OpenAI’s Friar: Strong AI Competition Coming from China,
We do internally feel that we have a path towards AGI. We had a big breakthrough in 2024 around reasoning models in particular. So you saw us move from a world of more chat models.
Chat bots touch beauty, fastest app in the world to 100 million users and today well over 300 million users. But last year we really started to see the progression into reasoning and what we call our oh series of models. So on preview was launched and now we're all the way to oh one. Hopefully you've all tried it. I'm one of those reasoning models. Do they really start to help you think about difficult problems? So you see the model. One of my favorite things to do is
actually watch what it's doing in real time. You see it go down, chains of thought, trying to answer the prompt that you've put in. You see, it sometimes hit a cul de sac. It doesn't quite get to the answer. And so it almost has to reverse course and try a different approach to solving that problem. A lot like the way we all do to ultimately give you the best answer. And frankly, I think often the answer is not a definitive singular answer.
It's an exploration. It's a way to continue a conversation. We are upping the pace. So we just went from 012, we previewed oh three to the second model. We like to confuse sometimes with our
nomenclature, but oh three, that just took three months. And if you think about the the length of time from chapter GPT three to chat 54, it took about two years. So the pace of innovation is really accelerating. What's it like to work in a place like that? It's exhilarating, it's inspiring, it's sometimes exhausting, and it's definitely done with a sense of privilege. I think all of us at Openai, I think a
lot about the magnitude of what we're bringing into the world. We think about it with regard to safety, with alignment, but also with the optimism of what it can do when you can put that sort of human intelligence into the hands of kids in schools, when you put it into the hands of teachers, when you put it into the hands of our colleagues at work, and we're just getting started. Absolutely. So you mentioned reasoning. Can you tell us a little bit about how this kind of really advanced reasoning that Openai is coming out with? How does that translate into real customer value for you? How are you seeing Opening Eyes users actually, you know, gain productivity, gain insights from that? Yeah. So reasoning is on the path to agents and we really feel that we're now moving into this era era of a gigantic technology and into a world of agents. So what we see today with the reasoning models is they are for some of the most difficult problems. We see it used in areas like pharma, for
example, deep research into new molecules that help us think about drugs and drug discovery. Those have been some of the areas that I have been most excited about the progress so far and how people are taking the O series and using it. But as I talked about with agents, I'm sure you're probably going to ask. Yes, unless you because everyone has a different sort of idea. But what's your take on what an agent is? And yeah, you know how Sam is that agents are going to going to be acting as employees, right. In 2025. They're actually.
And so where are you where do you think we'll actually see agents start to do the work of real people first? What sectors you think it's going to hit? So it's important with agents, first of all, to think about why is it even possible? And so that's why I made the point about with reasoning models that kind of cul de sac that are reasoning model goes down and then has to back up and try a different way to solve. Because in an old world of deterministic software, effectively the developer or the architect had to in some ways envision all the possible routes that a question or an output could take. Today with a reasoning model acts much more in the way that we humans think about problems. And so that starts to bring forth this idea of agents within our workplace, maybe even just as task workers alongside us every day. So some of the earlier things that we are looking at is how do we use agents to just solve day to day problems, like I'm a working mom, often I'm in that all the women in the audience just smile. You're in that moment. You're like, Oh, now it's 5:00.
I have still got multiple hours of work. I've got to get home. There's supposed to be something on the table to feed my children. Right now. My husband is really useful, but maybe
not all the time is the most useful. What am I going to do? And this is a moment when you think about an agent that can perhaps figure out What am I going to get delivered to the home? I don't want to have to be specific. It's like, Please get me something healthy. We had pizza last night, not Italian.
Let's not do too many carbs again. Won't know my budget, might know what local restaurants I typically have order from might come up with a new idea. So I would be an agent more in my personal life. But you are also starting to. See this, Don, of agents in our workforce. And so it may be areas like software development. It could be areas like research within
labs, deep science. We see many folks here at Davos heralding that age of agents in areas like customer relationship management, for example. So I think that you're going to see a lot of companies talking about new agents coming to the forefront of customer success is another one. But you've got to remember what that is
built on top of, and it's really built on top of these reasoning models that we open. I have brought into the world of the frontier technology and the Frontier Company, right? Let's talk a little bit about your bread and butter, which is the world of finance. Right. How do you top a record setting VC funding round, or what is the next what do you expect this year in terms of and obviously I cost a lot, right? The cost of computing increases your every year. So what can we expect? Will there be another make around 2025? Our technology is built on three things great people who are building the most frontier algorithms and models in the world. It's built on top of tremendous amount of compute, and I think we're just scratching the surface on that. And then, of course, data and in the
world of compute, in order to buy that compute to have access to it to control our own destiny, we have had to do an extraordinary amount of fundraising. Luckily, we also have a business model that supports that. I already talked about charging it for consumer over 300 million users today, and it's a workhorse when it comes to revenue, revenue growth and ultimately profitability. But we are seeing now enterprises of every size embrace this technology, and we really see ourselves also as the enterprise company. In fact, there's a really good symbiosis between those two areas because often when I meet customers and I'm meeting a lot of them here in Davos, it's actually their personal experience. They start with when I say, Hey, how are you using Chatbot? They'll actually give me a personal anecdote first.
And for those of you who sell into enterprises, you know that if you've won your customer heart and just their day to day being able to go in and then sell them into their enterprise environment gets an awful lot easier. And so that enterprise model is really building across every sector of the economy, every type of company, every scale of company. And I'd be happy to talk through examples. Morgan Stanley is a great example since we're talking financing and they're also a good partner in that front. But they're using our technology in areas like wealth management and areas like even their investment banking, and they've been doing it now for multiple years. So in terms of the future, from a financing perspective, I suspect we will continue to have to finance at pace, but we will do that on the merits of our business as well. How about an IPO?
What would that look like for open? You mentioned profitability. Know, how far do you think we are from that? You know, you're someone who's taken two companies public. What would the steps be for opening how to get there and how far are we from profitability? And so as I've told every company I've been associated with on an IPO, it's it's not a destination. An IPO is just a marker on a journey. And if you get wrapped around the idea that the IPO is the destination, it's a very kind of dangerous world to live in because there's a feeling of finality and management going public is just the beginning of another very interesting part of your journey. What are the positives, an IPO? Number one, I think it's a very strong credential, I think, moment for any company.
I love that moment of sunlight of being able to show your financials externally to show how your business is building. It's the best disinfectant in any room. It's a great way to fundraise because it opens the door not just to equity and selling your equity in that moment, but it also starts opening the door to many more areas of financing, starting with mezzanine debt, structured debt and so on. And again, in a world where we're buying a lot of compute, we need to get there fast because equity is an expensive way to raise capital and to deploy capital, we need to make sure we continue to bring down that weighted average cost of capital. And then the third thing that I like about an IPO, I say that somewhat tongue in cheek is it does add an element of rigor and discipline to the company's cadence, I think is a very good thing. The reason why I kind of caveated the
statement is you need to be very careful. It doesn't make you short term in how you think about your company. It's very easy to get wrapped around 90 day cycles.
But for folks who operate, companies do not work on 90 day cycles, particularly companies in our sector right now that are having to take a very long term perspective. Right. The the old series of models, the the work for that was that was really done probably two years previous when we kind of put shovels in the ground, built data centers, built up infrastructure to do it. And so this means that during an IPO process, it's incredibly important to bring along the right type of investor, the investor that understands the longevity that will be required to build the business probably a little bit more like biotech investors in some way. So it's always a potential station on the on the journey that we're on. But I don't want to make it the destination. All right. So we talked about funding, mentioned
the recent round, which was really the first major accomplishment right, under your tenure as CFO. Now, there's another big task at hand, which is the restructure of Openai from a nonprofit to a for profit and opening AI plans to make it a public benefit for profit corporation, but nonetheless a for profit. Why is this an important change for the company? Why do you need this? And particularly, why is it important to investors? Yeah, so first and foremost, as we think through the restructure, the most important thing for those of us at opening I focused on it is to also make sure that the nonprofit is front and center. Our mission is everything to create AGI that benefits humanity, that really benefits everyone, people in the world. And that is what the board remembered today. The board is the board of the nonprofit.
That is what they are most focused on in regards to shifting over to being a PBC. The reason that we are thinking through that is the rights corporate structure. It's not a done deal, but a PBC does allow us to balance both being mission focused as well as being shareholder focused, economically focused. It's not the be all and end all because we are proving that we are able to access capital, that we're able to continue to grow the business and we're able to attract great people. But those are the three tenants that I look at and thinking about why we might want to look more like a traditional company because it is fun to innovate, but it's better to stay innovating in the on the on the innovation edge that we want to be on, which is less about our corporate structure and much more about our models and the products that we're built on. Right.
And so, as you said, it's not a done deal. There are steps to go through. What are some of the challenges of navigating this restructure, especially as some people, including Elon Musk, have criticized the company for the move or the intended move? So for Elon and Sam, I've been quite public about this and others like Greg, our co-founder, that we see Elon as a competitor. We think he's a strong competitor, but we hope that he won't keep resorting to kind of using law and lawfare to compete.
I think even in his own words, AI and building AI is a very capital intensive business and I think even he recognized very early on that it would require us to be much more than a nonprofit. So I actually think he's even said that we don't we shouldn't be a nonprofit to be able to raise the capital required. So for right now, you're kind of said is a complicated internally.
We try not to distract. We want to make sure our researchers are able to do the job they want to do, to be curious, to be serendipitous. We want to make sure those that are taking that research and turning it into product are customer focused first and foremost, but also thinking about innovation, things customers may never have thought about. Make sure our go to market folks are free to do what they do. And so we really try to make sure our focus from a restructure perspective stays in the right context and at the right scale.
So there's the legal challenge from Elon Musk and regulatory approvals. But there's also just this changes the, you know, the structure for existing investors and, you know, kind of reassessing the equity in the company. So what what can that what does that going to look like with existing investors like Microsoft? And, you know, what kind of equity might we expect? I know Altman has said that some of the numbers out there are maybe not, not quite right, but what do you think may happen with equity at the company for him and others? Yeah. So if you look at the folks really involved from a restructure perspective, it's not that many parties at the table. As we said, the the nonprofit is actually the key focus for those of us at opening and making sure that the nonprofit remains well capitalized, well-funded, and can really help us live up to the very grand inspirational mission that we have for others like Microsoft, a really early partner for us, both from a capital perspective for helping us build infrastructure initially to be a partner even in how we think about some of our our go to market, that is more of a negotiation for the investors.
We came in last year, we actually raised that round with them sitting more atop. So for them it will be much more simple. It'll just be a conversion of the 6.6 billion they put in over the 157 billion post-money valuation.
That's the percentage of the company that they owned at that moment. It's easy math. It feels much more like a growth equity round. Well, I'm probably not going to get to the bottom of equity numbers today, so I'm going to move on. But I mean, you know, we put out there 7% was a number that was out there, I think Altman said. Yeah, that's that's not right.
Any kind of ballpark of what we might expect or nothing that we can discuss today. And I'm not trying to sugarcoat the question, but frankly, we don't have a definitive answer, so we're not really able to talk about it. Let's talk about global policy and infrastructure. So just on the policy front, last night, we all watched tech leaders, including Sam Altman, attend Trump's inauguration. How does the company plan on working
with this new administration? So we have released multiple blue papers now about infrastructure being destiny. And what we mean by that is we recognize that this new era era of AI is going to require a lot of both public and private sector relationship building and bringing it to to the world. That's because of the capital requirements, but also how we think about the right regulatory environment, How do we make sure we build this in a way where it benefits humanity? And so with that, our blue paper on infrastructure as destiny really leans into things like building on economic zones, recognizing a little bit with a US lens right now that it is about productivity, but also about national security.
But I stood in front of you as both a an American, a Brit and a European. So growing up in Northern Ireland turns out to have some really good advantages. And so from a US perspective, we see the productivity build already beginning to happen, frankly, and that's just good for people. Better standard of life, better quality of life, a better way to use their time.
But we also want to make sure that it's a Western alliance. Maybe one way to put it so that everyone is embracing it because we do have competitors that are going to come after this also from a national level. And I think we need to make sure also from a national security perspective, we're investing. So we've thought about things like
economic development zones, how to governments really reach out to their populace, help them think about reskilling and retooling. We're doing a lot of work in the education sector. We've struck some kind of incredible deals.
Aceh is a good example of a university in the US that has 280,000 seats deployed on it. My alma mater, Oxford, has also been a really good customer for us, for example, and we love getting down into that level because we see students embracing this technology. So that's been a second piece of the infrastructure is destiny. And then beyond that, how do we think about working in the right regulatory environment so that again, we deploy technology safely, but we also recognize that people want to be able to use it and that governments, for the benefit of their citizens, need to make sure they're on the forefront as well. So Openai recently put out an economic blueprint right of their the company's thoughts on what U.S. A.I.
policy should be. And a big message there was that the first sort of technological arms race and A.I. between the US and China and that the U.S. should invest and support the build out of infrastructure. Any any indication of if there is support for this idea in the Trump administration and why? Why are we hearing more rhetoric about the US versus China sort of AI competition now? I don't think it's rhetoric. I think it's the fact there is absolutely competition going on right now between us and China.
China is absolutely investing in this area. They absolutely know how critical it is to their economy, but also from a security perspective. So we should not be naive on that front. And we absolutely see already with the Trump administration a real willingness to lean in, I think to be very kind of on the economic front foot. And whether that comes from perhaps being more open from a regulatory standpoint, more open from just a business competition standpoint, we're excited to actually get to work. Let's talk about Europe maybe for a second, because you mentioned your European heritage and we are here in Europe. You know, it's no secret that sort of
policymakers in the EU are considered tougher on tech. We have seen deregulation come out of EU, right? Whereas in the US we don't have federal legislation really on the matter. Do you think that regulation can threaten the pace of innovation? And are we seeing maybe a changing appetite from some governments on that front? You know, I know the UK is not and is different, but the recent announcements right there about fast tracking air infrastructure, similar to some of the policy proposals that you're putting forward in the US. So what do you think about that balance
right in the EU of regulation being tough on tech versus innovation? Yeah, I mean I would kind of start if I'm sitting in the EU or the UK and think about productivity and what that means for the growth of my economy. Right. We've seen a real divergence between the United States and other economies around the world, and it really kind of hurts my soul in some ways to see the growth rates. The UK as particularly being quite a low growth rate, relatively speaking. And I think technology is certainly a major accelerant back to higher GDP growth, better quality of life for citizens.
We just did a survey that said what do people want when they think about AI? And it was really three things. Number one, they want a better quality of life generally. They want to make sure that they're getting jobs, that they're getting like a lift from the income that they take home every day. Second thing they want is better health care. And the third thing that they're looking for is just save me time, help me be more efficient, Help me just have that life where I have time to spend with my kids or have time to spend on the hobby that matters to me. And so we think that from a government
perspective, you really want to be mindful of finding that balance between overregulating, but then perhaps not allowing technology to bear the fruit that I think it's capable of. I think the US has found a really nice path on that. The blueprint that you're referencing has a whole conversation about the advent of cars and the society actually originally innovated in the UK, but at the time we used people walking around of cars using red flags to remind pedestrians there was this car coming. How could that be? Right? People don't walk that fast. It's because we also regulated the speed of the car, I think down to something like three miles an hour.
And so hence us took up that mantle, created an incredible automobile industry, and really took the baton at a moment when maybe the UK should have thrived on that innovation in particular. So it is but one example and one example is not everything. But I think it's a good metaphor for thinking about how do you balance leaning in to this new technology while doing it in a way that feels safe and and for the good of your citizens? I want to wrap up with a few more personal questions. So you are one of the top ranking female executives in tech right now. There is a lot of discussion about DEI in the tech field, about how, you know, how do women succeed in the tech industry? What's the best way to do that? How have you been able to navigate this and what are your thoughts on the topic? It's something very near and dear to my heart. I started my career as an engineer. In fact, my very first job out of
college or my first internship was working on a gold mine in Ghana, believe it or not. I wrote my masters on how to extract gold out of sulfide ores if anyone wants to have a conversation later. And I say that because I did that and I came back to the UK and thought, That is not a job I can be successful in because there were no other women.
And so I strongly grabbed hold of that statement about it's very hard to be what you can't see. And so I think in someone sitting and has the privilege that I have sitting in my seat, first of all, it's showing that women can be. And so hopefully that generation that's coming up behind can see what what they are capable of. I think the second thing is how do we lean into technology to go back to what we can do with AI? You really can start to create personal tutors. I just saw an amazing piece out of World Bank where they did a study in Nigeria for particularly they looked at the gender dynamic. So they created using Chhatrapati effectively an after school charter program.
And I think for girls, I think they zeroed in on the girls in particular. They saw almost a two year leap in education outcomes in just a few months of using the technology because it really does create a personalized moment. I think one of the things that hurts my soul a lot is when I hear particularly young girls say, I'm not good at math, which is just not true. They would never say, I'm not good at speaking German because you don't know yet. You are just starting your life. You do not know.
But it may not be. Talk to you in a way that resonates. Computer science is often taught first through gaming. My daughter, when she first took her first science classes, said, Mom, I just something like this. I want to build games.
Once she recognized that she could use it to create websites, she could pull her artistic side and she was fascinated, drawn to it. Today, I'm really proud of the fact that she is a hard scientist taking chemistry in university and doing computer science. So I think a lot of it comes down to how do we think about approaching people where they are and then in the workforce, How do I turn around at the table, reach behind myself and pull people up of all different types of diversity, not just women? Because I think in the end, the more diversity we have around the table, the better the products that we build, the better the outcomes we have for society writ large. And you cannot do that when you have a homogeneous point of view sitting in the room. This is a question that I got from
others, so I know it's on people's mind, but what is Sam Altman like as a boss? Let's see. Or where's the one when I knew it? Sam is a delight. He's super fun to work with because his brain runs at 1000 miles an hour. And so I kind of as someone who's really hyper curious, I love that because it keeps me on my toes. I he has a good heart. And I think that's really important.
And but he's always pushing. And in fact, I had a conversation with him recently. Talk about, you know, when you're having that personal moment where I said, Sam, hey, just something to know about me that when you're giving me feedback, if it's always on the thing, that could be a little bit better. That is good.
Bring it on. But sometimes it's really helpful to kind of also remind me what went really well. I will I will respond better to you. And and actually, he was like, that is really helpful because I'm someone who like tough feedback all the time, but I'm really going to use that and think about that. And I've seen him shift how we interact. And so again, it's like, this is why diversity is important, because we need to be able to talk to each other about different ways that we respond. And I hope actually that helps him think
about others that he's surrounding himself with. So yeah, overall it's been a wild and incredible and inspirational first kind of run at the company. It's an incredible group of people. You're literally working with the best in the world at what they do, and you really have to work with very different personalities. And I love it. And I would call that reinforcement learning. It is reinforcement learning.
2025-01-25 10:23
