When Ginni Rometty Took IBM, Only Two in Ten Had the Skills the Future Required

Lex Fridman opens with scale. Ginni Rometty spent more than forty years at IBM, starting as a systems engineer and running the company as chairman, president, and CEO from 2011 to 2020. IBM, he notes, is one of the largest technology companies in the world, with about 280,000 employees. What, he asks, are the biggest challenges of running a company that size?

Rometty turns the question. The biggest challenges, she says, are not in running companies like IBM. They are in changing them, and in knowing what you should change and what you should not. People rarely ask the second half. "What should endure, even if it has to be modernized, but what should endure?" Then came the part she found hardest: changing how work got done inside a company that large.

Fridman, a romantic about long-running companies, guesses that IBM's core is its public moonshots: Deep Blue, Watson, the research that shows off. Rometty corrects the size first. The research division itself is more than 3,000 people, not a side show. What should endure, though, is quieter. In the 1930s IBM built the Social Security system. It helped put a man on the moon. To this day, she says, banks do not run and railroads do not run without it. At its core IBM is a business-to-business company that does mission-critical work.

When she started as CEO she felt people were still looking for a word for that. Together they landed on *be essential*. She loved it because she cannot call herself essential; someone else has to determine that she is. The surface keeps changing. Today IBM is a software and consulting company, with consulting about a third of the business and the software bet placed on hybrid cloud and AI, a structure that would not have been true two decades earlier. Moonshots do not always work. Mission-critical work does. The identity that should last is doing the things the world cannot pause.

Size still brings the failure mode people expect. In a big company, Rometty says, people start treating the process as the client. "Your process is not your customer. There is a real customer here that you exist for." It is easy to fall into because people become masters of the process, and that is not right. The other tax of size is speed. When she took over in 2012, IBM had catching up to do, and cloud, big data, mobile, and social were hitting at once, an unprecedented pile-up of trends.

Her first move was to tell people to go faster, and faster, and faster. "Honestly, I've tired them out." That is when she learned a lesson she keeps repeating: at that scale, the *how* may be more important than the *what*. If you do not change how work is done, give people new tools, and help them with skills, they will just do the same thing faster. She uses a picture. You tell someone to run a marathon and they are wearing hiking boots. They are not refusing to run. The shoes are wrong.

Big-company shoe changes start with what she jokingly calls blunt clubs. Everyone does them: cut layers. She reduced layers of management by about half. The math is simple. Fewer decision points, faster decisions, more autonomy, and a destination that is not filtered through seven or eight levels. Grass grows back. Layers come back. You have to keep cutting. The disease those cuts are meant to treat is one Fridman has heard everywhere: nobody can say yes, and everybody can say no.

After the blunt club came the longer journey, the one that sounds like a buzzword until you try it at scale: agile. Not a forty-person pod off to the side, but small multidisciplinary teams working from the outside in, setting interim steps, taking feedback, pivoting, and applying that method to more than products. IBM ended up training hundreds of thousands of people on how to do it correctly. When you manage a couple of hundred thousand people, she says, you have to get a lot of them working that way.

Rometty's book and her style of leading are full of *we* and *us*. Fridman asks the sharp version: are there times a leader has to step in, almost autocratic, and make hard unpopular decisions? Of course yes, she says. There is a leader for a time, and there is a leader for a situation. "Anytime you have to run a company that endures a century and has to endure another century, you will do unpopular things. You have no choice."

She often felt she had to sacrifice things for the long term. Sometimes that meant job changes or reductions. Sometimes it meant changing the way IBM did semiconductors, a whole different philosophy. In a crisis, she repeats a line she used on herself: it is not a popularity contest. None of these jobs are, whether the company has one person or half a million.

Psychologically it is still hard. A new CEO walks into decades of tradition and into rooms of people who really are experts in their domains, and says the work has to be done differently. The temptation, when you have to change a company, is to throw everything out. That is why she keeps returning to what must endure. She knew on day one how much had to change. The further she got, the more she could see: a lot more had to change.

IBM had always done its best, she says, when it had a technology platform: mainframes, the PC, middleware, even services as a platform. It needed the next one. Skills were the uglier number. When she took over, the company inventoried who had modern skills for the future. It was two out of ten. Not that people lacked relevant skills for today. For the future, two out of ten. "Yikes. That's a big problem."

Her view of hiring changed four times, each time because reality pushed. Forty years ago she was in a hurry, squeezing interviews into a calendar so someone could get the work done. Then someone told her to hire the best people and the job gets easier, so she spent more time on it. Then came experts. She hired a lot of them. The epiphany arrived while she was running the company and had to change skills: if someone has been an expert at the same thing for thirty years, the odds they really want to change a lot are pretty low. In a dynamic industry, that is a problem.

The talent market made the same point from the other side. IBM needed cyber people. Unemployment in the United States was almost 10 percent, and she still could not find them. Schools were not teaching the right things. Then, almost by accident, she reviewed a corporate social responsibility program: a fledgling school in a low-income area, a high school paired with a community college, internships, and curriculum advice from IBM. They could hire those kids. "This is not CSR. I just found a new talent pool." IBM had been hiring 99 percent college graduates and PhDs. She is vice chair at Northwestern and is not against higher education. The sentence that changed her hiring logic was simpler: aptitude does not equal access.

The number-one trait she would hire for now is willingness to learn. Curiosity beats everything else on the résumé. Underneath that is a deeper change of mind: you do not think of buying skills, you think of building them. On the cloud and AI path she brought in plenty of outsiders, "shiny objects," and watched two groups form. Some did fantastic work. Others arrived ready to school IBM, without understanding that a slip can break a bank, or how hard the security and auditability requirements really are. Meanwhile some IBMers said they could learn, and some said they did not know how, which was exactly the thing to learn. "There was not one group as a winner and one was a loser. The winners were the people who were willing to learn from each other." Fridman adds that nobody is truly skilled in AI yet, because the field is still forming. She agrees completely.

Fridman reads a passage from her book. Her drive for perfection often meant she only focused on what needed to change, without acknowledging the positive, and that could keep people from trusting themselves. It took her a while to learn that just because she could point something out did not mean she should. She still spotted errors, but she became more deliberate about what she mentioned and sent back. She also tried to stop assuming her way was the best or only way. Giving other people control builds their confidence. Constantly trying to control people destroys it.

The flip answer about balance, she says, is that as the company gets bigger you have no choice: you have to let people find their way. A leader still has a few jobs you always do yourself. Values and value-based decisions are one. Young companies struggle here because the values are not deeply rooted, and when the storm comes they are easy to uproot. The story that taught her was early-eighties, first-time manager. A high performer was telling dirty jokes that offended women on the team. Her boss was unequivocal: it stops immediately, or you fire him.

The other thing she felt she had to role-model was learning. IBM's longtime motto had been Think. She turned it into Think Academy: the first Friday of every month, compulsory education, everybody, she did not care what your job was. When the whole company has to transform, everybody has to have some skin in the game. She taught the first hour of every month for four years. She had to learn something in order to teach it, and the point was the demonstration: if she could teach it, who in the company could not?

Fridman admits he wings things, pushes work to the last minute, and is almost always late, as if the anxiety of a deadline is what produces the performance. Rometty is a preparer, and the habit has deep roots. As a child she was not a memorizer. Her brother could read something once and be done. She always wanted to understand how something happened, whether it was an algebra theorem or anything else. Don't give her the answer. She wanted to figure it out so she could reproduce it and did not have to memorize.

Engineering school in the 1970s added a second reason. She was the only woman in the class. People still walk up and say they remember her. She tells them she is sorry she does not remember them: there were thirty of them and one of her. She learned that whatever she said in that room would be remembered, good or bad, so she had better study hard. Early on, preparation was a shield for confidence. The more she prepared, the more confident she felt, and she says that is probably still true.

The third reason evolved later, and it is the one she cares about most. If she is really prepared, then in the moment she can really listen, because she does not have to be assembling language on the fly. She can take what she already knows and help the situation. Preparation, in that telling, is not a performance. It is how you become present. Her husband, she adds, does not prepare at all. He calls himself a type Z; she is a standard A. They have been married 43 years, and that configuration seems to work.

Fridman asks whether hard work, so often tied to burnout, is also what made her happy. She enjoyed it, and she is careful to say so, because she does not think everyone should take her route as the only way there. As more people watch you, you are a role model whether you like it, know it, or want it. She learned to be explicit: just because I do this does not mean you must. Then she turns to power. Ask people if they want it and many say no, they would rather do good. "The irony is, you need power to do good."

She writes her life as three rings, like a pebble in water: the power of me, the power of we, and the power of us. The hard-work ring starts at home. Her great-grandmother came from Russia, never spoke English, and cleaned the Wrigley Building in Chicago. If she had not saved every dime, Rometty's mother would not have had a home or a car. Her grandmother made lampshades seven days a week in the 1940s and 1950s, and taught her to sew clothes they could not afford. Her father left. Her mother, who had never worked a day outside the home, was homeless, with no money, no food, and four kids, and climbed out of it. Rometty associates hard work with strong women, and with being sure you can always take care of yourself.

She still remembers the last two weeks of freshman year, when she had 25 cents left. She put the quarter in a clear plastic box on her desk and stared at it. This is it, she thought. No more money. Fridman asks whether that desperation played a role in the drive that took her to CEO. She had just asked another American CEO whether part of his success came from bad things happening, and he said yes. She will not say you have to have tragedy. It does teach one important thing: there is always a way forward, and it is in your control. Friends worry that their kids have never had hard times. Required? She does not know. It sure does not hurt.

The line people quote back to her most is one she shamelessly stole from her husband: growth and comfort never coexist. Mid-career she was already running a big business when her boss got promoted and called to say he wanted her to have his job. She said no way. She was not ready. It covered every product line, development, every function, around the world. He told her to go to the interview anyway. The next day the interviewer offered her the job. She asked to go home and talk to her husband.

She told the story at length. He asked, "Do you think a man would've answered it that way?" He knew her: in six months she would be bored, and all she could think about was what she did not know, while other people with less skill thought they could do it. She names the stereotype because it resonates with many women: they are often their own harshest critics. She took the job. When she went back to the mentor who had been looking out for her, he looked at her and said, don't ever do that again. Being uncomfortable was fine. Using it as a public shield was not.

Now she takes stock of what she can do, then looks for times to be uncomfortable. If she is nervous, she is learning something. She asks Fridman if he was nervous to meet her. He says he is still terrified. Fridman pushes back on treating self-criticism as only a women's problem, or as something to silence. He has it too, and he thinks the anxiety is also an entrance to growth. She agrees there is a balance. Extreme confidence makes other people value you more in the short term. Too much of it, and you stop growing. Her own test for a new job became this: if she was willing to ask lots of questions and understood enough, that was all she needed to know.

Her book is titled *Good Power: Leading Positive Change in Our Lives, Work and World*. The essence, she says, is doing something hard or meaningful in a positive way. She hopes one day she is remembered for how she did things, not just for what she did. Against a cartoon of bad power, good power means three things: you embrace and navigate tension instead of running from it, so you bridge divides rather than deepen them; you do it with respect, the opposite of fear; and you celebrate progress versus perfection, because chasing a private definition of perfect ends in polarization or paralysis.

Writing it was hard because she does not usually talk about these things, and nobody wants a scientific description. They want the stories. In retrospect the memoir fell into the power of me, the power of we, and the power of us. The we chapter is five principles for driving change well. The first is to be in service of something, which is different from merely serving. She had a knee replaced and could tell which surgeon cared that she could walk again and which one only cared that the hardware was fine. A waiter can serve food or care that you had a good time. If you are really in service of someone, you believe that fulfilling their needs will fulfill yours.

The other principles include building belief so people spend discretionary energy instead of waiting to be ordered, changing how work is done, and what she calls good tech. Everyone is a tech company now. If you invent something, you look at the upside and the downside. Quantum computing helps materials and risk calculations; it can also break encryption, so IBM worked equally hard on algorithms that would survive it. She was talking about AI ethics a decade before this conversation, and not many people wanted to listen. An older company, she says, realizes society gave you a license to operate and can take it away.

Ultimate good power, for her, is doing something for society, and how it is done. The leap from me to we is often more gradual than a conversion. Sometimes it happens because you cannot get the work done alone. Sometimes it is a jolt. In Australia in the 1980s or 1990s she gave what she thought was a great speech on financial services. A man walked up afterward, and she expected a technical question. He said he wished his daughter could have been there. Until that night she had wanted people not to notice she was a woman. She had refused women's conferences. Then she remembered you cannot be what you cannot see, and accepted that she was a role model whether she liked it or not.

Fridman quotes the old line that power corrupts. How did she resist the delusion that she was smarter than she was? Surround yourself with people who will hold up a mirror, she says, and do not keep a court of fans. A longtime colleague who had once been a competitor would tell her the truth until Rometty rolled her eyes, and then tell her not to roll her eyes. Diversity, as she defines it, is the same defense: you get a better product and a better answer. Inclusion is a choice you can make every day. When people say they cannot find the kind of people they are looking for, she says they are not looking in the right places.

Every company, she says, faces pressure to sacrifice what is good for the world for the bottom line. She has been vocal about stakeholder capitalism and helped on the Business Roundtable purpose statement. You only get to be a hundred years old by balancing stakeholders, short term and long term. People write about it in black and white. She has had to live in a gray world. IBM had been an originator of the commercial semiconductor industry, still doing its own R&D and manufacturing, and losing a lot of money, while it also had to fight a war on cloud and AI. Shareholders would have said shut it down. Those chips also powered systems that power today's banks. The answer was not stop, and not keep pouring money in. They sat in an uncomfortable spot until they could separate manufacturing from research and keep the R&D.

Working with giant clients, she says, is simple if you honor your promises over a long time. On a European trip, in a climate of national tech champions and arguments about data and security, she began explaining to Chancellor Merkel that IBM systems had no back doors and did not share customer data. Merkel stopped her: you are one of the good guys. That was not about Rometty personally. It was about a company that had acted consistently with its values for decades. Values, she says, are not fluffy. If she were starting a company now, she would spend a lot of time on why you do what you do, and what is not tolerable.

Fridman brings up the 22 consecutive quarters of revenue decline between 2012 and the summer of 2017. The urgency, she says, was always high, even if she looked calm. IBM has one of the world's largest pensions, a huge workforce, clients whose businesses do not run if IBM fails, and shareholders. It was perhaps the largest reinvention the company had ever had to undertake, and some headlines called it existential. In that stretch, 50 percent of the portfolio changed and she divested $10 billion of businesses. Net of divestitures and currency, growth was flat. Flat is not great. For a transformation that size, she was proud. Her little nephew asked if mean talk on television made her mad. She told him she was doing what had to be done. She knew it was needed for IBM to live a second century. Red Hat, a $34 billion deal and at the time the largest software acquisition ever, was a short-term hit and the foundation of the hybrid-cloud bet. They had been late to public cloud, found it tuned for consumers rather than mission-critical work, and concluded the real world would be traditional systems, private clouds, and public clouds that still had to connect securely.

Before the AI history, Fridman pulls her back to 2002, when she championed IBM's $3.5 billion purchase of PwC's consulting arm. PwC was heading toward an IPO; IBM swept in. You acquire hearts, not parts, she says: people can walk out the door. You also acquire a private-partnership culture and drop it into a public company that thinks in quarters. Someone asked how long they had been at IBM and got the answer 143 quarters. Clients who had never negotiated that way suddenly wanted to reopen prices every quarter, and margins moved fast. She learned that if people do not believe a target is possible, they probably will not make it happen, and she had to move some leaders. By the time she stepped away she had run the business for seven, eight, nine years. It was worth $19.5 billion and made $2.7 billion in profit. She gave the team a paperweight with a Thomas Edison line: many of life's greatest failures are people who gave up right before they were going to be successful. On negotiation she wants both sides a little fussy. If one side thinks it did great and the other thinks it did horrible, the deal is usually born bad.

Deep Blue beating Garry Kasparov in 1996 and 1997 is, for Fridman, one of the great moments in AI history. Rometty, who is not a chess player, saw it first as supercomputing. Chess has a finite number of moves, so the demonstration was how fast a machine could crunch through them. She grants it was a milestone because it put man-versus-machine in public consciousness, and because games have been used that way ever since. Fridman, who grew up in the Soviet Union, did not experience chess as computation. It felt like memory, pattern, intuition, guts. The pattern he hates is that the moment a machine crosses a threshold, everyone says it was just brute force. Rometty's line on what comes next is that AI is a reflection of humanity: teach it bad things from the past and you teach bad things for the future. Efforts to regulate the technology itself she calls a fool's errand. Regulate uses.

Watson on *Jeopardy!* was inspired by that chess match and, she thinks, helped pull AI out of another winter. It humanized the technology by playing two champions, lost first, then won, and showed natural language plus a bit of reasoning over a lot of data. It also set expectations very high. IBM took healthcare first and had to pivot. The challenges were not mainly the technology. They were people: do doctors trust it, do they have time to learn another way, and does dropping AI on unchanged processes just pave a cow path. They also picked very hard problems to start with. A lot of technology now starts simple, builds trust, and works up. She has no regrets. Watson was rebuilt many times, made more modular, and today sits as AI inside many things, an ingredient rather than a destination. Thousands of IBM clients still use those components. She would rather open jobs as an assistant than pretend the system replaces judgment. French bank unions, she notes, once voted for it because people did a better job.

On HAL 9000, she has looked for evidence that Kubrick named it after IBM, one letter off, and thinks the decoding came after the fact. What she does think is that it was an early picture of AI taught badly. Around 2015 or 2016 IBM published principles of trust and transparency: AI should be explainable, you should know who taught it, what data went in, and that it is being used. People will not trust the technologies, she says, unless they have that kind of transparency. Impostor syndrome, when Fridman asks, is something she hears a lot and has spent time helping people with. The bigger the job, you look left and right and think, what am I doing here. Then you remember the work, and that you are in a position to make something change. She does not dwell on the feeling for long. She focuses on the work.

A lesson from her mother: never let anyone else define you. Most of her self-discovery, she says, arrived a year or two after the moment, when she looked back. On gender she chose, as a product of the 1970s and 1980s, not to look at it as the frame. The women around her thought the way to get ahead was to work hard, and she worked at companies steeped in equal opportunity, where she was evaluated on whether she hired a diverse team. That does not mean it never mattered. A colleague pointed out that media sometimes names a woman CEO when something goes wrong and names the company when a man is in the chair. With so few of you, she says, much of what you do will be magnified. People still tell her they cannot wait for their daughter to read the book. She worked hard to write it for everyone. She just happens to be a woman.

Through OneTen she is trying to put a million Black employees into the middle class without college degrees, by getting them the right skills and upwardly mobile jobs. At IBM she once heard a highly accomplished Black colleague say that if he failed or succeeded, it felt as if a whole country in Africa were on his shoulders. She recognized a version of that burden in the headlines when she was named CEO, as if her success or failure would be a statement for the whole gender. Role models matter because the category, tiny or huge, is how people decide they can see themselves in the work.

Her family still jokes that she never had children because she had already raised her family. When her mother went back to school and to work, Rometty was the one who went to the teacher meetings. She is still "mama bear" to her siblings. She does not want anyone to believe you cannot do her job and have children. She names friends who did both: Marillyn Hewson at Lockheed Martin, Mary Barra at General Motors, Ellen Kullman at DuPont. The choice she and her husband Mark made, both coming from families that had been through a parent's absence, made it easier to focus on work, and she is honest that that is just math. It also made her empathetic about policy. Her sister, with a Northwestern MBA and a career at Accenture, hesitated to return after time with family because she thought technology had passed her by. Rometty went back to IBM and asked for everyone who had left for parental or family reasons and not come back. That became a returnship program: one month, three months, whatever it took. A lot of people discovered that not that much had happened, or that they knew more than they thought.

Advice for young people, and for anyone trying to build a life they can be proud of, is modest. A life you can be proud of is one where you leave something a little bit better. It does not have to be big. She reaches for Churchill: you make a living by what you get, and you live a life by what you give. Her own advice is to ask more questions than you give answers, and to be a sponge. She asked her husband the same question. He paused and said patience. He sees so many young people in such a hurry to somewhere they cannot name. If they had patience and let life unfold, they might be surprised where they ended up. She thinks that is a really good answer.