OpenAI CFO：Q3营收近500亿美元，推迟IPO计划

Bloomberg Audio Studios, podcasts, radio, news.

So, here's the latest this morning. OpenAI expecting to hit at least $70 billion in annualized revenue for the year. The Chad GBT maker delaying plans for its highly anticipated IPO. Bloomberg's Tom McKenzie is standing by from the International AI Summit in Dublin and has a special guest. Good morning, Tom.

Good morning, John. Thank you very much indeed. Yep. Very pleased to say I'm joined by the CFO of OpenAI, Sarah Frier, here at the International AI Summit in Dublin. Let's start with with the wall of demand that you've talked about previously, Sarah. Where are we in that story, particularly when it comes to shipping of AI products, Agentic AI to your European and global customers? What is that wall looking like right now?

Yeah, so that surge in demand you're talking about has been massive. U we just put out some research this morning saying that here in Europe for example we've gone over 50% of all of the tokens we see are agent what does that really mean gone from asking to doing and actually with our new DOTS product that's available to all enterprises proactive uses um if you think about businesses companies like BBBA huge bank deploying it to 126,000 people times really interesting digital native startup here out of uh Dublin out of Ireland itself all the way down tomemes um I spent Monday tonight with about 250 women starting, running, growing their own businesses and there they are seeing life-changing outcomes from being able to just talk to chat.

Is Europe behind on adoption?

Europe is definitely trailing somewhat. Um, if you look at frontier firms that we define as the top 10% of of firms on our platform, only about 10% of them are European in in jurisdiction. Um, that doesn't mean Europe is getting left behind. I think we're just at the beginning, but it definitely should be a call to action to European firms to lean in.

What is your message to European firms who might look at open weight cheaper open weight models or models from Mistrial particularly given the sovereignty concerns and say we'll adopt those instead of open AI. What is your message to them?

I mean first and foremost I'm a CFO so I am like you should get the right model for the right task at the right price for you and that's our goal. If you want a frontier model, it's probably going to come with a premium price because you're going to be doing um chip design. You're going to be doing breakthroughs in drug discovery. But if you want to be able to use a model at the low end um with enough intelligence, but maybe it's in your call center cuz it's doing a more routine task, I need to make sure we need to make sure you're getting that model at the right price.

We took Luna, which is our cheapest model series. We dropped prices by 80% about 6 weeks ago. Um we saw use go up 10x. So that tells you there's a lot of price elasticity of demand right now. Um the other thing I would say to folks that are embracing maybe a Chinese model is also to think about the safety and security elements. Right? We know that from a cyber perspective, Chinese models are getting very strong in what they do.

We think you want to be on u more western models frankly on the cyber front and so this is another area of huge opportunity but we need to sell solutions not just sell fear.

You brought out safety so I'll go there. There's obviously been a number of incidents, hugging face notably, but also Australian government websites, executives from Open AI and other Frontier Labs yesterday telling a committee in the UK and Parliament that there may be more to be disclosed. Don't know yet. Should Frontier Labs be the ones who face the consequences if their agents go rogue? How do you think about where the liability should land?

Yeah. So maybe just first of all on what we're doing to make sure inside of Open AI we're doing things that are more safe, more aligned, but also to have transparency when things happen. So I think in Hugging Face's case, we really wanted to get the information out into the world, but internally we are doubling down on how we think about safety and alignment, putting more compute to bear. Um, our newest model, Astra, is our most aligned model, but there's even times we pace the frontier.

So, we might hold back a next generation model, which we're actually doing at the moment because we want to do a few more turns on safety and alignment. As we think about um your question about liability and so on, I often think about it with a bit more of a car analogy, right? There's clearly the parts that go into the car and if a screw fails or a part fails, there's liability with the supply chain, the car itself.

Car companies have put seat belts in action and so on to keep us safe inside cars, but if something goes wrong because of what they did, there's liability there. And then there's obviously user liability too, right? If someone drives while perhaps having drunk too many Guinnesses, you know, that also sits with the user. So I think the world is going to break into pieces um to think about where the ultimate liability originated and and then with that who should be making sure it doesn't happen again. and and and to be clear on the compute side, are you shifting more more spend and more compute towards alignment away from R&D?

Because that could be one solution.

Think about when you're training a model um in pre-training or post- training. Safety and alignment is an inherent part of it. So, we've actually always done a lot of investment in that area. So, I wouldn't say it's net new vast amounts of more compute. There's probably more, you know, just, you know, when you look at before and after, but it's not a major shift. But generally speaking, we're spending a lot more on compute to keep driving frontier research.

And that could be across every type of model, not just an LLM, but voice models, image models, even specific models to areas like life science. And then, of course, a lot more on compute to serve our customers because our demand is is growing so quickly. Can you give us any more detail about this individual who's been detained by the FBI in the US putting a device in your office? A Saudi national. Do you know what the motivations were?

Who was behind that? Who's driving that? Is there more? Unfortunately, it's an ongoing FBI investigation. So, I certainly can't talk about it here, but we are very focused on security and I'm really grateful to our security team that helped um bring that to some closure for us.

On revenues, there was a debate. There was a mismatch in terms of what we saw. the FT reporting one number 50 billion versus 70 billion that was briefed to investors on ARR. What is what it was behind that? Is 70 billion the number we should be looking at by by year end 2026.

So what we said is in Q3 we hit close to $50 billion. Um it was an incred quarter, right? Could make a CFO's heart sing. Um we grew over 70% just quarter over quarter. our enterprise business doubled and I think it's worth remembering for OpenAI we have a very broad-based model right today we're about 5050 50 in consumer which we monetize through subscriptions 50% in enterprises going from the smallest SMBsmemes all the way through to large enterprises and governments and some of that is consumption API based revenue some of it is subscription and then in the consumer we have this massive opportunity around something like ads that's just getting going So when I look at Q3 momentum into Q4, it's continuing at pace which is good to see.

Um and then when I look at areas like cyber as an example, I think that's going to be a big driver of our business.

Very briefly, is that what's being missed by people like Torston SL at Apollo putting out a note this morning referencing a survey saying we're still looking for the AI ROI that we're still looking for that and that the costs continue to surprise.

Yeah. I I me