Satya Nadella on Open vs Closed Models: Who Wins | Why Companies Need to Own Their Own Intelligence

0:00 And Dario, and Ilya, and Sam, and Greg, and Elon were all, I guess, one happy family. They were really the believers in scaling laws, and that's what I believed in back in 2018, 2019. I think the biggest change now I have is I used to always think that alignment would just fall out. I no longer believe that. Now, this show, honestly, is quite an emotional one for me. I started 20VC from my bedroom in London. I was 18 years old. I had no money. I had no contacts in my phone book. And today, I have the chance to sit down with Satya Nadella, CEO of Microsoft, one of the greatest companies of the last three or four decades. And I will just say on Satya, the warmth that he was able to imbue in this interview in such a short space of time was almost unlike anything I've seen before. The clarity of his arguments and his ideas, his ability to go very deep, very quickly. It was almost unlike anything I've ever seen before. It was a show I'll always remember, and I really hope you enjoy it.

0:58 Ready to go? >> >> Satya, thank you so much for doing this with us. I would say, you know, thank you for coming to the studio, but you're actually hosting us here. So, really appreciate you taking the time. >> Well, like, thank you so much for having me on the show, Harry. It's been a long time coming, but I'm finally glad we were able to do it. >> Now, you wrote a seminal piece 10 days ago or so, and you said that we should assume a model is compromised.

1:30 And I just wanted to expand on that. Does that mean that we can't or shouldn't rely on the disclosures of the model provider companies? >> No, I think the the challenge, the way I see it right now, is the models are becoming more and more capable, right? Scaling laws are working. Except, one of the assumptions I had was that you'll have alignment just fall out of scaling, and it doesn't seem to be. there's more science challenges there that need to be tackled, right? And mechanistic understanding is also elusive. And so given that, I think the best posture to have, if you want to continue to use these sophisticated models, is to treat these models in a trust architecture. I mean, you after all, zero trust is not something new, right?

2:12 We've had zero trust architectural approaches. So, the claim at least I'm making is let's create an engineered zero trust environment that can contain any malfunctioning agent before it does any harm, right? So, that's the point. Because if you really want to have agents that are more capable and yet less aligned, then you need to have a trust system that you can rely on. >> Can I ask, what does that trust system look like? Who's the architecture of it?

2:42 Who's the arbiter of it? >> It's a great point. And so, the that's why I kind of look at it and say, you leave AI safety outside of this, right? Because that alignment, for example, at least my take is that it's it's it's really mostly has to show up in training, right? In in fact, maybe even in pre-training. I don't think you can even RL your way to alignment. You have to truly get alignment to be something that truly gets embedded in the weights right at the very beginning of it. And so, that's the the model itself, hopefully at some point will get aligned.

3:14 So, if you leave AI safety, then the trust architecture is going to be just, you know, what a company like Microsoft does or all the cyber folks do. So, what we have done, for example, is we built these two things, right? Agent 365 and perception, right? What does it do? It gives every agent an identity. it then logs all of its activity, all its tool calls, in fact, every model call and the COT in it. In fact, that's where I'm also asking that all the COT better be in English and not in New Release.

3:42 and then you can check using multiple models. it better be tamper-proof, right? You can't have the agent go modify its own logs. and then on that, you set policy. And then you behaviorally monitor the activity of the agents so they conform to the policy. So, that's the architecture. >> We don't have that architecture today. Is that just symbolic of where we are in the progression? >> exists. So, there two I think people conflate what I will call an engineered security zero trust architectural system for containment and AI safety. Right?

4:18 Because I think people say, "Well, AI safety is hard and elusive and not yet achieved. So, therefore, there must be nothing we can do on security." And I'm not saying that that these two things are all got to come together at the same time. In fact, I would just say, "Hey, we have to deal with the situation as it exists. So, let's build a trust system outside in fact of the model and then just be super aggressive in setting policy and monitoring for policy deviation.

4:45 >> On the monitoring side, if we think about chain of thought, how reliable is that? >> It's not, right? We know that the chain of thought itself can be unfaithful and and and in fact it can be made worse when the model company says, "Well, it's unfaithful and I'm going to hide it." >> >> and and so it becomes even more problematic. And that's where I think a little bit of the push for standardization and the COT at least for trusted partners being available as is so that it's inspectable, it's auditable, and I can check using multiple models and what have you, right? So, that's going to be one of the In fact, that's one of the things that may even drive more open weights quite frankly because if you can't trust It's funny. Closed trade models are more powerful, but if they kind of don't recognize the need to be more transparent especially as their trusted partners use it, then we may have a problem.

5:41 >> Can I ask you, when we think about that rise of open source, I completely agree with you in terms of closed being more powerful, open source stays predominantly Chinese models. Is that a concern then when we think about safety being baked into pre-training, when we think about disclosures as we said there, and then juxtaposed with the majority of open source being Chinese? >> Yeah, I mean I think there two sets of things. I do believe, if you sort of step back, somehow AI safety not being there is not a good thing for China, it's not a good thing for the United States, it's not a good thing for the UK. So, I think that there's a part of everyone because this is not something that somehow an unaligned open weight Chinese model will probably attack even China. So, it's not like you can rest think that this will only be used as an attack vector. So, setting that aside, I'm pretty pleased with some of the progress even the open weight models in the Wester making and I'm you know obviously you know Jensen's working on it, the other folks are working on it. we will have even with the MAI models that will be not open weight. We'll talk I think of it more as licensed weights. Think of it that way. I want to turn this into just a simple straightforward software business as opposed to a network effects intelligence business. So, I think there will be different business models and I ultimately trust in your ability to use those models and control those models is what's going to determine what gets used.

7:05 >> As a venture investor, I have to predict where traffic will go today between closed, open. When you think about the two different dimensions which is like traffic and dollars, if we look forward say 24 months, what percent of traffic goes through open versus closed do you think? >> That's a good question. I I you know I think it depends on what the frontier looks like right which is I think most things seem to get saturated and when it gets saturated open weights or or even just cheaper models of even closed weights seems to work because especially one of the things is if you can train on your own traces then you really don't need the latest model to be able to solve that same problem if the problem has not changed right?

7:50 So, if you you know if you have an eval that you want to continuously keep moving forward, then you may want to pay. But even there by the way, I'm very thrilled about this Hydra fusion stuff we now have in GitHub where you know >> this. >> Yeah, because the cool thing there is it's a learn model, it know it's anyway we are now going into these agent swarms and multi multiple sub agents. So I can call a lot of sub agents that even happen to be local models. and so I think my suspicion is that there will always be a market for frontier.

8:23 there will always be some new thing. Science maybe the next thing that you know frontier is being set at. but I think predominantly token usage will be commodity token usage. >> Does that mean it's open source predominantly? >> It It maybe It maybe open weights or as I said it does It has It can also be cheap closed weights, right? Because there'll be lots of price competition. Let me put it that way, right? Because if you look at my own usage inside a co-pilot, we use a lot of closed source cheaper models along with open weight models.

8:55 >> You said about kind of the importance of owning your own intelligence for some of your amazing customers that you have. You're talking about some in the UK. Is that the model environment that we have moving forward? Specialized intelligence for companies with their own data. >> Yeah, so it's I I think this is to me one of the most important things, right? Which is what's the future of a firm in an era of intelligence, right? Because if you go back even to Hayek, his fundamental description of a economy is local intelligence. Tacit knowledge is distributed throughout the economy, right? Without that, you know, there can't be only one firm and it can't be one intelligence. And so that's sort of the foundational thing. In fact, capitalism works because of that. Price discovery works because of that.

9:38 or pricing mechanism works because of that. So therefore, I think it'll be important for every firm to be able to accumulate knowledge at increasing scales using AI versus leak. That's why I called it the reverse information paradox, right? Which is you know, the you know, Kenneth Arrow kind of described this paradox of information where if you're a seller and you disclose the secret without a patent or a trade secret protection or copyright protection, you leak the value. Now for the first time, the buyer needs the protection because the buyer, even as you know, most people think of it as oh, my data, it's even my exhaust of using it, right? Thumbs up, thumbs down signal, right? Who like you who you're creating value for the model company versus capturing that value for yourself. So, I think one of the things that at least I ascribe is that we need to have for this to be stable more of that surplus that gets created by use of intelligence to remain at the firm.

10:43 And by the way, it doesn't need to me only by embedding your knowledge in weights. It can be in context. It can be in your e-vals, right? You may have a benchmark. In fact, one of the things that I'm very gung-ho about is we all become every firm will have a private e-val benchmark, which is their core IP. and every model comes in and tries to sort of hill climb on that e-val. And that to me is how a firm works.

11:08 >> I I love this private e-vals benchmarks for each company. This is for a rarefied air of companies, so it would seem. Your Harveys or your Legoras or the kind of very forward-thinking AI companies. Will that disperse to every company in the world you mentioned to your >> happens. Yeah, it happens Harry Harry always, right? Which is, you know, there were there was a day when I remember in the late '80s and the early '90s that the few companies that had deployed PCs and were doing >> >> forecasting which felt magical. They were able to put in an Excel spreadsheet, send it in email, and avoid the fax machine and interoffice memos.

11:45 And it felt like oh, wow, when will this happen? And it took 10 years, maybe, and it spread out throughout. This time around, what is considered oh, I'm an AI native company and I do this. I think, you know, 5 years from now, it'll be pretty passé, I I would imagine. And then that's kind of what I mean that these tech cycles always start like that, right? Which is even cloud. Remember in the beginning of cloud, it was like the big deal was oh, who's cloud native and who's not? And you know, if you're not cloud native. And now everybody is, right? Even Lloyds, you know, which is what 500 years old is cloud native. And so it happens.

12:16 >> Do we move into an implementation economy then? Cuz when we think about these customers say as you said some of your very big customers they will need heavy FDE support much more so than they have done in traditional years when the last 10 years. Will we have a very heavy implementation economy to transition existing customers into >> Yeah, this it's it's interesting like the the implementation economy like systems integrators have always existed, right? You know, we've depended on implementation economy for lots of other technology diffusion. So this one I think has different contours, right?

12:49 Because it's the combination of tech and business if I could say. It's kind of but it's not classic management consulting is not classic systems integration and it's not classic sort of technology support. It's sort of a like for example, think about this, right? You want to start like when I look at my FDE engagements, right? Even take the Unilever example I talked about at the keynote today. It's fascinating. They've built an agentic system to say, "Hey, we're going to really have order to cash automated using an agentic system." Plus they trade claims. And by the way they start at the using a frontier model and then they substitute for in fact I think they're using currently an MAI model for all the trade claims because it's sort of a lower cost and it's by the way they control it, right? Because they're able to train on their traces.

13:35 And so we had some FDEs but a lot of who got deployed and now their IT team itself is you know managing it all. But it started with what's the business, what's the eval, what's the way to conceptualize the agentic architecture. So it's sort of business consulting combined with technical consulting and I think yeah, you're right in saying that there will be a new set of skills that will get built in the ecosystem. Today I think all of us have quote unquote our own FDEs but FDEs will become like you know, the traditional systems integration will become new FDE.

14:11 >> You've mentioned obviously MIA a lot. M A I >> Yeah. >> M I A >> Yeah. >> Gosh. But when we look at a lot of different players it's kind of this realization that you need to be full stack from chips to models to application. Is that the future? You have to have vertical ownership. And what does that mean? >> Yeah, you know it all depends on the industry structure, right? you know, I think that I would say today, you know, there would be who knows five, six open weight models closed models all competing maybe five, six open weight models all competing and so therefore at some level if you're building an application like the agentic system, you could even have something like what I said which is a router with even a decision engine built into it and then you just route to the best model for a given task. So that's a possibility.

15:00 of course there's always the risk that there may could be one vertically integrated model and app app tier and the the and the all the way to the silicon and system. It's unlikely to me, right? I I thought at some point that that might be the case that there will be one. It's if anything it's going the other way which is every day you wake up and you say wow there's more competition in every layer of the stack whether it's in silicon, whether it's in systems to manage the silicon, in the model layer, in the app layer.

15:32 And so I think that's a healthy sort of direction of travel but you never know, right? Because this is one of those things where it all depends on what becomes the dominant design. but my hope is that this becomes a little bit of a horizontal specialization with multiple winners in each layer and then the ability for people to compose. >> The thing I didn't see is the depth of each layer. Like what I know now that I wish I'd known 12 months ago is just like the infinite demand for inference.

15:58 >> Yeah, good point. >> And I wish I'd known it 12 months ago I would have done some very different investing. Speaking of that, we've had like Mustafa Suleyman saying training runs of a hundred billion are coming and only a handful of labs can pay for that. A hundred billion dollar training runs. Do we have the energy infrastructure and power to enable this scale across the stack, but especially when we >> Yeah, I mean the way I come at it is ultimately it has to come from the world GDP inflecting and then that allowing someone to then have as a percentage of R&D, right?

16:36 Because when somebody says training run, it's just a percentage of R&D, right? Sort of companies that are heavy R&D spend what? 15, 20%? you know, on R&D and so therefore yeah, at some point if you have enough revenue, you can spend enough on training. and then of course you'll have enough revenue if you're contributing to the world economy growing so that you can get your fair share, so to speak. And then you also bring up the other side, which is the supply side, which is we have power shortages, we have silicon shortages, we have, you know, all kinds of things today.

17:09 They will all they're long lead items, right? You just can't, you know, snap your finger and have power generation or you can't have a new fab. and so now, interestingly enough, given the demand, everyone's building. and so the question is when will all of this come online? What will be the demand supply? No one knows, right? Quite frankly, any one of us who are sitting here today will just be speculating. But my feeling is that if you start by thinking of the stable state would be just like today, right?

17:36 Income statements are suddenly not going to be different, right? For a company. It's going to be, "Hey, I have some revenue. I have some cost of revenue and I have to have some profit." and if that is going to be true, then the question is what is the world GDP growth rate? That's the bigger thing for me. >> In terms of the inflation of GDP growth rate, the world has kind of stabilized 2% a year for a very large amount of time.

18:01 Does AI have the ability to to seriously inflate that in a short amount of time? >> I think so. and I somewhat hope so. because that is the point, right? When I look at it let's take even as a percentage of any of our GDPs, the amount of money we spend in our health care, right? So that's why I look at even the deployment of Copilot in NHS and say, "Wow, if it can improve patient outcomes and reduce cost in the health care, that is by definition, you know, helping the GDP of the United Kingdom."

18:35 >> Mhm. >> And so, I think that there are these pockets of the economy which are ripe for adoption of AI and really In fact, there's the least amount of displacement, the most amount of benefit. and so, that I think is where I think we'll have to find the best fit. software development is another one, right? Which is It's fascinating to see how much the demand is elastic for software development, right? Which is that means the IT backlogs were huge and none of us even recognize it.

19:09 >> Actually, I was told that you had to be the top one or two in a market to win in venture. We're in, you know, Foundry, which is fifth, and it's winning. >> Fifth? >> Yeah. >> This breaks all rules of venture. The question that I have, again, slightly naughty nor on schedule, but it's just like, will we see the tipping in terms of capability advancement in alternative functions, finance, sales, marketing, that we've seen in coding? Because coding's been the mother load.

19:34 >> I think that that's a good point. It it You know, people talk a lot about, you know, do you have a verifiable reward and you know, it's easy to do these because a lot of the magic has been RL on top of these pre-trained models. That is what, you know, has helped coding. can we crack that in knowledge work seems much harder. but again, all the examples I talked about sort of, you know, in contact centers, in these supply chain systems, in these marketing {slash} or cash-to-order systems, I think it's it's possible and we're seeing it. is the demand as elastic as coding? I think it may be science may be as elastic, right? Or engineering in general. Like, by the way, you know, if I now look at even what's happening inside not just in software development, let's take silicon development. That's another thing that's going through a radical shift, right?

20:22 Your ability to tape out much faster, simulate everything in silico is much better. Your first spin can completely come out >> But even just like optimization of maintenance reports like you said with Rolls-Royce >> one which I would never It's unbelievable. Like I mean, it's a thing that they've been dealing with forever. Just schedule maintenance and I had not realized that wow, that's a big optimization problem that you can go now run. >> So, I've I've watched, yeah, many many of your shows and we're going to do a quick fire round.

20:48 >> Yep. >> and people haven't asked some of these questions which I've always wanted to know and I think everyone wants to know, but you know, people haven't asked them. What's the exercise routine given the travel schedule, Satya? >> Well, I mean >> You're in good shape. >> Well, I I try to be. I don't know whether it's all well hidden. >> >> but you know, the one thing I've gotten into the habit, you know, this has been maybe 25 years even on, which is every morning I get up and you know, I used to be able to run. Unfortunately, my knees have gone way and so now at least I walk with incline. even this morning I was able to get a half hour in and so that's what in spite of any jet lag, wherever I am, I get hit the gym every day.

21:28 >> Love it. I too have bad knees. My age as well. what have you changed your mind on in the last 12 months? >> Last 12 It might be because after all, I was one of the first obviously the when when Dario and Ilya and Sam and Greg and Elon were all I guess one happy family, they were really the believers in scaling laws and that's what I believed in back in 2018, 2019. and I think the biggest change now I have is I used to always think that, you know, alignment will just fall out. I no longer believe that. So, I think this alignment thing maybe a harder, different problem. and so yeah, that's the one thing I've changed my mind on.

22:10 >> What looks crazy today, but will be incredibly common and accepted in 5 years? >> You know, it might be that that thing that we just talked about, these Evals, right? Each What is a form? I think a form would be an AI benchmark. >> Final one for you. There are many things to be excited about in the next 10 years. For you specifically, like for me it's that my mother's got MS. I'm so excited for the drug development that could be capable. what are you most excited for?

22:37 >> I mean that's a great one, I think. One of the things that I've always been I would say been looking forward to is a general purpose technology even as powerful as this and people say, "Oh, this is not just another general purpose technology." Whatever this is. I hope this spreads, diffuses throughout the world faster than anything else and throughout all sectors of the economy unlike anything else and truly for once there is a real shift in the frontier of that global growth that happens where it's not about just counting winners and losers, but it's kind of like everyone gets lifted. Like that's been my dream. In fact, my father was a development economist and you know, it was always about convergence growth. When will the global south get something that just suddenly boosts that? I hope this is it.

23:33 >> Got you. This has been such a joy to do. Thank you so much for taking the time. >> Absolutely. It's a pleasure, Harry.