Phil pods with Cognizant Co-Founder Frank D'Souza: Why AI is breaking the IT services industry

Sure, Phil. First of all, thanks for having me on the podcast. I was the founder, co-founder of Cognizant Technology Solutions in the early 1990s. I spent 26 years at Cognizant, including 12 years as CEO. We grew that company from a startup to 280,000 people, $16 billion of revenue, and about $40 billion of market cap when I left about six years ago. I created Recognize, which is a private equity business that focuses entirely on investing in the next generation of digital services businesses, and I've been doing that for the last six years. It's great to be with you.

You know, Phil, I think there were two or three things. The first is that we founded Cognizant at a moment when the industry was going through a massive operating model change, and that was the offshoring moment in the early 1990s. There was a whole confluence of things that happened at the same time, low-cost bandwidth, the internet, and so on, that enabled global delivery to happen at scale. We were in the right place at the right time. The second part was that we executed really well. We had an incredible team, and we spent a lot of time on this idea of talent density, putting the best team against the highest opportunities. The third thing is that we created for ourselves and with our investors the opportunity to invest in the business. We had a rule of 40 business before we talked about the rule of 40, which was that we maintained our margins at a level consistent between 19 and 20%, below where our competitors were, to give us the room to reinvest in growth. Our investors should expect to have faster growth as a result of that investment. So I think it was those three things: we were at a moment when the industry had tremendous tailwinds, we had the best team in the industry, and we had the space to make the right investments. Those things together created some magic.

Yeah, I would absolutely, Phil. When industries, whether it's our industry or others, go through moments of big change, you have the well-documented phenomenon of the innovator's dilemma, which is that the incumbents tend to focus on the current operating model, the rules of competition that they are comfortable with, and it's very hard for incumbents to reinvent themselves. At those moments in time, the challengers have an advantage. They don't have the baggage of the past. They start with a clean sheet of paper, and they're able to disrupt the status quo. You see the pecking order shift at these moments of great disruption.

Yeah. In the white paper, I talk about this idea that code in some senses has never been more valuable. If you think about some of the most valuable companies in the world, we talk about Anthropic, OpenAI, and SpaceX, these massive IPOs that are lining up, companies that have achieved revenue growth in periods of time that have never been seen before. What underpins the value of many of these companies is their code, their product which is embedded in code. At the same time, the act of producing code, the cost of that, is falling to zero. Large language models today are incredibly capable of writing code, and the marginal cost is tending towards zero. So you have this paradox where code has never been more valuable, coding is becoming lower and lower cost. If you think about that in the context of any other industry, if you were the CEO of any other company where your product has never been more valuable and your factor of production is becoming cheaper, that would be a formula for tremendous value creation. The issue in our industry is that we have historically anchored pricing on the input, not on the output. You don't go to buy a new car and say, 'I'm buying four tires and a transmission and spark plugs.' You just go buy the car. In our industry, for some reason, buyers have got used to selling the tires and the chassis and the spark plugs. So I think the industry has this great opportunity to move to output and outcome pricing, and then have a situation where the factors of production change and create tremendous value.

I think so. The industry has been talking about this for a long time, Phil. It's been accelerated by AI and generative AI in particular, but as long as I've been in the industry, we've been talking about moving more towards output and outcome-based pricing and commercial models. The industry understands how to do this. There are some places where it's quite possible, and by moving to output and outcome pricing, you align interests between the service provider and the client much more strongly. So I think it's possible. But it's not an easy transition for the incumbents because everything about the industry right now, the culture, the way people have been trained, many of the incentives and metrics that we traditionally use to track the health of a business are anchored on what I call the scarcity era, where producing code was the scarce resource. We anchored on utilization, numbers of people, and many input measures. Making the shift is going to require firms to focus on a new set of metrics and then change the culture and the compensation accordingly.

Absolutely. I've always believed, Phil, and I continue to believe that one of the most perhaps even underappreciated aspects of the services industry is the fact that because any individual company has clients across a range of industries, technologies, and functional areas, that diversity of work attracts the best and the brightest people to come work in the industry. People want to have breadth of experiences, to experience different technologies and industries. Structurally, I believe the industry is advantaged in that it can create more interesting career experiences for consultants. I believe that we're going through a moment right now where everything is changing, technology is new, AI expertise is scarce, but this is a moment when the industry can shine by attracting the best and the brightest to come and do their best work for the great clients that the industry has.

You know, Phil, when I talk about the J curve, what I mean is that there's a gap. For the existing work that any firm is doing, if you're really deploying AI and using the tools to their fullest potential, you're likely going to see a compression. For a given amount of work, you'll be able to do it with fewer humans because AI can fill that gap. If your pricing model is based on an input price, which is the number of billable hours, then you're likely going to see some compression in revenue. You might have this J curve of revenue, where you find some compression before you start to work your way out of it. Most firms that have been in the industry for a while face some degree of J curve. If you're not facing that J curve, it's probably a sign that you're not deploying the latest AI technology in the most effective way. If your revenue per employee or profit per employee is flat or declining, it's a pretty good sign that you're on the wrong side of the J curve. If it's starting to flatten out or increase, then it's probably a sign that you're through the J curve and on the other side of it.

Yeah. I believe, Phil, that if you step back, there is Jevons paradox, this idea that as the cost of something comes down, the demand for that thing goes up. I think that over time, that is true in software. The world is clearly becoming more software and technology intensive, not less. The need for software is going to be substantially greater than it is today. The challenge right now is that in many of the public companies, we are still using scarcity-era metrics when we're talking to investors, and it has not been made clear how to dimension the J curve. If we were clear with investors that we are in this moment of transition, which everybody intuitively understands, but that there is a path through it, and we were able to dimension that path, we could explain to investors that you have to accept that there is a productivity gain from AI. That causes a short-term compression on the existing revenue base, but in theory, if done right, should improve margins. Once we get through that, we come out the other side. That's a message and an explanation that investors are capable of understanding and would welcome.

Yeah. It has always been the case that great technology innovation leads to a great tailwind for the services industry. That continues to be the case here. The frontier labs that continue to innovate and produce ever better models, ever better vertical slices, ever better functional versions of those models, every time there's that level of innovation, it creates an opportunity for services firms to focus on what I call the first mile and the last mile. To take that technology and make it real, make it do something important, and do that in a safe, ethical, compliant way requires a set of services around the edges. The courage that the services firms need to have right now is to say, 'Some of the things that we did in the past, AI is better at doing. AI is generally better at writing code, writing test cases, and those kinds of things than humans. That is progress.' The services firms need to stand up and say, 'These are the parts of my business where productivity is going to have a dramatic effect, but these are the other opportunities that are emerging where we think we can grow.' On balance, that's going to be a net positive for the industry.

I view it as a tremendous opportunity, Phil. I'm very much in the latter camp. In the first days of generative AI, there was this great fear that LLMs were going to do all the things that humans do, and there would be no need for services firms. The fact that these large language model frontier lab companies are actually partnering to create their own services business is first and foremost an acknowledgement that the technology still needs humans in order to make the technology work and do something productive for the client. These services businesses and deploy codes to me are a great acknowledgement that the future belongs to humans and agents together deploying solutions for a client, not one or the other. That model is now clear, and the investments that you've seen make that model clear, which then makes it an equal opportunity for the services firms to partner with the frontier labs to create that same set of services.

Oh, for sure. And I think, Phil, that's not just in our industry, not just in software or technology. It's almost a cliche now where people are saying, 'You're not going to lose your job to AI, you're going to lose your job to somebody else who's using AI.' The jobs are changing. For anything, whether you're a software developer or anything else, are you skilled at using this technology in the most optimal way possible, freeing yourself up to do the things that are uniquely human? Sometimes I talk about art when I'm asked about artificial intelligence. We talk about artificial intelligence, but I don't think we yet have artificial wisdom. Wisdom is still the exclusive domain of humans because wisdom is the sum of all of our human experiences as individuals and the sum of humanity. Computers haven't been able to capture all of that. The wisdom of Phil First doesn't exist in any large language model anywhere. The knowledge that Phil has, some of that may exist in a large language model. The collaboration between agents and humans is a collaboration between artificial intelligence and human wisdom. Those things when they come together will allow us to discover new things, will allow us to make the world better. That's the great promise of this new technology.

Yeah. So, Phil, you and I have had some debates about this issue offline, this idea of what I call the talent geometry. The industry used to focus on pyramids of humans organized in pyramids. I believe that the future is going to belong to teams that look more like a diamond shape. You call that a doer model. Whatever you call it, today's AI automates many of the tasks that less experienced folks used to do. If you think about the old model of software development, and I use software development as a proxy for everything else because large language models are particularly good at it, coding was usually done by junior developers, testing was usually done by junior developers, and that's going to increasingly become automated. The real productive work is going to be done by a diamond-shaped team of people where experience is going to matter. There are two important things to keep in mind. Number one, this is a tale of two middles. The middle of the diamond is not the same skills as the middle of the pyramid. The people in the middle of the pyramid were largely people managers. Their job was to manage up and manage the teams below them. In a diamond, the middle is not about people management. It's about the first mile and last mile skills needed to direct and manage AI. The middle of your diamond is really directing and managing AI. The middle of the pyramid is managing and directing humans. It's a very different skill set. Number two, I don't know if over time services firms will have fewer people. That seems logically inconsistent, but the reason is that we still need to find a way to bring people into the industry to train them to make them functional at that middle level. The gap that gets created at the bottom as we move from the pyramid to the diamond gets filled by very deep apprenticeship programs. We're going to have to bring entry-level folks into the industry but train them in the same way that we train other engineers. If you imagine civil engineers who build bridges or nuclear engineers, you wouldn't dream of bringing somebody in from college and saying, 'We're going to give you two or three months of training, now go build a bridge.' We have very structured apprenticeship programs where they get to watch and learn and sit alongside experienced engineers. The same will be true of software going forward. One idea I've put forward is that those apprenticeship programs perhaps should be collectively done by the industry as opposed to individually by firms. This may be an area for the whole industry to come together and create an industry-wide apprenticeship program so that everybody can build a common set of skills that benefit the whole industry.

Yeah. You've brought up a very important, maybe the most important point here. Even if you put aside the issue of how we bring people into the industry, you can't build a sustainable culture in a firm by hiring middle-level folks from the outside and trying to integrate them. You build culture from the bottom as people come in and grow with the organization. To build a truly sustainable firm, even if there was talent in the industry that you could just hire from the outside, individual firms absolutely are going to need to build the internship programs from the bottom up in order to build a sustainable culture. We also have to think a lot about culture here and how firms will build culture in a world where the talent geometry looks more like a diamond.

Yeah, look, I think the industry has grown dramatically over the last decade, and maybe there has been a little bit of that what you describe, but I don't think it has to be. At moments like this, at times of great change, we are at a moment in time like none of us have seen before in our careers. AI, or let's call it generative AI, is the most profound technology of our lives. The ability for a services firm to put together a compelling business and a compelling culture to do great work for customers has never been higher. People want to come to a place where they wake up every morning and say, 'This is the place where I can do the best work.' That opportunity has never been higher. We are at an incredibly exciting moment in time. The great technology services businesses have that opportunity right now. Clients need help. The rules of deploying AI are just forming. Wisdom about how to deploy AI is scarce. If you can build a firm that is on the bleeding edge of that, two things will happen. You'll attract the best people. Those people want to stay, and therefore a great culture can emerge, can be reinforced, can continue to exist.

I think that's a little unfair, Phil. If you think about the industry and the growth of the industry, we've had great leaders building great firms. The industry is facing an incredible moment of change, and that requires clear-minded, courageous leaders. I think we have them in the industry, and those people will emerge. Right now we're in the midst of that transition, and when you're in the middle of a storm, it's sometimes hard to see clearly. When we look back on this industry five years from now, there will clearly have been winners and losers, as always happens in times of great disruption, but I have no doubt that great leaders will have taken their firms through this transition as well as they have in the past.

You know, Phil, I don't think it's one or the other. I think it's both. In this transition, both have relative advantages and continue to have relative advantages. The smaller companies have the benefit of being small and nimble, and that brings a set of advantages. But you can't ignore the fact that the big companies have incredible client relationships, incredible distribution, and an ability to invest both organically and inorganically that the smaller companies don't have at the same magnitude. The question is not whether one group is going to win at the expense of the other. The question is whether the leaders of these respective firms are going to truly understand what assets they have at their disposal—people, capital, client relationships—and capitalize on those assets in a way that's unique and different. The big firms have incredible client relationships. Yes, they have a much bigger change management challenge than a smaller firm because they've got a hundred thousand people, hundreds of thousands of people in some cases, that they need to take to the next level. But they've got tremendous assets as well. That's the chessboard right now. How you make the next set of moves is going to count. If you're the CEO of one of these businesses, how you play the chessboard has always been important, but I think at these moments it's going to be even more important. If I go back to the Cognizant example, when we started the business in 1994, we were tiny. There were 700 other firms just in India at the time. By definition, we were behind 700 other firms, but we played our hand smartly, and as a small company, we were able to outmaneuver many of the larger ones. Having said that, many of the large firms continue to thrive and survive. It's not one or the other. It's how you play your cards.

Yeah, Phil, the firm you're referring to is Ciklum. It was our first investment at Recognize. When we invested, a significant amount of the firm's talent was in Ukraine. This was before the war. We've had to manage this business through a very difficult period of time, and I'm incredibly proud of the Ciklum team. I'll go on record every day saying that this is a team that's resilient, incredibly committed to the mission, that wakes up every morning in very difficult circumstances and says, 'How do I do the best for my customer? How do I deliver well?' It's been a privilege to work with this team over the last several years to manage this business through a very difficult circumstance. Against that backdrop, the transformation that we have accomplished at Ciklum has been incredible. We've taken this business through the J curve. When we invested, the business was largely a staffing business. They had begun the transition from a staffing business to an output outcome project-based business, but they were very early in that journey. In addition to managing the business through the war and having to diversify the geographic locations from Ukraine to other parts of Eastern Europe and to India, we've also had to manage the business through this J curve. We've been able to show that you can manage the business through the J curve. Revenue at Ciklum declined for a while but now is very much on a growth trajectory. In the process, we've moved the entire company from a pyramid to a diamond-shaped talent model. We've been able to protect margins in the business, both gross margins and operating margins. Revenue per employee has gone up. We've completely reshaped that business from a traditional low-margin staffing business to a diamond, high-value, AI-first development shop. We've done all of that against the backdrop of a war in the country. There's a lot to be learned there. I hope that I never have to manage a business through a war again, but I'm incredibly proud of what we've done at Ciklum. That should be the topic of a podcast all on its own because it's such a great story.

You know, I think we're going to be talking about the firms that made it through this transition and perhaps those that didn't. Those that made the transition to talking about their business in terms of what I call outcome density, the business value that the firms deliver per person per quarter. We'll be talking about firms that showed that they could drive up revenue per employee, gross margin per product. We'll be talking about the firms that talk about their business in terms of share of revenue from fixed-fee engagements, from certainty bundles, from outcome-based contracts. Firms that can credibly talk about their business being tied to proprietary IP that is code or data models or agents, rather than staffing or people or utilization. Those will be the firms that make it through this J curve. We'll be talking about the firms that did that successfully and the firms that didn't quite make that transition.

Yeah. Any moment of big disruption like this is going to feel chaotic, both from an organic standpoint and an inorganic standpoint. I do think that M&A is a very valuable tool for firms to make a transition. You can try to change and make the change organically, or you can try to find companies that have already made the change and acquire those companies as a way to seed the change within your organization. You'll see that pattern. You'll also see the pattern of perhaps weaker companies that aren't going to make it being acquired by stronger companies who have the ability to either reshape those companies or reshape their cost structure to make them more efficient. You'll see all of that. The last thing I would say is that very large-scale M&A in the services industry has been difficult because of the cultural challenges of integrating two large firms together. So I don't know how much of that you'll see or will be practical going forward.

Phil, thanks so much for having me, and thank you for what you're doing for the industry. I think what HFS is publishing is very much on point and on the bleeding edge of thinking around how the industry is emerging. Thanks for everything that you do for the industry. Appreciate it.