Hi everyone, thank you for tuning in. I'm Shiv Verma. I'm the CFO of Robin Hood. Very excited conversation today. We are live here in the headquarters of OpenAI in San Francisco to have a great conversation. We are hosted by Sarah Friar, the chief financial officer of OpenAI today. Sarah, thank you so much for having us.
Welcome. No, Shiv and team, we're so happy to have you. You're a great customer, but also now a great partner. So, thanks for being here.
Exactly. So, the reason we're here is exactly that. We are now an investor in OpenAI through our fund, Robin Hood Ventures Fund 1. Super excited to allow customers to have access to it. When we take a step back, our mission at Robin Hood is to democratize finance for all. And the vision of the fund was to invest in frontier technology companies and I can't think of a better company than OpenAI to be in the portfolio. So, we're going to learn more about it. Just want to have our customers and investors hear from you directly and let's dive in.
Sounds great. Let's go. So everybody knows OpenAI since the viral ChatGPT moment in 2022, but just to level set, could you tell us a little about the scale of OpenAI today and what you're really excited about?
Sure. So you're right, viral chat moment at the end of '22 and I think OpenAI today is still at its core a research and deployment house. Research is the core of all that we do. We've had many breakthroughs since then, to reasoning at the end of '24 that really brought us to this era of agent technology. Codex as our coding agent is kind of the first and really the biggest takeoff that we've seen. But of course we now see this world where everyone's going to have a personal assistant at their side. Today we serve over 900 million consumers on ChatGPT. But on the enterprise side, which is getting close, it's not quite 50% of our business today, but it's getting close. We're at about 60/40. We are starting to see enterprises from the smallest in the world, the entrepreneur that's just getting going all the way up to literally the largest enterprises in the world. Yesterday I was with a group in the banking sector. So folks like BNY, the oldest bank here in the United States, BBVA, a huge bank that services Europe and LATAM, they have I think 120,000 people utilizing our products inside BBVA. And then folks like American Express and Travelers. That's just one sector. I could go sector by sector, but really the scale of the opportunity just continues to grow.
So the consumer side, many people are familiar with the consumer side, but I think one thing that's more recent is the ad business. Would love to learn more about that, why you think that's great for customers, why it's great for advertisers, what's the vision there, how big could it be?
Yeah. So as you say advertising I also hear access and I kind of loved how you started by saying what Robin Hood stands for which is this idea of democratizing access. And in our mission statement OpenAI is AGI for the benefit of people everywhere and that means access. So today of those 900 million consumers 95% of them get it for free. Now to do that and to keep driving a great experience, we're looking at ads as a way to monetize, but it's really to create more access. So, one of the things I see in our daily activities, a free daily active user uses about seven turns per day is what we say internally, but it means kind of seven questions.
If you look at our next tier up, once you start paying a subscription, so our gold tier, they go to about 15 turns per day. Our Plus which is our next up subscription gets to about 3x and then Pro where you see a lot of devs for example about 11x. But what is that telling me as a CFO? It says as people understand what intelligence can do for them they use it a lot more. And it kind of reminds me of back in the day when I first got my cell phone it was a flip phone, you know, I was like, well it makes calls and that was quite useful but as I started to get a smartphone and could see that this thing was not just making calls I could text on it. I could communicate in different ways. But now I can order food. I can order a car. I can buy a new pair of shoes. Like all the things I love to do. You start to see the value. And so going back to ads, ads is a way to help us monetize in a way to create more cash flow that we invest back into the business and give more access to those users that want to use it for free.
Now, on the other side, of course, advertising will only work if you have great advertisers who want to be on your platform. And what we see within ChatGPT is there are many things that people are doing where there's a very natural motion to bring an advertiser in like help me find a great weekend getaway. That is an amazing time to go work with a Booking.com or an Expedia for example. Help me find a new pair of shoes. That is a great moment. What are they for? Oh, it looks like you're trying to get healthy. Looks like you might be trying to run. Here's some ideas of running shoes that might work specifically for you. And I think what we're going to see in that advertising motion is way more conversational advertising, which we've never had before because you actually could never really do it at mass scale. And so I'm pretty excited about both sides of the network. And to your point, we're still very very early. I think our ad platform's been alive for about eight weeks.
But as of actually last Friday, we just went from the US into Canada, New Zealand, and Australia. So we're starting our move around the world as well.
Oh, that's very cool. And we're happy to be a pilot partner in it for our advertising team. What gets us excited is it's high intent. So customers are trying to solve a problem and then you can surface the solution to that problem. There's no better place that we'd like to try to tie that. So super excited to see how it goes out.
And maybe just one more thing to say on it. There's still a north star that there has to be trust. The user has to know that the model is returning the best possible answer, not the sponsored answer. We don't want to go down that track. We're not selling your data to advertisers, but rather we're using the intent of the conversation to bring an advertiser like Robin Hood to you in a moment where you're asking about the market. Like, how could I begin to invest? You know, I have late teenagers now, early 20s, and one of the things I'm doing is a lot of financial literacy. How do I teach them to think about the market? And a Robin Hood account is a great way to do that and a great search that they might do on ChatGPT and find their way to you.
100%. So that's the consumer side. If we go to the enterprise side, you know, you mentioned it could be as big as the consumer business. Would love to know a little about that. What are you most excited about? You know, you mentioned Frontier. There's also Codex, your coding tool. Before I left, I had it doing a couple things for me. Would love to learn a little bit about the size and scale of the enterprise business. Codex. Anything else you want to share there?
Sure. So, let's dive in. There's a lot in there. So, first and foremost, I'm actually going to go to Codex first because I think we're all so excited about it internally right now. So, Codex just hit four million users. Just to get a little sense of how fast it's growing, it went from 3 million to 4 million in about 14-15 days. It went from 2 million to 3 million in about a month. So, actually the speed is, you know, it's going exponential. That's always a good sign that customers are loving what they're getting. Codex was this beginning of showing what agentic behavior could be like. So what does it mean as a developer when you can natural language say to an agent, help me build this website. You know, create the product requirements document for me, the PRD, help me build it. So the agent goes off and builds it, but now help me quality test it, like bug bash it. Help me write documentation, usually the thing developers really don't love. I'm right there with them. And so it's doing a lot of those tasks for you. And so what I see is more and more our more senior developers are actually orchestrating multiple agents at a time to do work for them. Even as a CFO I'm orchestrating multiple now agents to go off and do work for me alongside my amazing finance team. So Codex growth is just skyrocketing. We just launched Codex Labs. That's really about unlocking more deployment into the world. But that's only one piece of our enterprise business. From there, there's an underlying API business. So, what's an API? It's just a way for a company to get access to our models, to our technology. So, usually their tech teams will effectively back into our API to bring that into their world.
And then, of course, there's a lot of customers who need our help kind of trying to figure the whole thing out. And so Frontier is the platform that we're deploying that allows an enterprise to go from maybe a more simple ChatGPT deployment. So that might give you lots of productivity moments into becoming truly an AI native company. So think of Frontier as the whole kind of shebang in the end. It is the big sale and that's another place where we're going to market with a lot of partners.
No, that's great. I think what we love internally is originally it was developers who adopted it. Now it's the entire company. And so as you go out further and further, I think these tools are not just for developers, but for anybody who wants to use technology, which is phenomenal.
Yeah. I mean, let's talk finance for a moment. We are two CFOs. I promise it's about to get really interesting. But as a CFO, like some of my favorite use cases, but I'd love to hear what you're doing at Robin Hood. We just raised a fairly large funding round. And I'm excited to create access through Robin Hood as well to the general public. One of my favorite moments during that fundraising round is my team knows me well. I often joke I'm not a micro manager, but I am micro-interested. And so my treasurer kind of just showed up with a dashboard. And I was like, 'Wow, I'm really impressed.' Like I didn't know you could build dashboards, too. Felt like a whole like he'd created this whole treasury app. And he's like, 'Well, I know that as we start to see funds flowing into our bank accounts, you're going to keep asking me how much has arrived, what hasn't arrived, you need to make some calls, all of the things.' And I was like, 'That's amazing.' And so, he had gone off literally while he was getting the tires changed on his car. I kid you not.
He'd used our Codex tool and our internal dashboarding tool.
And just prompted it through word of mouth to create a dashboard that would show funds in, funds not yet arrived, by account, by amount, the whole thing. And I just find moments like that, you're really unlocking your team to do some great things.
Yeah. So, what have you seen recently?
Oh, we're seeing, so first we are fully AI-led. And we are using it across the firm in finance. The thing I love is it's organic. So we went to every team and we said our north star is just adoption. We just want you to try it. And once you see how powerful it is, people start to use it more. And then what gets really exciting is they start to champion it to other people. Hey, look at this cool thing that I did. Look at that. And so when I look at the finance team, our FP&A, we have a lot of reports and dashboards and there's constantly things that you're monitoring. All of that's automated now. And it gets you in real time and it's much more efficient. Treasury team, they love it. They're looking at risk. They're looking at analytics. We have tens of billions of dollars of funds that flow through our platform every single day. And so they're using that to monitor. I was talking to our corporate controller. They're using it on the technical accounting side to give the first draft of memos. He's like, you know, he's like, I know enough to be dangerous, but I'm also going through it and refreshing. Our corp dev team is using it to go through massive amounts of data rooms and having it summarized. We have an internal team called benchmarking or competitive intelligence. Similar to Amazon, they created essentially their own website that has a repository of every single report we've done going back a decade that you can now go in and search in natural language and find different things. And so the use cases are just coming up organically day after day. So yeah, it's been really fun to see what people are doing.
It is super fun. We do a hackathon with the finance team every quarter and you know, you just see people getting unlocked but getting a little kind of geeked out as well. Thank you. Thanks for doing that.
Thank you for the great technology. So another product that we saw yesterday you launched was the images new product. Super excited for that. Would love to learn a little bit more about that and yeah when's that coming up?
Yeah so Image Gen 2 is our new model for images. And so for everyone who has access to ChatGPT on the consumer app. Please just go wild at the moment. If you click into ChatGPT you'll see a little icon that says images. If you click on that, what we're really trying to do is help people prompt in a way that they get something even better than they could imagine. I mean, of course, you can do a very simple prompt yourself that just says, you know, take this image and turn it into a birthday card. I do that all the time. Because usually I've forgotten to buy the real thing, so people get a lot of text birthday cards from me. But we actually have a lot of example templates in there. There was one today I was using as an infographic and I was kind of blown away by it. I gave it one simple prompt and I very quickly had something that looked like I'd worked on it for maybe days, weeks. And so that is an example of, you know, you started off, it can have a very consumer look and feel, right? Consumers have come to love images and videos and so on. We see that across social media, but then very quickly you see all these applications in areas like education, right? We find teachers tend to love image gen because when they're creating class prep and you know if they're going to teach about the rain cycle there's nothing better and you know ChatGPT you can actually go into study mode and say I'm teaching an elementary school class or I'm teaching a PhD so it will kind of graduate where you're at but then most of that is way more engaging particularly at the younger kid level if you can turn it back into images. So, it's just one example that I really love. But behind the scenes, what are we doing? We are continuing to make ChatGPT better and better. You'll start to see it become more and more of a personal assistant for you. And then it comes back to our business model. In order to do that, we have to keep monetizing the platform. That allows us to create more cash flow. That allows us to invest back into the business. And I'm sure we'll talk about compute in a minute because that's the big daddy of all the investments that we do.
