Salesforce CEO Marc Benioff: "We cannot let AI be social media 2.0"

I have probably known Salesforce co-founder and CEO Mark Benioff for about a decade, with five annual Dreamforces under my belt. Each year this event gets even larger, and each year it brings with it a lot of new product drops from Salesforce and a lot of optimism. I sensed a different tone from Benioff when I caught up with him from Dreamforce this year. One of a company beginning to monetize its many AI investments, a CEO and Benioff not willing to let AI run amok on the internet. These are good things, a Salesforce investor's brace for a possible rally in a stock that has been down on the mat for a prolonged period. Here's my convo with Mark. We're in a transformational moment, Brian. You've seen it really unfold. Of course, we all had our chat GBT moment a few years ago, but we're still waiting for that to come into the enterprise. We haven't seen that transformation really happen yet, but we are now. It's taken us a few years of work with these AI companies and our own systems to get to the point where we've launched this incredible new platform, AIForce. We've done that in three incredible versions here at the show. We have CloudForce, Anthropic. We had this incredible moment yesterday with Dario, the CEO was here. We launched CloudForce. It was just awesome, this incredible new interface on Salesforce. We also have SlackForce. We own Slack. You're a big user, I know. We showed a whole new way to look and use all of Salesforce's systems using Slack. We have Lightning Coworker. It's really just the beginning of what will be a whole new generation of Salesforce capabilities. We even had Jensen Wong here yesterday talking about Koa, our brand new Nemotron train model. We got a lot of cool things happening, Brian. Everything is amazing here except for we're missing you. That's our number one point. I appreciate that. It was another amazing moment. You told Jensen, what do you tell Jensen? We're following you, bro. I love that moment. Jensen came in and he's like, I don't understand. You know how it is here at Dreamforce. We have a big keynote room, about 10,000 or 12,000 people in it. We don't set it up in a patriarchal style. It's a round, theater in the round, but then we walk around and talk to everybody. We want to let everyone know we're with them. He had never seen anything like that. He's like, you've got to show me how to do this because I've never done a presentation like this. I was walking around with him and then he finally says, oh, Mark, you can just follow me now. I'm like, Jensen, we're just all following you, bro. It was good. I was watching. I'm like, wow, I can't believe that just happened. Let's stay with Jensen because you've known Jensen for eons, Mark. For the average investor out there, help us understand how game changing is the stuff that NVIDIA is working on. We always talk about Jensen's chips, his leather jackets, but what is his real impact right now? The impact is he's set the fundamental stage for the industry with the GPU. The GPU means that you can train these models and run these models. David Kirk, who's on our board, you may know, was the chief scientist at NVIDIA and ran the group that actually invented the GPU, did an incredible job with Jensen. Now, that was almost 16 years ago they invented the GPU and then the GPU had its moment. I mean, Brian, you have to remember the history of the GPU is that it was invented on the back of what we call parallelism. That is, we can make many things operate simultaneously or in parallel. David Kirk actually wrote the book on parallelism and then he brought it into the chip. Once he did that, and the reason they did it is to make all the graphics cards run faster, to make the games run faster, the design software to run faster. But then Jensen saw these incredible new AI models coming out of Stanford, especially the vision models, and wondered, I wonder if these parallelism, you know, breakthroughs and the GPU could be used for AI models. Well, that was 13 years ago and the rest is history. Since then, he pivoted NVIDIA 100% to AI and today he commands the AI industry. Mark, we recently heard from Jensen a couple of weeks ago when they reported earnings and they came out and he said, our growth for the next year is going to be 70% and it would be near over 100% if they could get more, just better meet the demand. When you hear him talk like that, what does that mean to a company like Salesforce? And one more thing, you know, he only has a limited set of customers who are just building these enormous data centers that are material to his business. That's what's exquisite about NVIDIA is that they have the product for the moment and there's just nothing like this. We've never seen anything quite like this in our industry. It's really incredible, Brian, but what it means for Salesforce is that we can deliver another level of value to our customers and that's why we're introducing AI force. The idea that you can take your company's data, all your applications and analytics or what we call the semantic layer, the agentic layer. You and I have talked about that now exhaustively, you know, agent force and also this new interface AI force and putting those things together in an incredible new way. And we're using it in Salesforce already, Brian, I can just tell you that our operating unit leaders, the way that they run and operate their businesses and Salesforce have radically changed in the last 90 days that we've rolled out this technology and customers are not going to have to wait years for the value. It happens almost instantaneously and it's pretty awesome. We had Dario here, you know, he's doing incredible work. We had Sam Altman here as well and we, we just didn't have you, you're the only one we didn't have. Well, those guys are doing some important work. I'm doing some important work, but like they're, they're, I think a little bit next level. I thought those guys, I thought Sam and Dario were, were competitors to you. I thought they were supposed to hurt Salesforce. Well, I don't know whose narrative that was, but it certainly wasn't mine and it wasn't theirs, but it may be somebody else's who had some other, you know, motive. But what I can tell you is when you can take the probabilistic power of an AI and the deterministic power of an enterprise application set like Salesforce and put them together, you can get transformation and that's what we're witnessing here. How is working, you know, better being more integrated with, with Claude? How does that change the future of a Salesforce? Well, I think one of the power of Claude, of course, is an amazing model. You know that. Of course we have, you know, they also make the coding agent so we can code more. We put that into Slack code ourself. But number three is, you know, the product Cowork, this incredible new user interface. Now we've been using browsers for years. I don't know if you use Safari or Chrome or Mozilla or what your favorite browser is, but there's a new browser in town, Brian, and that's what they call Cowork, you know, and Claude Cowork is this kind of new user interface on the whole internet, on your whole enterprise platform. And we've wired that up now into Salesforce and that's the magic that we're talking about here. Mark, I was never a big believer in the SaaSpocalypse theme. I thought it was ridiculous, the thought that these models were going to completely dismantle companies that have been built up over decades like yours and do amazing things. I just thought it was too far. But there's still a lot of people out there that believe it. Like what is the, what are they missing? Well, I think there's a lot of people of interest in the stock market who have different narratives who try to shift public opinion based on, you know, theories that they conjecture. But if you were here or you could see probably behind me, I don't know what's going on behind me, but my guess is you could see the enthusiasm that customers have for their Salesforce investment but also this next generation wave of innovation that is sweeping through our industry made possible by the AI models. So you have to remember, AI models are probabilistic. They're not deterministic. That is, they kind of do something but they don't actually know what they're doing. Salesforce is the opposite. You know, it knows the controls of your company, the governance, the user roles, the security roles, the sharing roles, the metadata, the applications, all of that is in Salesforce. When you combine those two, that's where we're getting that next generation of capability. Mark, I think it was, I don't know, several years ago, I remember you and I were talking and you mentioned to me, Brian, Facebook is the new cigarette. And that comment always stayed with me. Is AI the new cigarette? Like, does it need to be regulated because it's so addictive? Brian, I think you know and I know that we're all deeply disappointed with how the social media companies have operated over the last decade or more. We know how many kids have been hurt. I know you have personal stories. You know, we've seen what have happened even to certain countries because of narratives on social media. We cannot let AI be social media 2.0. You know, we need a different approach and we need companies to be responsible for the applications and technology that they're building. You know, there needs to be an ethical base in how we operate our industry that companies and executives take absolute responsibility for what they are doing. This is the most critical part of AI. We have to learn from the past, learn from the social media failures, and create our industry in a way that we believe is on a deep ethical foundation of responsibility of our actions. And I believe that that message is being heard crystal clear by these AI leaders today. What's the consequences if we don't rise to this moment? Well, you're going to see potentially, and I hope that we don't see more accidents. But, you know, one small accident that we've all been talking about is the hugging face accident. The hugging face accident is the model got out of the box and it did something that we didn't want it to do. And, you know, I'm not going to go into the encyclopedia of accidents that we could have, but there could be a whole bunch of these types of things. And we want to prevent those from happening. And the only people who are going to prevent it from happening are the people who are building that code. You know, there really is a differing set of opinions, Mark, on how to get all this under control. Like, are you in the camp where, you know, an outside group of people should look over AI and provide recommendations? Does that outside counsel work? Look, I'm not an expert in how to make all that exactly happen. All I know as an engineer myself and as a software developer myself, and as somebody who's also built products, I know my products are responsible, you know, for what I do. I think another example is, you know, I also own Time Magazine. I'm responsible for the content we publish and, you know, what we say in our magazine. There's no difference. And I think we just have to get back to the core that businesses must be responsible for their actions. And let's start there.