Audio · 2026-04-17 · 57m · 24 moments

Scaling Global Organizations in the Age of AI with ServiceNow Chairman and CEO Bill McDermott

Few teens are business owners, but by age 16, Bill McDermott had purchased and was running a local deli. Now he runs leading global technology powerhouse ServiceNow, a company that is defining how the world’s largest organizations transform for the digital age. Sarah Guo sits down with ServiceNow CEO Bill McDermott to discuss his journey from child entrepreneur to CEO, and how he navigates his role as a leader in the age of AI. Bill argues that human connection is still a vital part of being a s ✦ AI generated

Apple Podcasts ↗

GUEST Bill McDermott HOST Sarah Guo HOST Elad Gil

timeline · colored by role

Data◆

Enterprise platforms are far more cost-effective than trying to replicate their functionality with language models.

McDermott argues that replacing an enterprise platform with a language model is roughly 10x more expensive when you account for rebuild costs, human capital, and GPU/token expenses.

transcript

Bill McDermott: The cost to replace an enterprise platform in this SAS apocalypse that people talk about is an extraordinary expense to try to match what some of these really important enterprise platforms do. Because you'd have to rebuild it. You're getting a model. I'm going to write code now. So I'm rebuilding your model. Let's take that cost. And then let's take the cost associated with the human capital doing that instead of something else, because the platform was doing the work for you. And then let's add up the cost of the GPU factory, the business model associated with the language model company, and the tokens that will materially affect their business model. We've actually done the math on this. And so for a simple application on our platform, it would be 10 times greater in cost to try to replicate it with a language model.

Anecdote◆

Customer-centricity is the single determinant of business success — keep them coming back and you win.

McDermott distills his earliest business lesson from running a deli at 16: everything comes down to knowing your customer, serving them exactly how they want, and making them return. He illustrates with three distinct customer segments and a memorable anecdote about a kid choosing his store over 7-Eleven.

transcript

Bill McDermott: I think the biggest thing that I've learned in my life, and especially there, is it's all about the customer. In the end, the customer and the customer alone determines whether you win or lose. And it's a very simple equation. If you keep them coming back, you got a good chance. And if you don't, you lose. And, you know, back then, one of the most interesting parts about that store was knowing your customer. And knowing your base. And I really had three main customers. One was the blue collar worker, like my dad, as I said in the book, was rich on Friday and dead broke by Sunday morning. There were the senior citizens, and this was the early days of DoorDash, I guess, because they never wanted to leave their house, and we delivered, and no one else did. And the hard part was getting the kids to walk a block and 1/2 past 7-Eleven to come to my store. And these were the early days of video games and Asteroids and Pac-Man. And you tell that to people today, most of them never even heard of these games, but kids would stand there for hours, plucking their parents' quarters into those games. And that's how I got them to walk a block and 1/2 past 7-Eleven and come to my store. At the end of a long day, one of the young people said to me, when we want to have good food, be treated with dignity and respect, and play video games, we come to your store. And when we want to steal stuff, we go to 7-Eleven. So if you take care of your customer, you know who your customer is, and you give them what they want just the way they want it, no matter what business you're in, you've got a really good chance.

gives example · 1provides context · 1supports · 1

Claim

The pace of change today is accelerating irreversibly — it will never move this slow again — and leaders must find inspiration in that challenge rather than fear.

McDermott argues that while the current speed of change feels stressful, it is the new normal and will only accelerate. Leaders should reframe it as inspiring rather than frightening, because overcoming challenges leads to being superpowered. He grounds this in his own origin story of fighting for his first shot at Xerox.

transcript

Bill McDermott: And in this environment we're in today, which is changing, and it's dynamic like never before, and it's fast like never before, I tell them, it is fast, but it'll never move this slow again. So this is the way it's going to be. It's kind of stressful. Yeah, it is, and it is in one way, but it's inspiring in another. You know, we have to learn to get like real inspiration from the challenges that are in front of us, because if we're able to cope with them, and especially if we're able to overcome them, we could superpower.

Claim◆

Leaders must preserve human connection in the age of AI, because AI exists to serve people and amplify human ambition, not replace it.

Bill McDermott argues that despite AI's transformative power, the essence of leadership remains human connection. He warns that society is losing the human power to connect, and insists AI should serve people, not diminish human ambition.

transcript

Bill McDermott: I think somehow in the mixture of things with AI and today's society, we're losing a little bit of the human power that we all have to connect with each other and do magical things together and just make sure we remember AI is to serve people and to make the human ambition greater, not to take it away from us. And so that's my feeling on just how things have changed, but in many ways, the essence of leadership, human connection, people are just more important now than ever.

supports · 2

Claim

AI is meant to serve people and amplify human ambition, not replace human connection — and in the age of AI, the essence of leadership remains human connection.

McDermott contends that amid the AI revolution, organizations are losing sight of human connection. He argues that AI's purpose is to serve people and elevate human ambition, not to supplant it, and that human connection is more important now than ever. This is illustrated through his lifelong friendship with the Xerox manager who gave him his first break.

transcript

Bill McDermott: So I think somehow in the mixture of things with AI and today's society, we're losing a little bit of the human power that we all have to connect with each other and do magical things together and just make sure we remember AI is to serve people and to make the human ambition greater, not to take it away from us. And so that's my feeling on just how things have changed, but in many ways, the essence of leadership, human connection, people are just more important now than ever.

explains mechanism · 2extends · 1gives example · 1supports · 6

Mechanism◆

The key differentiator is that language models think while workflow platforms act — an LM can recommend steps but cannot close the case across multiple departments, data sources, and compliance functions.

McDermott illustrates with a compensation case example: a language model can instantly suggest resolution steps, but closing the case requires traversing data across sales, HR, finance, legal, and compliance — which is what a workflow platform like ServiceNow does. He frames this as 'AI thinks, but workflow acts.'

transcript

Bill McDermott: One such example would be, let's take an example that's real. Every company today has a compensation issue. So, okay, we have a compensation issue on aisle 7 in XYZ Corporation. The language model will do a great job of telling you in a compensation situation like this, please consider step one, two, and three. Excellent. Does it in milliseconds. Fantastic. Only problem is it doesn't actually close the case. So now in aisle 7, we got to have the data, the context, and we'll have to go in and out of several different iterations of departments, because let's assume it's a salesperson in aisle 7, and by the way, that's connected to the HR department. And because it might be a problem, probably finance is involved in it, or legal, or compliance, or risk. And now we have data and a workflow that's gone through many different iterations, databases, and functions to come back and remediate the issue and close the case. That's what a workflow platform like ServiceNow does. And it was also important to have AI think, but workflow acts.

explains mechanism · 3

Fact

Replacing an enterprise platform with a language model costs roughly 10 times more than maintaining the platform, because you must rebuild the workflow, retrain or replace the human capital the platform frees up, and pay GPU tokens.

McDermott presents a cost analysis arguing that replicating a simple ServiceNow application with a language model would be 10x more expensive when accounting for platform replacement, human capital costs, and GPU/token expenses.

transcript

explains mechanism · 1

Mechanism◆

Attempting to replicate an enterprise platform's capabilities with a language model costs 10x more, and businesses will never forgive software for making mistakes the way they forgive people.

McDermott lays out a detailed economic argument: replacing an enterprise platform with LLM-built code incurs the rebuild cost, the opportunity cost of diverted human capital, and GPU/token expenses — totaling roughly 10x. He adds that while people accept human error, software errors are unforgiven, making deterministic platforms indispensable.

transcript

Bill McDermott: The cost to replace an enterprise platform in this SAS apocalypse that people talk about is an extraordinary expense. Let's take that cost, and then let's take the cost associated with the human capital doing that instead of something else, because the platform is doing the work for you. And then let's add up the cost of the GPU factory, the business model associated with the language model company, and the tokens that will materially affect their business model. We've actually done the math on this. And so for a simple application on our platform, it would be 10 times greater in cost to try to replicate it with a language model. Now, the other thing about that is, if I give you the language model and it makes a mistake, you call me up and you say, hey, Bill, the language model made a mistake. And I say to you, the language model works. And you say, yeah, it works, but it made a mistake. Well, it's probably right, but it's not deterministic. And incidentally, it doesn't have the context of all the data that has materialized in the company and our relationship for a couple of decades now. And so what we're learning too is people that run businesses understand that people make mistakes. They never will forgive software for making a mistake. So we're kind of in a situation where at a very small fraction of an IT budget, a premier platform is a bargain.

Claim

The essence of leadership is human connection, and AI must serve to amplify human ambition, not replace it.

McDermott argues that in the age of AI, human connection and care for people are more important than ever, and that AI should serve people and amplify human ambition, not diminish it.

transcript

Bill McDermott: We have to put tremendous care into people, tremendous attention to detail when it comes to uniqueness in human beings and what they are capable of doing. And without him, I wouldn't have gotten my shot. And I never forgot that. And today, when I have CEO meetings, he still comes to them and he's my friend for life. So I think somehow in the mixture of things with AI and today's society, we're losing a little bit of the human power that we all have to connect with each other and do magical things together and just make sure we remember AI is to serve people and to make the human ambition greater, not to take it away from us. And so that's my feeling on just how things have changed, but in many ways, the essence of leadership, human connection, people are just more important now than ever.

Mechanism

Businesses will never forgive software for making a mistake, while they accept human error — so determinism and context are critical for enterprise platforms, making them a bargain relative to non-deterministic language models.

McDermott highlights a fundamental asymmetry: business leaders accept that people make mistakes but have zero tolerance for software errors. Since language models are non-deterministic and lack full enterprise context, premier deterministic platforms provide irreplaceable reliability at a small fraction of IT budget.

transcript

Bill McDermott: Now, the other thing about that is, if I give you the language model and it makes a mistake, you call me up and you say, hey, Bill, the language model made a mistake. And I say to you, the language model works. And you say, yeah, it works, but it made a mistake. Well, it's probably right, but it's not deterministic. And incidentally, it doesn't have the context of all the data that has materialized in the company and our relationship for a couple of decades now. And so what we're learning too is people that run businesses understand that people make mistakes, they never will forgive software for making a mistake. So we're kind of in a situation where at a very small fraction of an IT budget, a premier platform is a bargain.

explains mechanism · 1supports · 1

Claim

ServiceNow positions itself as the AI control tower for business reinvention that integrates with all hyperscalers, all language models, and all systems of record — providing a fabric that connects every node rather than competing with them.

McDermott articulates ServiceNow's platform strategy: rather than being displaced by hyperscalers or LLM companies, ServiceNow integrates with all of them as the unifying 'control tower' that connects every system of record, language model, and cloud, enabling a true agentic business.

transcript

Bill McDermott: Our perspective is these language model companies are fantastic, and they're very important, and they're going to do fabulously well. We said the same thing with the hyperscalers. It wasn't that long ago that people were like, hey, why wouldn't the hyperscalers just eat software? Because everything is going to go to a workload in these great hyperscaler companies. And they are great companies. They're fabulous companies. And so our perspective was we have to be the AI control tower for business reinvention that integrates with all the hyperscalers, with all the language models and all the systems of record, because they too are important companies, and the data that resides in them is very important. So if we can be that fabric that connects all those nodes, now we can help you run a true agentic business.

gives example · 1provides context · 1

Claim

ServiceNow positions itself as the AI control tower for business reinvention — integrating hyperscalers, language models, and systems of record into one fabric, and extending into security to address the $1 trillion/month cybercrime economy.

McDermott describes ServiceNow's strategy as becoming the connective fabric across all major technology layers — hyperscalers, LLMs, and legacy systems of record — to enable an agentic business. He frames the expansion into security (via Armis) as a logical extension, given that cybercrime is now the world's third-largest economy at $1 trillion per month.

transcript

Bill McDermott: And so our perspective was we have to be the AI control tower for business reinvention that integrates with all the hyperscalers, with all the language models and all the systems of record, because they too are important companies, and the data that resides in them is very important. So if we can be that fabric that connects all those nodes, now we can help you run a true agentic business. And on top of that, we made the move into security, not because we're going to do all things in security, but because US is the world's largest economy, China is #2, and cybercrime is #3. It's a $1 trillion a month problem. And so we feel that being able to integrate to all the good security platforms, but manage IT and OT, which is the operating technology with Armis now becoming a part of ServiceNow, we're going to give corporations the ability to have a full purview of the whole landscape.

Prediction

The enterprise software companies at risk from AI-generated code and agents are departmental point solutions serving a single low-value-add function, not platform companies spanning multiple departments or serving as hard-to-replicate systems of record.

When asked about segments at risk, McDermott identifies departmental point solutions — single-function companies with low value-add that are not high CEO priorities — as vulnerable. Multi-department platforms and serious systems of record are safe due to their data, context, and integration moats.

transcript

Bill McDermott: I think that it's true that there will be companies that will be at risk. It's also true that they theoretically should be more valuable because of the data and the context around that data. But if there was a risk, I believe it's more departmental companies. So if you think about companies that do something that span multiple departments or are a serious system of record that is very hard to replicate, they're safe. And they'll be a part of the future. If you're serving one function, and what you do is not tremendously high value add, and it might not even be a high up on the priority list of a CEO, I think that would be the chance where the vulnerability equation really rises quickly.

Prediction

AI will dramatically reduce net-new headcount in enterprises because AI agents handle the bulk of routine work, freeing humans for critical thinking and judgment — ServiceNow already resolves 90% of customer service cases with agents.

McDermott predicts that in five years, AI agents will absorb the bulk of routine operational work, sharply reducing the need for new hires in functions like finance and HR. He cites ServiceNow's own experience: 90% of customer service cases are now agent-resolved, shifting humans toward higher-value judgment work.

transcript

Bill McDermott: I think that in terms of employee base, I think you're going to see that the net new added headcount will be dramatically reduced. And that's because the company will be far more productive. The agents are real and they will take on a tremendous workload. So where to keep up with growth in a growth company like ServiceNow, you'd have to hire thousands of people in finance and HR and the supporting functions and services just to keep up with it all. And now the agents are going to be able to do a lot of that. So you're going to invest in things that really matter, like humans that engineer great innovations. ... For example, 90% of our customer service cases now are managed by agents. And that means only 10% are actually involving people. And so there's a lifting and a shifting and a changing of the guard in terms of what people do for the critical thinking and the judgment calls that they have to make instead of the tactical work of just grinding out details.

supports · 1

Claim

Businesses will never forgive software for making a mistake, even though they accept human error.

McDermott highlights a fundamental asymmetry: companies tolerate human mistakes but have zero tolerance for software errors, which makes deterministic enterprise platforms essential.

transcript

Bill McDermott: The other thing about that is, if I give you the language model and it makes a mistake, you call me up and you say, hey, Bill, the language model made a mistake. And I say to you, the language model works. And you say, yeah, it works, but it made a mistake. Well, it's probably right, but it's not deterministic. And incidentally, it doesn't have the context of all the data that has materialized in the company and our relationship for a couple of decades now. And so what we're learning too is people that run businesses understand that people make mistakes. They never will forgive software for making a mistake.

explains mechanism · 1

Mechanism

ServiceNow's strategy is to be the AI control tower for business reinvention, integrating with all hyperscalers, language models, and systems of record.

McDermott explains that ServiceNow's competitive moat is being the fabric that connects all nodes — hyperscalers, language models, and systems of record — enabling a true agentic business rather than competing with any of them.

transcript

extends · 1provides context · 1supports · 1

Prediction

Departmental point solutions serving a single function with low value-add are most at risk from AI-generated code and agents.

McDermott identifies that companies serving a single department with low value-add functions face the highest risk, while cross-departmental platforms and complex systems of record are safe and will remain part of the future.

transcript

explains mechanism · 1supports · 1

Prediction◆

AI agents will dramatically reduce net new hiring in growth companies, shifting human work from tactical tasks to critical thinking.

McDermott predicts that AI agents will take on a tremendous workload in finance, HR, and support functions, reducing the need to hire thousands of people to keep up with growth, while humans focus on engineering and judgment calls.

transcript

supports · 1

Claim

Leadership is the greatest profession in the world. The essence of leadership — human connection, people — is more important now than ever. AI is to serve people and to make the human ambition greater, not to take it away from us.

McDermott argues that in the age of AI, human connection and leadership are more essential than ever, not less. AI should serve to amplify human ambition, not replace it. He illustrates this through the story of Emerson Fullwood who broke company policy to hire him at Xerox, a gesture of faith in human potential that McDermott still carries forward.

transcript

provides context · 1

Mechanism

A language model thinks, but a workflow platform acts. The language model gives good advice in milliseconds but cannot actually close the case by navigating data and workflows across departments.

McDermott distinguishes between language models and enterprise platforms using a compensation-issue example: the LM offers good step-by-step advice instantly, but cannot resolve the case end-to-end. An enterprise workflow platform like ServiceNow handles the multi-department, multi-database process to actually close the loop. The insight that 'AI thinks, but workflow acts' was a major unlock for CEOs.

transcript

Bill McDermott: Every company today has a compensation issue... The language model will do a great job of telling you in a compensation situation like this, please consider step one, two, and three. Excellent. Does it in milliseconds. Fantastic. Only problem is it doesn't actually close the case. So now... we got to have the data, the context, and we'll have to go in and out of several different iterations of departments... That's what a workflow platform like ServiceNow does. And it was also important to have AI think, but workflow acts.

explains mechanism · 1supports · 1

Data◆

For a simple application on our platform, it would be 10 times greater in cost to try to replicate it with a language model.

McDermott argues that when you factor in the cost of replacing an enterprise platform's functionality — including rebuilding the application, the human capital diverted from other work, and the GPU/token costs — a simple ServiceNow application is roughly 10x cheaper to use as-is than to rebuild with a language model.

transcript

Bill McDermott: The cost to replace an enterprise platform in this SAS apocalypse that people talk about is an extraordinary expense... Let's take that cost, and then let's take the cost associated with the human capital doing that instead of something else, because the platform is doing the work for you. And then let's add up the cost of the GPU factory and the tokens that will materially affect their business model. And so for a simple application on our platform, it would be 10 times greater in cost to try to replicate it with a language model.

Claim◆

People that run businesses understand that people make mistakes. They never will forgive software for making a mistake.

McDermott argues that enterprise customers have zero tolerance for software errors (which they expect to be deterministic), whereas they accept human mistakes as normal. This makes reliable enterprise platforms essential despite the appeal of generative AI.

transcript

Bill McDermott: People that run businesses understand that people make mistakes. They never will forgive software for making a mistake. So we're kind of in a situation where at a very small fraction of an IT budget, a premier platform is a bargain.

Claim

ServiceNow aims to be the AI control tower for business reinvention — the fabric that connects all hyperscalers, all language models, and all systems of record.

McDermott describes ServiceNow's strategic position as a neutral 'AI control tower' that integrates with all hyperscalers, language models, and record systems. Just as people worried hyperscalers would eat software — and instead they became strong partners — ServiceNow provides the fabric connecting all nodes. This extends to security via the Armis acquisition, addressing the $1 trillion/month cybercrime problem, with IT/OT full-purview coverage.

transcript

Bill McDermott: Our perspective was we have to be the AI control tower for business reinvention that integrates with all the hyperscalers, with all the language models and all the systems of record... So if we can be that fabric that connects all those nodes, now we can help you run a true agentic business. And on top of that, we made the move into security... because cybercrime is #3. It's a $1 trillion a month problem.

provides context · 1

Prediction

At-risk companies in the AI era are departmental, single-function vendors with low value-add; platforms that span multiple departments or are hard-to-replicate systems of record are safe.

McDermott identifies the vulnerability profile for enterprise software companies: those serving a single department with low value-add are most at risk from AI. Conversely, cross-departmental platforms and serious systems of record are safe and will be part of the future. ServiceNow's horizontal and vertical coverage across the enterprise creates a 'pretty huge moat,' evidenced by 85+ billion workflows in flight.

transcript

Bill McDermott: It's true that there will be companies that will be at risk... If there was a risk, I believe it's more departmental companies. So if you think about companies that do something that span multiple departments or are a serious system of record that is very hard to replicate, they're safe. And they'll be a part of the future. If you're serving one function, and what you do is not tremendously high value add... I think that would be the chance where the vulnerability equation really rises quickly.

provides context · 1

Highlight slides

Enterprise Platforms vs. LM Replacement: 10x Cost Gap✦ from: Enterprise platforms are far more cost-effective than trying to replicate their functionality with language models.The Three Cost Layers of LM Replacement✦ from: Enterprise platforms are far more cost-effective than trying to replicate their functionality with language models.Customer is the sole determinant of winning or losing✦ from: Customer-centricity is the single determinant of business success — keep them coming back and you win.Three customer segments, one playbook✦ from: Customer-centricity is the single determinant of business success — keep them coming back and you win.Serve them exactly how they want it✦ from: Customer-centricity is the single determinant of business success — keep them coming back and you win.The Core Claim: AI Must Serve Human Connection✦ from: Leaders must preserve human connection in the age of AI, because AI exists to serve people and amplify human ambition, not replace it.The Warning: Losing Human Power✦ from: Leaders must preserve human connection in the age of AI, because AI exists to serve people and amplify human ambition, not replace it.AI Thinks, But Workflow Acts✦ from: The key differentiator is that language models think while workflow platforms act — an LM can recommend steps but cannot close the case across multiple departments, data sources, and compliance functions.The Compensation Case: Think vs. Act✦ from: The key differentiator is that language models think while workflow platforms act — an LM can recommend steps but cannot close the case across multiple departments, data sources, and compliance functions.The 10x Cost of Replacing Platforms with LLMs✦ from: Attempting to replicate an enterprise platform's capabilities with a language model costs 10x more, and businesses will never forgive software for making mistakes the way they forgive people.Software Errors Are Never Forgiven✦ from: Attempting to replicate an enterprise platform's capabilities with a language model costs 10x more, and businesses will never forgive software for making mistakes the way they forgive people.AI Agents Will Slash Net New Hiring in Growth Companies✦ from: AI agents will dramatically reduce net new hiring in growth companies, shifting human work from tactical tasks to critical thinking.Before vs. After AI Agents✦ from: AI agents will dramatically reduce net new hiring in growth companies, shifting human work from tactical tasks to critical thinking.Enterprise platform vs. LLM rebuild: 10x cost gap✦ from: For a simple application on our platform, it would be 10 times greater in cost to try to replicate it with a language model.Cost components of LLM-driven rebuild✦ from: For a simple application on our platform, it would be 10 times greater in cost to try to replicate it with a language model.Enterprise Zero Tolerance for Software Errors✦ from: People that run businesses understand that people make mistakes. They never will forgive software for making a mistake.The Premium Platform Bargain✦ from: People that run businesses understand that people make mistakes. They never will forgive software for making a mistake.

Related episodes

Video · a16z Podcast

Ben Horowitz on the Global Race for Tech, Power, and Influence

→

Video · a16z Podcast

Software in the Age of Agents | The a16z Show

→

Audio · All-In Podcast

SpaceX-Cursor Deal, SaaS Debt Bomb, New Apple CEO, SPLC Indictment, Colon Cancer Spike

→