nearly empty London hotel lobby and updated Friedman's thesis. Global trade had collapsed, and one of Infosys's main rivals, Satyam, had just been exposed in an enormous accounting fraud. "The world has got flattened," he said. "Because the world was flat, we all got flattened: the crisis moved around the world faster."
Roberts noted that it was the kind of enigmatic line visionary businessmen get away with. Nilekani still defended markets without apology: "We should not throw out the baby with the bath water... we lived for many years under a very stultifying socialist regime and have seen what it is like." Three months later he left the private sector entirely, to build the largest government database in human history.
The Computer Boy
He had tried government once before, and on that occasion he had hidden his credentials.
In 1999, S. M. Krishna became chief minister of Karnataka and set up a series of task forces. His office asked the managing director of Infosys whether he would lead the one on Bangalore. Nilekani agreed and put ₹2 crore of his own money in. The work was mostly unglamorous: garbage, roads, the parts of urban government that decide whether a fast-growing city stays livable. In a January 2000 interview with Rediff, the interviewer suggested he was living out "the educated urban man's dream" of having money and power to remake a metropolis. He brushed that off and described his job in managerial terms. "There is no shortage of reports on what should be done," he said. "There are a lot of reports on how to plan Bangalore, how to build a ring road and how to tackle garbage. What is really required is an action-oriented focus." Money, he added, was not the binding constraint. "It is more a question of time."
He was careful in that first public role not to talk about technology. He later wrote that he had been "wary of being labelled a 'computer boy' who saw every problem as something that could be solved by writing a piece of code. After all, what do computers and software have to do with clearing garbage or provide soft drinking water?"
That passage appears in Imagining India: The Idea of a Renewed Nation, published in 2009 with a foreword by Friedman. At 528 pages, it is part memoir and part policy brief, and its organizing idea was that nations change when ideas take root. At TED in Long Beach in February 2009, he gave the short version. He sorted India's ideas into four kinds: those that had arrived, those in progress, those in conflict, and those the country had to anticipate. The ones that had arrived included people as human capital rather than as a burden, entrepreneurs as role models rather than villains, and English as the language of jobs rather than of colonizers. "If you don't really invest in your human capital," he warned, "the same demographic dividend can be a demographic disaster." Among the ideas to anticipate, the first he listed was technology in governance. "It's important to prove," he said near the end, "that growth and democracy are not incompatible."
He went on The Daily Show with Jon Stewart in March 2009 to promote the book. That year Time put him on its list of the world's 100 most influential people. Within months, the man who had once avoided talking about computers in public was being asked to solve, with computers, one of the oldest problems in the Indian state: it did not know who its people were.
One Page, One Billion
Nilekani has often said the mandate was a single page. Its substance was that every resident of India should get a unique identity. The Planning Commission had issued the notification creating the Unique Identification Authority of India in January 2009, and the annual plan for 2009–10 set aside ₹100 crore. In July 2009 Nilekani resigned from the Infosys board to become UIDAI's chairman, with the rank of a Cabinet minister, at the personal invitation of Manmohan Singh. A trade publication greeted the appointment with polite skepticism. National ID projects were not new, and pilot programs had already run in at least two states "at huge costs." Nilekani's own early remarks were narrow and technical. "The key issue is to ensure there are no duplicates," he said. His job, he insisted, was "about the infrastructure," not about delivering benefits. Others would build on top of it.
He framed the problem more fully on October 27, 2009, in the 43rd A. D. Shroff Memorial Lecture in Mumbai. Shroff, a co-author of the 1944 Bombay Plan, had defended competition and private enterprise in the decades when the planning consensus held. The UID, Nilekani argued, was the missing third leg of a platform whose other two legs, mobile phones and online banking, already existed. Without proof of who they were, he said, the poor were "chained to the places they work." He described what that meant in practice: duplicate ration cards, ghost workers on rural employment rolls, a man named Venkatanna whose identity had been used to claim the same housing benefit twice. Every subsidy that leaked through a false or duplicate identity was money taken from the poor and given to whoever held the forgery.
Within roughly a month of starting, he had set the number: 600 million people enrolled within five years. Here is how he later explained to an audience at MIT why the target had to be so large and come so early.
— Nandan Nilekani, MIT Tata Center, November 2014Whatever we did, it should reach such an irresistible scale that it could not be undone.
The logic was political. The government supported the project, and the next one might not. The only protection a program like this had against a change in power was being too large to dismantle. Size was the strategy. "Scale has to be intrinsic to the design of the solution that you have in mind," he said at MIT. "Scale is not an afterthought. You don't build a mousetrap and then start thinking about how you can catch every mouse on the planet."
Designing for scale meant cutting the design down to almost nothing. The system recorded name, address, date of birth and sex, plus an optional mobile number and email, along with a photograph, ten fingerprints and both irises. It recorded no caste, no religion and no income. He called it "a pure ID system, identity as a platform," and he guarded the list of fields closely. "Everyone around you is trying to force you to make it more complicated," he said. "We're not Facebook or something, trying to create all kinds of data about everybody. Any one data item we added to the ID had to be collected one billion times."
The name the team chose, Aadhaar, means "foundation."
The resistance came quickly. In December 2011 Parliament's Standing Committee on Finance, chaired by the BJP's Yashwant Sinha, rejected the National Identification Authority of India Bill. It called the bill "directionless" and questioned whether UIDAI had any legal authority to collect biometrics at all. The project's critics inside the government were barely more patient. Home Minister P. Chidambaram objected that UIDAI's enrollment overlapped with the biometric census his ministry was running through the National Population Register. Nilekani had just asked for an additional ₹15,000 crore. Singh had backed him so far. The bill went back to the Cabinet.
He kept enrolling people anyway. That October, The New Yorker had sent Ian Parker to write about the project and the man; the piece ran as "The I.D. Man" and called Aadhaar "the biggest social project on the planet." By August 2013, Soutik Biswas of the BBC counted 500 million Indians enrolled at 25,000 centers and 400 million numbers issued for a ₹120 billion project. Ten of India's twenty-eight states were already using the number to send pensions, scholarships, rural employment wages and cooking-gas subsidies directly to recipients. The country Nilekani described to Biswas was one where only 58 percent of children were registered at birth and 40 percent of villagers had no bank account. Biswas noted that he spoke "with a brisk, evangelical flair."
"In the West, IDs are taken for granted," Nilekani told him. "You have millions of people who have no ID and no acknowledgement of their existence. They can't open a bank account, they can't have a mobile phone connection. It's an identity divide."
The surveillance worries were brought up again in that same interview, as they would be every year for a decade. "A lot of attention has been made to design the ID so that it doesn't become a massive data collector," he said. "It's a simple ID system that protects the residents."
In January 2014 the total passed 560 million. He stayed until mid-March. "I had set a goal which I had promised to the prime minister that we will cross 600 million Aadhaar, which I did," he told an interviewer later. "After that, I said 'Mission Accomplished.'"
The phrase has an unlucky history, and he probably knew it. The comparison doesn't fully hold, though. Aadhaar's mission really was accomplished, in the narrow sense he had set: the thing could no longer be undone. Narendra Modi's BJP won power that May, and the BJP had attacked the project while in opposition. In government it took Aadhaar over completely. By April 2017, 1.14 billion people had a number. By 2025 the count was 138 crore. The irresistible scale had resisted even the party that had campaigned against it.
Bangalore South
Nilekani's next decision is the one his admirers find hardest to explain.
In March 2014 he joined the Indian National Congress, the governing party of a government that was about to lose badly, and the party named him its candidate for the Lok Sabha seat of Bangalore South. The New York Times found him readying himself for "an uneven, uphill battle." At fifty-eight he had the same neat mustache he had worn for decades, but he had traded business suits for kurtas and vests. He told the Times that his political model was Michael Bloomberg, the billionaire who became mayor of New York. A newcomer, the anti-corruption Aam Aadmi Party, was campaigning on the idea that clean people should go into politics, and Nilekani addressed that pitch directly. "The A.A.P.'s argument is that there should be clean people in politics," he said. "I represent that very well. Plus, I have competence."
He could have taken a much easier route. He had the option of entering the Rajya Sabha, Parliament's indirectly elected upper house, where technocrats are routinely placed without having to face voters. He turned it down.
— Nandan Nilekani, to The New York Times, February 2014My mission in life, at this point, is to bring more change into India. If I'm going to be an effective agent of change, I must have the credibility and legitimacy of having won a popular election.
He was the richest candidate in the entire 2014 election, with declared assets of ₹7,710 crore. His opponent was Ananth Kumar of the BJP, the incumbent in a constituency with a large BJP base. The national mood was a fury against Congress-era corruption scandals, and it ran in Kumar's favor. The result was 633,816 votes for Kumar and 405,241 for Nilekani: 56.9 percent against 36.4. He lost by 228,575 votes.
"The Modi wave blew him away," said T. V. Mohandas Pai, a former Infosys colleague. Pai had welcomed his entry into politics but had publicly doubted the timing. "I had said 'I wish he had done more work in Bangalore before contesting,'" Pai said. "Had he worked in the constituency and built support base he would have won the election."
About nineteen months later, Nilekani sat down with Manu P. Toms of VCCircle and was more comfortable with what had happened. "Good thing I lost," he said, and laughed. Had he won, he added, "obviously, I would have had to deliver." Then, more seriously: "At the end of the day I have realized that perhaps I am not cut-out for it. I like to work on problem-solving, with a three- to five-year horizon. These are very complex problems."
The defeat says a lot about him. He had a five-year time frame, a measurable target and the kind of competence that keeps a large database running. Electoral politics has none of those. Its feedback is noisy, its cycles are long, and its currency is loyalty, which he could not produce at scale. He had approached the electorate the way he approached a deployment, sure that legitimacy was something one could get by finishing a project. The voters disagreed. It was the one system he couldn't modernize, and when it rejected him he stopped trying to win it over and went back to working underneath it.
Plumbing Is Not About Politics
His favorite word, Toms reported in January 2016, was "plumbing." "Plumbing is not about politics," Nilekani said. "It is cutting across horizontally and fixing something. That is what I did in the government."
After 2014 he worked almost entirely below the line of public credit, as an adviser, committee chair or board member, in both public and private roles. He had already chaired the committee that designed the Goods and Services Tax Network, the IT backbone of India's GST, and the committee behind a national electronic highway-tolling system. In Rebooting India: Realizing a Billion Aspirations, written with Viral Shah, a colleague from his Aadhaar years, he listed twelve projects. Aadhaar and direct benefit transfer were done, and four more (eKYC, payments, GST and tolling) had been "planted," he said, while he was still in government. "I had done a startup inside the government," he told Toms, "and it is now a successful startup." The book's argument was that India's institutions were "19th century systems" serving a billion people who no longer had time to wait.
Payments were the most important of the four. Nilekani advised the National Payments Corporation of India, the bank-owned utility that would build the Unified Payments Interface. He argued for an open architecture in which any bank or app could plug in and every payment address could reach every other. In 2015 he gave a speech that the people who heard it still cite. He called it the "WhatsApp moment" for Indian banking. WhatsApp, he noted, used phone numbers as identity and phone contacts as a social graph, and offered free messages. "WhatsApp does 30 billion messages a day," he said. "All the telecom operators on the planet put together do only 20 billion SMS a day. In other words, one company started by a Ukrainian migrant with 40 engineers has 50% more messaging volume than all the telcos put together." The same thing, he predicted, was about to happen to banks. A journalist in the room recalled sensing "both excitement and scepticism among the audience."
Ten years later, the doubters had been proven wrong. In January 2025 UPI processed 16.99 billion transactions worth ₹23.48 lakh crore, a monthly record. In fiscal 2024 it processed more than 131 billion transactions, worth over ₹200 lakh crore. It accounted for roughly 80 percent of retail digital payments in India, connected more than 640 banks and had expanded to Singapore, the UAE, France and other countries. Aadhaar-based eKYC, the instant identity check that the number made possible, had helped Reliance Jio, a late entrant into Indian telecom, grow fast enough to become the market leader. Nilekani's pieces of infrastructure connected to each other and grew together. Business journalists started calling him, half-jokingly, the Chief Technology Officer of India.
He defended the infrastructure aggressively. When Facebook pushed Free Basics, a program giving free access to a handful of websites, he and Shah called it a "walled garden" and wrote that with it, "Digital India is as good as dead on arrival." When the Supreme Court in 2015 restricted Aadhaar's use to ration shops and cooking gas, he wrote an op-ed calling for an appeal. He opened it with Gandhi's talisman, the instruction to recall "the face of the poorest and the weakest man you may have seen" and ask whether a policy would help him. He argued that the court's order failed the test. In March 2016, as the Aadhaar bill moved through Parliament, he wrote a sentence that sums up his approach to institutions: "Every entrepreneur knows that if you wait for all lights to be green, you will never leave home." He had taken the job in 2009 assuming a law would follow eventually. The law arrived seven years later.
The critics kept writing. Engineers published technical rebuttals of his arguments. Privacy advocates warned that seeding one number into every database would make profiling trivially easy. In January 2018 Nilekani answered them in the Hindustan Times. "These fears are over-hyped and baseless," he wrote. He used the analogy of a mobile number linked to Ola, Zomato and WhatsApp: the telecom company doesn't learn where you ride, what you eat or whom you message. Then he made an admission. UIDAI, he wrote, "had originally considered tokenisation in the early days of Aadhaar in 2010, but it was an idea then ahead of its time." It was being added now, and every organization would receive a token instead of the number itself. He presented the change as proof that the architecture was flexible. The more skeptical reading was that the critics had been right about a real gap, and that he had treated the gap as a scheduled upgrade. Both readings can be true. Nilekani has always treated legitimacy and design as one problem.
In 2016 Paul Romer, then chief economist of the World Bank, called Aadhaar "the most sophisticated ID programme in the world." The Indian government now says the system has saved ₹3.48 trillion, about $39.2 billion, by cutting fraud and bypassing corrupt intermediaries. The figures are large, mostly unaudited and plausible.
The Founder Comes Home
The company he had left was not doing well.
Founders had run Infosys from 1981 to 2014, passing the CEO title from Murthy to Nilekani to Gopalakrishnan to Shibulal. In 2013, with the company struggling, Murthy came back for a year as executive chairman, bringing his son Rohan in as an executive assistant. In August 2014 the board hired Vishal Sikka, a former SAP executive, as its first CEO from outside the founding group. Sikka was an outsider with a Silicon Valley style, a California office and plans to take the company into AI and automation. He acquired an Israeli automation company, Panaya, for $200 million, and the digital-experience firm Skava for $120 million. The founders, who still held a large block of shares and the moral authority of having built the place, started raising questions about governance, severance payouts and the Panaya deal. Murthy was the most persistent.
On August 18, 2017, Sikka resigned. In a note to his board colleagues he blamed a "drumbeat of distractions" and "false, baseless, malicious and increasingly personal attacks." It was the most public breakdown in the history of India's best-governed company, and it came from inside the founding family.
Within days, Nilekani returned. On August 24, 2017, he became non-executive chairman, replacing co-chairs R. Seshasayee and Ravi Venkatesan. Seshasayee, Venkatesan and Sikka left the board, as did the directors Jeffrey Lehman and John Etchemendy. Forbes India would profile him that year under the headline "Mr Congeniality," arguing that his natural affability had helped him bridge many divides. The returning founder moved the company's center of power from California back to its headquarters in Bengaluru. In January 2018 the board installed Salil Parekh, who had come from the French consultancy Capgemini, as CEO. Parekh was an outsider too, but he would run things from Bengaluru, with the chairman a few floors away.
That year, 2017, also changed his relationship to his own fortune. In November he and Rohini signed the Giving Pledge, committing half their wealth. At the time they were reported to be worth about $1.7 billion, most of it in Infosys shares. Only two other Indians had signed, Azim Premji and Kiran Mazumdar-Shaw. Rohini Nilekani, the quizzer from IIT, had by then become, in The Ken's description, India's most prolific woman philanthropist, and they wrote their reasoning together: "We see inequality is increasing sharply in most countries. We see the young and the restless in this interconnected globe, unsure of their future, wanting more but anticipating less." They had already put $10 million into EkStep, a nonprofit education platform they co-founded with Shankar Maruwada in 2015. With Ratan Tata, Nilekani had backed a microlending company called Avanti Finance, each putting in ₹50 crore.
On August 24, 2021, four years to the day after he returned as chairman, Infosys became the fourth Indian company to reach a market value of $100 billion.
There is a less flattering side to all of this. Infosys, the company he chairs, has repeatedly been the vendor behind the national digital systems he spent years arguing for. In 2013, problems followed when Infosys took over the MCA21 corporate-affairs portal from Tata Consultancy Services. In 2020 the finance ministry summoned Infosys executives over glitches in the GSTN portal, the network whose design committee Nilekani had chaired. In June 2021 the new income-tax filing portal, built by Infosys, broke so badly and for so long that the government summoned CEO Parekh and pushed the filing deadline to December 31. Taxpayers were still reporting failures on the portal's first anniversary. On the corporate side, the company had paid $34 million in 2013 to settle a U.S. civil case over its use of visitor visas, while denying any "systemic visa fraud," and in 2019 it paid California $800,000 over similar claims. The plumber's own company had built pipes that leaked. The contradiction is clear, and the public record shows no comment from him on it.
Root and Branch
His case for AI followed the same pattern he had always followed: a fast technology, a slow institution, and the gap between them. "Models will be a commodity," he told Rest of World in July 2025. "We have to bring the conversation back to making it useful to people." He described a future in which a farmer in Bihar could speak to a computer in Maithili, in his own dialect, using his own colloquialisms, with no keyboard and no literacy required. He spoke about AI the same way he had spoken about the ID: as plumbing for a country that is still largely oral.
At the 2026 investor event in Bengaluru he gave the corporate version. He described "fundamental root-and-branch surgery" on legacy systems, and he acknowledged the comedy of the moment. He described two people in a quarrel, one using AI to inflate a one-paragraph email to ten, the other using AI to shrink it back to one. "So both have used AI, but what have we achieved? Nothing." He predicted that within five years "there'll be more AI legacy systems than any other legacy system," and that someone would have to clean those up too. He argued for teaching young engineers without tools first. Otherwise, he said, "it's like the guy who never knows how to calculate because he was born with a calculator." He used self-driving cars as an example of how much context matters. The first DARPA challenge was in 2004, he noted, and twenty years later robotaxis operated in a handful of American cities. "By the time they come to Bangalore, it'll be 2047, because dealing with Bangalore traffic will be a different level of context altogether."
The core of the talk was a distinction he has been drawing his whole career. "There is no opportunity gap," he said. "It is not an opportunity risk. It's an execution risk. You get it?"
MIT Technology Review had visited him a month earlier. "At 70 years old, Nilekani should be retired," the piece began. Instead he was promoting a "finternet," a global financial infrastructure in the spirit of what he had built in India, developed with the Bank for International Settlements, along with a digital layer to stabilize India's unreliable electrical grid, which he called "the next UPI." "It sounds like some crazy stuff," he said. "But I think these are all big ideas, which over the next five years will have demonstrable, material impact." Five years again. The horizon had never changed.
He told the same reporter what seeing the systems in use meant to him. "It reinforces that what you have done is not some abstract stuff, but real stuff for real people."
Then the summons came. Leaked examination papers had become a national scandal: millions of students sit for the tests that decide medical seats and engineering seats, and too many of those tests had been compromised. On July 26, 2026, the Prime Minister announced a task force to make the system leak-proof and technology-driven, and named Nilekani to lead it. That was three years to the day after IIT Bombay put his name on its Main Building. The boy who had missed his own entrance exam because he was sick would now redesign the country's.
It is the anecdote from the investor stage again. A system fails, nobody knows its insides, and someone calls the man from Phoenix.
The difference is in what his systems do when he is not there. The legacy systems in his anecdote need their old men; that was the point of the story. Aadhaar, by design, needs no one in particular. In a village hours from the nearest bank branch, a farmer goes into the local store to collect a welfare payment. There's no passbook and no clerk to bribe, and he doesn't need to know anyone's name or speak anyone else's language. He presses his thumb to a small scanner on the counter. The machine checks the print against one of 138 crore records, finds him, and the money comes.
The scanner does not ask who built it.
Part IIThe Playbook
What follows are the operating principles visible in fifty years of Nilekani's work, in a garage-or-apartment startup, a $19 billion multinational, a Cabinet-rank agency, a failed campaign and a decade of committees. Most of them are about one problem: how to get large, slow, distrustful institutions to adopt something new before they can organize against it. They are drawn from the record above, and where possible from his own explanations of what he was doing.
The Principles
- 1.Set the number before you have the method.
- 2.Make it irreversible before the next government arrives.
- 3.Collect the minimum; every field costs a billion times.
- 4.Build the rail, not the train.
- 5.Leave home before the lights turn green.
- 6.The problem is deployment, not technology.
- 7.Audit your tacit assumptions on a schedule.
- 8.Choose your arena by its feedback loop.
- 9.Name the moment with an analogy people already believe.
- 10.Design ownership to dilute yourself.
- 11.When you return, lower the temperature and move the center home.
- 12.Treat the critics' complaint as the next release.
Principle 1
Set the number before you have the method
Within about a month of taking the UIDAI job, before the architecture existed, before the biometric vendors were chosen and before Parliament had passed anything, Nilekani committed to 600 million enrollments in five years. That was more than half of India. He explained at MIT that the target came first because "it allowed them to crystallize their thinking about how to reach those goals." A number that big rules out every approach that cannot reach it, which in practice rules out nearly everything government had tried before.
The same habit shows up at Infosys. The doubling from $1 billion to $2 billion in twenty-three months did not happen by accident. It came out of planning at three horizons, where the five-year target forced the three-year strategy, which in turn forced the quarterly operations. It also shows up in the 2014 promise to the Prime Minister, which he treated as a personal contract and declared "Mission Accomplished" only once the count passed 600 million.
The risk is that a number can become a vanity target. His defense was to tie every target to a reason outside the organization: enrollments needed to cross the threshold at which the program could not be undone, and revenue needed to show that Indian companies could operate at global scale.
Tactic: Before designing anything, write down the scale that would make your project unkillable, then throw out every plan that cannot reach it.
Principle 2
Make it irreversible before the next government arrives
"Whatever we did, it should reach such an irresistible scale that it could not be undone." This may be the most important sentence in Nilekani's public career, because it treats speed as a defense rather than an operating metric. He knew that the government that appointed him might lose power and that the opposition was hostile. In December 2011 a parliamentary committee led by the opposition rejected his enabling bill outright. He kept enrolling people. By the time the BJP took power in 2014, Aadhaar covered more than 600 million people and states were sending pensions and wages through it. Dismantling it would have meant cutting off welfare payments to millions of people.
This is a corporate idea applied to the state: build switching costs until reversal becomes politically impossible. It is also a debatable one. Critics argue, with some justice, that "too big to undo" is just as true of a system that should be undone. The principle is morally neutral. Its value depends on the quality of what is being protected.
Tactic: Identify the specific actor who could kill your initiative and calculate the adoption level at which killing it would cost them more than tolerating it, then reach that level first.
Principle 3
Collect the minimum; every field costs a billion times
"Any one data item we added to the ID had to be collected one billion times." It is the clearest statement of how scale changes design, and it led to Aadhaar's spare schema: name, address, date of birth, sex, an optional phone number and email, and biometrics. There was nothing about caste, religion, income or bank balance. "Everyone around you is trying to force you to make it more complicated."
The discipline worked in two directions. It made enrollment fast enough to hit the target. It also gave him his strongest argument against surveillance fears: a database that never collects your religion cannot reveal it. When he later wrote that Aadhaar was designed so "it doesn't become a massive data collector," he was arguing from the original data dictionary.
The same instinct to subtract shows up in the 2008 sustainability meeting, where he steered the carpool proposal from a website to SMS, the channel people already used, and away from adding another login.
Tactic: For every feature or data field you propose, multiply its cost by your target user count; if you would not pay that total, cut it.
Principle 4
Build the rail, not the train
From his first weeks at UIDAI, Nilekani insisted that his job was "about the infrastructure," and that others would build applications on top. He called Aadhaar "identity as a platform." UPI followed the same design: an open, interoperable layer where any bank or app could connect and the public utility competed with nobody. Aadhaar-based eKYC was a narrow function, instant identity verification, and it helped Reliance Jio scale to market leadership in telecom. Nilekani did not build Jio. He built something Jio could use.
⚙
The Stack, Layer by Layer
Each layer was designed to be useful to the one above it, and none tried to own the end user.
| Layer | What it does | What it deliberately does not do | Phase |
|---|---|---|---|
| Aadhaar | Proves a person is unique and who they say they are | Record caste, religion, income, or purpose of authentication | Foundation |
| eKYC | Releases verified name and address to regulated entities | Learn why the check was done | Foundation |
| UPI | Moves money between any two accounts via any app | Compete with the apps built on it | Rail |
| Account Aggregator | Lets users share financial data by consent | Store the data centrally | Rail |
| ONDC | Lets any buyer app find sellers on any platform | Become another walled marketplace | Market |
| Finternet / energy layer | Proposed extensions to global finance and the power grid | Not yet proven | Speculative |
What made the approach work was restraint. Each layer declined to capture value that it could easily have captured. That restraint made the layers trustworthy enough for competitors to share, and shared adoption gave them their scale.
Tactic: Define the narrowest function your platform must perform, publicly commit to not competing with the businesses built on top of it, and let their growth become your distribution.
Principle 5
Leave home before the lights turn green
"Every entrepreneur knows that if you wait for all lights to be green, you will never leave home." Nilekani took the Aadhaar job in 2009 knowing there was no statute behind it. He enrolled hundreds of millions of people while a parliamentary committee called the enabling bill "directionless." The law passed in 2016. By then nearly a billion numbers had been issued and the program could point to more than ₹10,000 crore saved on cooking-gas subsidies in a single year.
The principle has a corporate parallel. Infosys priced its 1993 IPO, watched it fail to fill, and let Morgan Stanley stabilize it rather than withdraw and wait for a better market. The shares opened at ₹145 against a ₹95 offer.
The obvious risk is that moving without authorization invites a backlash that uses the lack of authorization against you, which is what happened in the Supreme Court cases. Nilekani's answer was to move fast and also to argue in public. He wrote op-eds, testified, and treated legal legitimacy as a parallel project rather than a precondition.
Tactic: Separate the permissions you genuinely need from the ones that are merely comfortable, start on the first set immediately, and run the campaign for the second set alongside the work.
Principle 6
The problem is deployment, not technology
As far back as the Bangalore task force in 2000, Nilekani said "there is no shortage of reports," and that what was missing was "an action-oriented focus" with accountability and monitoring. He hid his technology background so he wouldn't be dismissed as a "computer boy." Twenty-six years later, on the investor stage, he made the same argument about AI: "The technology is far ahead of its deployment." He cited Clayton Christensen's "technology overshoot." He noted that brownfield productivity had moved about 1 percent while greenfield had moved 15. "If you think that some better product coming along is going to solve this — nothing is going to happen, because the problem is here, not there."
The point is easy to agree with and hard to act on. Nearly all the money, prestige and press attention go to the frontier, and almost none to integration, change management and data cleanup. Nilekani has built a career out of the second list.
Tactic: When a new technology arrives, spend your first month mapping the organizational, legal, and data obstacles to deploying it, not comparing vendors.
Principle 7
Audit your tacit assumptions on a schedule
"Companies run on a set of explicit, tacit assumptions about the market, customers, about the future, world and technology. What happens to them often is that some of those assumptions are changing." Nilekani's three planning horizons at Infosys were mainly a tool for catching beliefs that had quietly stopped being true. The five-year scenario exercise existed to ask whether the company's theory of its own business still held.
It paid off. By 2006, 40 percent of Infosys's revenue came from services that had not existed five years earlier. His 2009 revision of Friedman, from "the world is flat" to "the world has got flattened," was the same habit in public: he kept the framework and updated the assumption it rested on.
Tactic: Once a year, list the five beliefs your strategy depends on that nobody has said out loud, and assign someone to try to disprove each.
Principle 8
Choose your arena by its feedback loop
The 2014 campaign is the clearest failure in the record, and Nilekani's own diagnosis is the most useful lesson he has offered from it. "I like to work on problem-solving, with a three- to five-year horizon," he said after losing by 228,575 votes. Electoral politics rewards different things: years of local relationship-building, as Mohandas Pai pointed out, and coalitions held together by loyalty rather than by results. He had brought a project-manager's theory of legitimacy into an arena that doesn't recognize it.
He didn't try again. He withdrew to advisory roles, committee chairs and the NPCI, where he could still shape national systems through work with measurable outputs on his preferred timeline. The years after the defeat (UPI, the WhatsApp-moment speech, Account Aggregator, ONDC, the G20 task force on digital public infrastructure) were arguably more consequential than a parliamentary seat would have been.
Tactic: Before entering a new arena, ask how long its feedback loop is and what it actually rewards; if neither matches your temperament, find the adjacent role where they do.
Principle 9
Name the moment with an analogy people already believe
In 2015 Nilekani did not argue for UPI from first principles. He used WhatsApp: 30 billion messages a day, 40 engineers, half again the volume of every telecom carrier combined. Everyone in the room used WhatsApp, and every banker in the room had watched it wipe out SMS revenue. The "WhatsApp moment for banking" made an abstract forecast concrete. The 2018 privacy defense used the same method, with Ola, Zomato and a mobile number. So did his description of the grid project as "the next UPI."
Analogies like these move an argument from technical claims, which institutions can always dispute, to pattern recognition, which they can't easily dispute. People don't have to believe a new idea. They only have to see that it resembles something that has already happened to them.
Tactic: Pitch your next initiative as "the X moment for Y," where X is a disruption your audience has personally experienced and lost money to.
Principle 10
Design ownership to dilute yourself
Infosys introduced India's first employee stock option plan in 1993, the same year it went public, and made roughly 500 employees millionaires. The founders' stake fell with the IPO, with the stock options and with the Nasdaq listing. By 2024 the entire promoter group held 14.61 percent and the state insurer was the largest shareholder. The Nilekani family stake was 2.31 percent in 2018. The pattern carried over into his philanthropy, where the 2017 Giving Pledge committed half the family's wealth, and into his platform designs, which leave the value capture to others.
The dilution bought something the founders could not otherwise have had: an institution whose legitimacy did not depend on them. That made the 2017 crisis survivable. When the founders and the board fought, there was an independent board to absorb the conflict and a large shareholder base that wanted stability more than it wanted anyone in particular to win.
Tactic: Write your ownership and control structure on the assumption that you will eventually be the problem, and make sure the institution can outvote you.
Principle 11
When you return, lower the temperature and move the center home
Nilekani came back to Infosys in August 2017 in the middle of a public fight between his co-founder and the board, and he had loyalties on both sides. He did not run the company himself. He took the non-executive chair, accepted the departure of the board members associated with the conflict, moved the center of power from California back to Bengaluru, and hired a new outside CEO, Salil Parekh, within about five months. The "Mr Congeniality" label was partly a joke and partly a description of his method: lower the temperature first, then restructure.
◷
The Recurring Summons
A career punctuated by being called in when a system failed.
1979
Joins Patni Computer Systems after missing the IIM entrance exam through illness; hired by Murthy after solving his puzzles.
1981
Co-founds Infosys with six colleagues; marries Rohini Soman.
1999
Infosys lists on Nasdaq, the first Indian company on a U.S. exchange; Nilekani heads Bangalore's urban task force.
2002
Becomes CEO; revenue grows roughly sixfold to $3 billion by 2007.
2009
Leaves Infosys to chair UIDAI at Cabinet rank.
2011
Parliamentary committee rejects the Aadhaar bill as "directionless"; enrollment continues.
2014
Crosses 600 million enrollments; loses Bangalore South by 228,575 votes.
2015
Delivers the "WhatsApp moment" speech predicting UPI's disruption of banking.
2017
Signs the Giving Pledge; returns as Infosys non-executive chairman after Sikka's exit.
2021
Infosys crosses $100 billion market value, four years to the day after his return.
2026
Tells investors AI carries "execution risk," not opportunity risk; named to lead the national exam-reform task force.
A returning founder is tempted to take the controls back. Nilekani's restraint, taking the chair rather than the executive role and recruiting an operator instead of operating, let the company recover without becoming his company again.
Tactic: If you are called back into an institution in crisis, take the role that lets you set direction and appoint people, not the one that runs operations, and move decision-making physically closer to where the work is done.
Principle 12
Treat the critics' complaint as the next release
Nilekani's handling of Aadhaar's privacy critics was combative in tone ("over-hyped and baseless") and responsive in substance. By his own account, tokenization had been considered in 2010 and shelved as "ahead of its time." When criticism peaked in 2018, the virtual ID and per-organization tokens went in, and he presented them as proof of a flexible architecture. Earlier features, including SMS alerts on every authentication and the option to lock one's own number, read as similar answers to earlier objections.
⚖
Conventional Wisdom vs. the Nilekani Approach
| Conventional wisdom | Nilekani's approach |
|---|---|
| Get legal authorization, then build. | Build to irreversible scale while campaigning for the law. |
| Collect data now; it may be useful later. | Every field costs a billion collections. Collect almost nothing. |
| Own the customer relationship. | Own the rail; let others own the customer. |
| Win the argument, then ship. | Ship, then fold the critics' best objection into the next version. |
| Bet on the frontier model. | Bet on the deployment gap: models are a commodity; integration is scarce. |
| Seek elected office to gain legitimacy. | Tried it, lost, concluded the feedback loop was wrong, and returned to building. |
There is a cost, and it should be stated. Fixing a problem in the next version means the earlier versions shipped with it. For a billion-person identity system, those earlier versions held real people's data. The approach works best when the people in charge actually listen to the critics they publicly dismiss. Nilekani appears to have done both.
Tactic: Keep a running list of your harshest critics' technical objections, rank them by validity rather than by tone, and ship a fix for the top one before you rebut the rest.
Free playbook
Get The Business Model Playbook
58 business models, one visual page each: how the money flows, the metrics that matter, and who runs it. Free when you join the Faster Than Normal email.
Free. No spam. Unsubscribe anytime.
Part IIIQuotes and Maxims
In their words
— Nandan Nilekani, MIT Tata Center, November 2014Scale has to be intrinsic to the design of the solution that you have in mind. Scale is not an afterthought. You don't build a mousetrap and then start thinking about how you can catch every mouse on the planet.
— Nandan Nilekani, The Indian Express, March 9, 2016Every entrepreneur knows that if you wait for all lights to be green, you will never leave home.
— Nandan Nilekani, to VCCircle, January 2016Plumbing is not about politics. It is cutting across horizontally and fixing something. That is what I did in the government.
— Nandan Nilekani, Imagining India (as quoted by the BBC)I was wary of being labelled a 'computer boy' who saw every problem as something that could be solved by writing a piece of code. After all, what do computers and software have to do with clearing garbage or provide soft drinking water?
— Nandan Nilekani, Infosys AI Day, February 17, 2026There is no opportunity gap. If anything, the opportunity is bigger than ever before... So it is not an opportunity risk. It's an execution risk. You get it?
Maxims
- Number first, method second. A target large enough to rule out every existing approach forces you to invent a new one.
- Speed is armor. Grow faster than your opponents can organize, until undoing the work costs more than living with it.
- Every field is a billion fields. At population scale, simplicity is the precondition for anything to work at all.
- Own the rail. Infrastructure that refuses to compete with what is built on it becomes infrastructure everyone builds on.
- Don't wait for green lights. Start with the authority you have and campaign for the authority you lack while you work.
- The gap is deployment. Technology always runs ahead of institutions, and the value lies in closing that distance.
- Schedule your doubts. Tacit assumptions expire quietly, so review them on a fixed calendar.
- Respect the feedback loop. Choose arenas whose rewards and time horizons match how you work, and leave the ones that don't.
- Borrow a moment people already lived. An analogy to a disruption the audience has felt persuades better than a forecast.
- Build to outlast yourself. A system that needs its creator to keep running hasn't been finished yet.
