은행 및 금융 서비스에서 OpenAI의 AI 통합

### 금융에서 AI의 역할: OpenAI의 CFO 사라 프리어가 전하는 인사이트
* 💼 Money 2020에서 OpenAI의 최고재무책임자(CFO)인 사라 프리어는 **모건 스탠리와 같은 사례**를 통해 금융 기관이 **AI 모델을 통해 자산 관리 생산성을 향상시키는 방법**을 설명했습니다.
* 📊 금융 서비스는 **OpenAI를 가장 많이 도입한 업종** 중 하나로, 수익과 효율성 증대를 위해 기술에 조기 도입하고 투자하려는 의지가 높다는 것을 보여줍니다.
* 💰 AI 도구의 가격 전략은 진화하고 있으며, 기업 사용자는 **1인당 약 2,000달러**를 지불할 가능성이 있지만, 많은 사용자가 **한 달에 60달러 정도**만 지불하는 것으로 보아 상당한 가치를 인식하고 있습니다.
* 🚀 OpenAI는 고급 AI 모델 개발 비용과 경쟁력 유지의 필요성 사이의 균형을 맞추는 것을 목표로 하며, **성장을 위한 현금 흐름과 투자자 지원 모두의 중요성**을 강조합니다.
* 🔍 Facebook은 **광고 기반 수익과 같은 대체 비즈니스 모델**을 모색하는 동시에 전 세계 사용자들이 **AI 도구에 액세스**할 수 있도록 하는 것을 최우선 과제로 삼고 있습니다.
* 🛠 OpenAI와 Microsoft의 파트너십은 필요한 컴퓨팅 리소스를 제공하는 데 중요한 역할을 해왔으며, **소비자 혜택을 강화하기 위해 이 관계를 확장하는 것**에 대한 논의가 계속되고 있습니다.
* 📈 앞으로 OpenAI는 **새로운 AI 모델을 지속적으로 혁신**하고 개발하면서 제품 속도를 유지하면서 비용의 균형을 맞추는 데 주력할 계획입니다.
* 🔒 다가오는 선거로 인한 잠재적 영향에 대비하면서 방대한 사용자층에게 **안전한 정보를 제공**하는 동시에 정보에 입각한 의사결정을 장려하는 것이 중요하다고 강조합니다.

금융 기관은 생산성을 높이고 고객 성과를 개선하기 위해 AI 기술을 빠르게 도입하고 있습니다.

So much of the focus to date on open air has been, I guess, a personal use of chat GPT. But let's start with the banks and finance. How much of that is made up in your business? Yeah, thank you. And it's so great to be here in your right Money 2020. I've seen it progressed through the years. What an incredible event it is today. And we're here because our customers are here. Right? AI is happening right now. It's not experimental. It's not something that people are just playing around with banks, financial institutions, fintechs. People are using it today in their business. There's the Morgan Stanley case study. Yeah. You know, this week alone or past week, Bank of America has talked about how many patents it's got in machine learning and artificial intelligence. But do you actually have a tangible sense of what it is those financial institutions are doing with your large

language models? Yes, absolutely. So, Morgan Stanley is a great example. And their wealth management business, they're using our models both to help wealth advisors be more productive. They're using it as a way to create better financial advice and outcomes for customers. We're seeing folks like Klarna use it in customer service. The CEO of Korn has been very loud and proud on this front. That's another great case study in terms of productivity improvements. And then we have banks like BBVA that are using it all across their business, and that's really our message here. It's just get started, get Enterprise rolled out to organization, see what your people do with it, help them just get started. Whether they're marketing and finance and product, everyone can have really interesting, even if we all just getting started, can you help our audience understand how meaningful contribution the financial services sector fintech makes to open air revenue? Yeah, so today if you look at our

largest verticals, areas like education, health care, but financials is probably third. And again, I think that's because they are typically early adopters. They're often willing to take that risk because they see the upside in terms of driving more revenue. But they also are very good at managing their cost and efficiency. And it's great when you get an early adopter like Morgan Stanley because it tends to lead the way. I can think of a wealth management client today that's coming to us to say, What do we need to do? How do we get started? Banks in particular can be sensitive to pricing. There's a lot of curiosity, not just in sort of as a consumer, how much I'm paying on a monthly basis for chat access. But at the enterprise level as well, reports are, for example, at the corporate per user level, you're thinking about $2,000 per head. How are you managing us?

Give us insight into the pricing strategy there. So pricing is super interesting because we're really trying to think about value and what is this person getting from this tool. And I don't think I actually don't think we've done a great job of that yet ourselves. Despite that, we have 250 million weekly active users. And of that a, you know, five, 6% have actually converted to be plus customers. So they're paying. But if you look at the value when we were rolling out oh one, our reasoning model and this is a model that stops and thinks for you, it actually does hard problems. I was watching a lawyer in action using it to create a brief and at the end we said to him, What would you have paid for that? Like if you had a paralegal doing that? He was like easily 1000, $2,000 per hour. And this is someone who's using it. If it's an enterprise, proceed probably

60 bucks per month. And this is someone who would have paid $1,000 to $2,000 per hour. So I think that there's a lot of value that is in the product today. But we are just trying to make sure people can get started, can actually see the outcomes. And over time, we believe that value to price will come into alignment. There's a balance, right, between what's of value to your customer, but also, you know, you have to account for open eyes spending towards that endlessly, particularly on the compute side. What is that balance in what works best for you and your customer base? Yeah, so it, our first and foremost, what's most important to us is to stay on the frontier, building the frontier models, making sure that we are bringing ultimately AGI to the benefit of humanity. To do that, it's expensive. We have to invest in large-scale compute. And so to me, there's really multi two ways to finance that.

It's either through the free cash flows of the business, spoken like a good CFO, or it's through raising equity and debt financing because investors can see the long-term potential of this business. So, on the former, on the cash flows of the business, we want to make sure we keep growing that business. I think we have been wowed at just the pace of growth, particularly on the consumer side. It's about 75% of our business today. But even our enterprise businesses, they are young, but they are already doing an incredible amount of annualized revenue, and we're really excited by the potential there for our Bloomberg television and radio audience worldwide. We're in Las Vegas at Money 2020, and we're speaking to the Open Air CFO, Sarah Friar. And you talk there about the consumer business. Something very interesting is the future business model adds. Imported tiers, for example, very specifically segmented pricing. How do you think about that, Sarah?

Yeah, so I think you always want to stay open to alternative business models or ways that you can layer a new business model on top. Now, the key for us is always access. How do we make sure as many people as possible get access to this tool, and that's on a global stage, by the way? And so, to do that, sometimes you do pivot away from just pure subscription models to models like ads. My last company was all ads, so I've definitely experienced this. I think ads have their place, but you have to be really mindful of where. I think in areas like e-commerce, ads are great, right? If I do a chatgpt prompt for black high-heeled shoes, something I probably would do, I actually don't want a history of the black high-heeled shoe. I want five stores I could buy from right now, probably e-commerce. So, that's why companies like Shopify are great customers of ours as well.

But there are other places where actually the ad model doesn't make as much sense. You want to get the consumer to the answer they need as fast as possible. And I think that's where chat you get is a really very different platform from say, something like Google Search. You are still relatively early in this role, but it's been two years since the original deal with Microsoft was negotiated and that compute relationship is critically important. Yes. Will those terms change? Are they fluid or are you just sticking to the contract that was on your desk when you arrived? Now, and it's actually longer than two years. Microsoft and Open Air have been partners, I think, for actually almost five years. So they really came to us when we were a research lab. And the deal that we've worked with them is they do provide compute exclusively and we give them the IP around artificial intelligence. So it's incredible. The products they're rolling out today are really built on top of opening as AI.

As we go forward and as we get bigger, we absolutely see a maturing in that relationship. And so for consumer benefit, we want to make sure consumers always get access to what they need. That will probably mean compute for more parties over time. And we did discuss the Oracle deal or Oracle discussed it a few quarters back, and I think that's a good starting point for just how do we maximize compute so we maximize the impact for consumers. The catchphrase or buzzword of this year I think has been ship products, but you're the CFO. And so there's a tension between the cost of doing so in the need to move quickly. I think Sam has denied recently reports that the next model will be out by year-end. What can you tell us about the path forward, the cadence of new models to come? Yeah, I mean, I think you hit the nail on the head, ship product, product velocity.

That is the mantra internally to opening. I and it's something I've just been so wowed by since I started is just how fast we do ship products, right? Even in my short tenure, I've seen oh one come on our reasoning model, that huge step forward from kind of what has been more the Taxi Beat model series. We've launched things like advanced speech. As to the model itself, the O series for oh, and now for Mini, for example, for a mini, which is our distilled model is 100th the costs of what the original Statue 54 model was. That is incredible for developers and that's why you see the API products be so successful. And again, it goes back to how do we get this into the hands of many developers are a force multiplier. And today I'm super proud. I think every single A.I. unicorn is built on top of opening ISE API. And I will tell you how we are the

Frontier model. When I started covering open AI, there were around 500 people at the company. It's probably nearer 2000 now. Yeah. Part of that is the CFO's role to keep talent, long-serving talent, happy. My understanding is that tenders will be a big part of that, giving employees liquidity. What will be the cadence and sort of increments of that going forward? Sara. Yeah. So we are a company that has done tenders to date, and part of that is because we are in a competitive market, particularly for research talent. When I think about what keeps you on that front edge of the best models out there, it is compute. But more importantly, it's people, it's great researchers. And so, in order to compete with companies that have liquidity already in their stock, public companies, we have taken this approach to tenders a little bit like others in the space. Space is a great example, one we've covered closely.

Yes. And so we want to be able to keep doing that. We want to do it thoughtfully and mindfully, knowing that the other rule one is to keep it on the field. Yeah, make sure we have money for compute. So it's always a balance. But we do think it's important to give our researchers that access probably in the television and radio audience all around the globe. We're in Las Vegas and we're speaking to the Open Air CFO, Sarah Fryer. Open Air is a software company, or it was. Now a lot of the focus is on Sam and the team's ambitions with safeguarding infrastructure. How involved are you in that? Talking about the sort of construction of five gigawatt datacenters and the financing of such things. Yeah. Was that a surprise to come in and sort of think, okay, I need to get a handle on that project? Not a surprise, but definitely a stretch. It's new territory stretch from a

capital perspective, from a capital and also just my own learning. Frankly, I think we're all learning in this space. Infrastructure is destiny. It's this wonderful phrase that Chris Lehane has managed to get out there in the world. And we do think that this build is important. It's important for our competitiveness, it's important for world productivity. It's important even with a national security lens. And so you are right. One of the key jobs I need to do is to figure out that capital allocation story. It's going to be both working with partners. It's going to be raising financing, but it's always making sure that we are staying ahead of what will be required. Call it two or three years out because you can't just turn on compute today if you need it. You actually have to have thought about it probably about three years ahead. On that, if I may, one of the things I

heard from some of your investors is you did a very good job early in explaining the basics of the plan, but the ultimate goal is AGI. That's correct. How confident do you feel you have the sort of infrastructure in place or a plan to have it for AGI? I think we have the plan in place. I think if Sam were sitting on the seat, he would tell you AGI is closer than most think, but I would. What would you say? I would agree based on what I'm seeing, like one of the best meetings I get to go to once in a while is the research meeting, and it would blow your mind to see what's already coming and what, as we have learned how to take reasoning models like GPT-4 and combine them with the T models and the interplay between those. You're now really starting to see some incredible outcomes, page HD level outcomes where you have in your pocket

human intelligence, that is sea level and physics and biology and chemistry and English literature, like whatever the job is you need to do if you're a health care researcher, access banker, if you're in education, the tool that you are now carrying, the power there blows my mind. Sarah, you touched on it a moment ago, financing the need to raise capital. Yeah, you've just done a pretty astonishing large round, but that a follow-on. I mean, you must have a good sense of the cadence of needing to read on an annual basis. I don't know how you planet. Yeah. So it goes back to what you said, which is really understanding in your plan. What are your big expenses? Compute is the big s, but we also need to run a company, so we have real operating expenses. We're right in the guts right now of our FY 25 planning. That is usually a three-year outlook in most companies.

And for us, it needs to be because we have to make those compute decisions. And with that comes then, okay, what is our balance sheet going to look like? What's our cash burn rate? At what point can we generate enough free cash flow to fund the business? Not ready to tell you that today. That's for the next time we talk. And then on top of that, how do I help keep bringing our syndicate of investors along? I call them. Maybe with us. We are less than a week from the election, and I think about my own use of force. I have not used it related to the election searching for information. But are you preparing for that election, and what impact might it have on open air? Yeah, we absolutely are. But we have to be very mindful from a safety perspective. Today, if you do searching around the election, you'll actually see that often we will not turn a response or we'll

return with a caveat that says to be mindful of your sources. And I think I learned a lot of that, frankly, at next door. Right. You cannot you need to be careful of not being paternalistic or paternalistic around people. People need to be able to find information, make their own educated decisions, but at the same time you also need to be very aware when you have a platform that today services 250 million people every single week. We have to recognize that they're going to want to do things and we need to provide avenues but in a safe and secure way.

OpenAI CFO Says AI Isn't Experimental Anymore

Welcome to the Bloomberg Technology YouTube channel. Subscribe for original series and mini-docs on game-changing technologies, profiles on innovators, live conversations with experts, and breaking technology news & analysis from Bloomberg TV.

Bloomberg Technology

현재 은행과 금융 기관에서 생산성과 고객 성과를 개선하기 위해 AI를 활용하고 있으며, 모건 스탠리와 클라나 같은 유명한 사례도 있습니다.

금융 서비스 부문은 OpenAI의 매출에 크게 기여하는 분야로, 3대 업종 중 하나입니다.

이 분야는 AI를 조기에 도입하고 비용 관리에 중점을 두어 매출 성장을 주도하는 핵심적인 역할을 하고 있습니다.

이 회사의 주간 활성 사용자는 2억 5천만 명이며, 5~6%는 유료 고객으로 전환합니다.

사용자는 시간당 1,000~2,000달러를 절약할 수 있는 도구의 가치가 상당하며, 엔터프라이즈 요금은 한 달에 60달러 정도입니다.

이 회사는 시간이 지남에 따라 가치와 가격을 조정하는 동시에 대규모 컴퓨팅에 투자하여 회사의 성장과 미래 잠재력에 자금을 지원하는 것을 목표로 하고 있습니다.

대체 비즈니스 모델 및 글로벌 액세스

이 도구에 대한 글로벌 액세스를 제공하는 데 초점을 맞추고 있으며, 순수 구독 모델에서 광고로 전환할 의향이 있습니다.

Microsoft와의 파트너십은 거의 5년 동안 지속되어 왔으며, AI IP를 대가로 독점 컴퓨팅을 제공하고 있습니다.

회사가 성장함에 따라 컴퓨팅 파트너십을 확대하고 소비자 영향력을 극대화할 계획입니다.

OpenAI는 최고의 연구 인재를 확보하기 위한 경쟁에 힘입어 직원 수가 500명에서 거의 2,000명으로 빠르게 확장되었습니다.

이 회사의 API 제품은 큰 성공을 거두었으며, 모든 AI 유니콘이 OpenAI의 API를 기반으로 구축되었습니다.

인재를 유지하기 위해 OpenAI는 직원 공개 채용을 실시하여 유동성을 확보하고 컴퓨팅 리소스에 대한 수요와 공급의 균형을 맞췄습니다.

연사는 경쟁력, 생산성, 국가 안보를 위한 인프라의 중요성을 강조합니다.

이들은 자본 배분, 자금 조달, AGI 달성 목표를 지원하기 위한 미래 수요에 대한 계획에 초점을 맞추고 있습니다.

연사는 추론 모델과 AI 도구의 혁신적 힘을 결합한 인상적인 결과를 강조하며 AGI 계획에 대한 자신감을 표명합니다.

선거 및 비즈니스 의사 결정에 대비하기

이 발표자는 현금 소진율과 잉여 현금 흐름 등 회사의 재무 건전성을 평가하고 의사 결정을 내리는 데 집중하고 있습니다.

또한 투자자의 참여를 유지하기 위해 노력하고 있으며 다가오는 선거에 플랫폼이 미칠 영향도 염두에 두고 있습니다.

발표자는 대규모 사용자층에게 정보를 안전하고 안전하게 제공하는 것이 중요하다고 강조합니다.