- 18 hours ago
Category
🗞
NewsTranscript
00:00What brings you to Europe and to the UK?
00:02Yeah, well, the incredible demand in Europe and meeting with all of our customers and our partners.
00:07We just had a fabulous customer event called Frontiers.
00:11And, you know, the UK is just the UK, Germany, France is growing tremendously.
00:16Sweden and Spain have great momentum.
00:18And so there's just so much going on and so much momentum.
00:21It's been a great week.
00:22ChatGPC and OpenAI have been known, obviously, for the consumer part and winning that consumer story.
00:27There's an increasing focus on enterprise.
00:29When does enterprise become the biggest revenue stream for OpenAI and what kind of time frame?
00:35Is that the aim?
00:36Well, I mean, the aim is obviously to meet both customers, our consumers and our enterprise customers.
00:41But you mentioned the demand around ChatGPT.
00:43And I think that is really such a momentum that comes into the business, into the enterprise business.
00:48Because as you use ChatGPT at home and you think about it for your personal life, then you want that
00:53same experience at work.
00:54And so today our enterprise business is 40 percent of our total revenue.
00:57It is growing.
00:59We expect it to be 50 percent by the end of the year.
01:01And there's incredible momentum.
01:03And that 50-50 balance, you're comfortable with that?
01:05Yeah.
01:06Or you want it to be lopsided more towards enterprise?
01:07I think we want to meet both consumers and enterprises.
01:10And the focus is, you know, especially right now, there's so much happening in the enterprise space.
01:14What we see is I actually think we're at a tipping point in enterprise AI because as these models advance
01:20so quickly and the innovation cycles continue to accelerate, there's this opportunity to bridge that innovation and the capability with
01:28how enterprises are thinking about adopting the capability.
01:30If you think about a couple of weeks ago, we launched GPT-55, which by all accounts is the most
01:36intelligent, the most intuitive, and also the most efficient model.
01:39But yet when I'm here in Europe and actually across the globe, what I still hear from CEOs particularly is
01:45it's hard to keep up with the compounded innovation of what's happening now.
01:49And so I think the momentum, there's so much potential, really.
01:53There's kind of a vertical wall of demand in the enterprise that is left to be captured, and that's really
01:58exciting.
01:58Does that momentum mean that you beat Anthropic on enterprise this year?
02:02Well, what we're focused on right now is meeting the customers in where they are right now.
02:06And what I additionally hear from companies is just this transition between using AI as a productivity capability.
02:13And the real change now, Tom, is that what I hear from companies is they want to move beyond that.
02:17They want to move to bring intelligence into the complex workflows.
02:22This week, we announced a new company called the OpenAI Deployment Company.
02:25It was really designed to meet that moment, which is moving from productivity to true complex workflows requires a different
02:33operating model.
02:35So bringing a forward deployed engineer on site, working with a company so that you can understand their tools, their
02:42workflows, their systems,
02:43and then connect these models is really how you bring that true agentic capability into the enterprise.
02:49That's the real economic unlock.
02:51And we're excited about the OpenAI Deployment Company as a way to accelerate that.
02:54What do you think investors and maybe some of your competitors miss in terms of the demand that you're seeing
03:02and your interaction with CIOs and CEOs?
03:04Is it the API use?
03:06Is it revenue per customer?
03:08Is it the contracts that you're signing around enterprise?
03:11What are investors and your competitors missing, do you think?
03:14Well, I think what we hear from our customers, honestly, is they really love the breadth of our offering from
03:19OpenAI and the capability.
03:21For example, you can use the API.
03:22You can use ChatGPT.
03:24Codex is really having a moment right now, especially.
03:27I think this is probably one thing that requires a little bit more emphasizing, but we've seen the weekly active
03:32users go from 2 million to 4 million in a matter of 30 days.
03:37And here in the U.K. and Germany, in fact, 9x and 10x, respectively, growth, 50% of that work
03:43in the U.K. is actually not coding, not the software development lifecycle, but knowledge work.
03:48And that is really the big opportunity, which is to transform the way the world works.
03:52And I think that's where we see so much momentum.
03:54Does that pace of growth continue?
03:55Yeah. Oh, I think it will accelerate.
03:57You know, the momentum around Codex right now is just accelerating through all regions.
04:02And this is this is the moment, actually, I think.
04:05And what I hear a lot from customers that work with us is they love, like I said, the breadth
04:09of the offering that we have, but also the way that we really lean in.
04:13We are a research and deployment company.
04:16The incredible research is on fire.
04:18The pace of acceleration can't be denied.
04:20But what we're also equally focused on is the deployment capability.
04:23And for us to bring the power of intelligence to everyone in the world and every person, we need to
04:29do both.
04:29We need to help bring the model and we also need to help with the deployment and the adoption.
04:33And really what we hear from our customers is that is a differentiator for us in how we work with
04:37them.
04:38You're based in Silicon Valley.
04:39You're here in Europe and the U.K.
04:40Is Europe behind on adoption?
04:42No, actually, what stood out to me this week that I think is really, really incredible is this move from
04:49the productivity era to the intelligence era.
04:52Every single CEO, CIO, every executive that I met with was thinking way beyond productivity.
04:58They're thinking about transforming industries, unlocking capabilities.
05:01If you think about BBVA, BBVA is a great example of a company that started with ChatGPT and grew to
05:08over 100,000 users with ChatGPT, but now transforming material workflows across different areas of the business, whether it's customer
05:16service, it's IT.
05:17They're leaning in in a really accelerated way.
05:20And that stood out to me that this is a transformational moment and the companies in Europe are leaning in
05:25and really wanting to do that and asking for how can they operate and think about this capability, again, not
05:31just as a technology, but as an operating layer.
05:34And that is where companies really start to compound the advantage and move ahead.
05:38And that's what really stood out to me here.
05:39I want to ask you about trust.
05:41Yes.
05:41The unveiling of Mythos, Bioanthropic, at least partial unveiling, has raised questions about safety once again amongst some of these
05:48models and for some of these monies around some of these models.
05:50You've got the court case right now, Elon Musk, of course, in court, pursuing that court case against Sam Altman
05:55and some of the OpenAI team.
05:58Does trust come up in your conversations with CIOs and CEOs, with the executive suite, with your clients?
06:03And if so, how do you address that question?
06:05Yeah, I mean, what customers are talking about is cyber right now.
06:08And if you think about as these models become so advanced, so much capability to unlock, but also requiring safeguards
06:16in terms of that capability.
06:17And we've been thinking about this for a few years already with our cyber preparedness framework, which is really at
06:23the essence of this.
06:24This is about democratized access.
06:26This is about iterative capability.
06:28And this is about making sure that defenders have this capability and accessibility.
06:32But we're also creating cyber resiliency.
06:34And that's been a big topic this week.
06:37And it's one that we are really excited with our model 5.5 cyber that we're giving trusted access even
06:42to European, even to EU institutions and businesses as well in a trusted access framework.
06:47Because Mistrat, the CEO of Mistrat, Arthur Mensch, has said, ultimately, essentially, he said to European companies and governments, you
06:55can't trust U.S. tech.
06:57What is your response to that?
06:58My response is we have been very clear about the cyber preparedness framework for the very beginning.
07:03We're very clear, very transparent about the capabilities and also very active in building cyber resiliency.
07:09And that program in terms of the cyber framework and the fact that you released that under this framework to
07:15certain companies, you started to do that in Europe already.
07:17Yes.
07:18How many banks have you partnered on that?
07:20Can you tell us?
07:20I can't give you the exact numbers, but I can certainly tell you that BBVA is involved in it and
07:24Deutsche Telekom and Telefonica.
07:27And we, as I said, the principles around democratized access, iterative capability, and making sure that we give the capability
07:33to the cyber defenders so that they can continue to advance the cause is always part of what we do.
07:38And we'll continue to roll that out in a trusted framework.
07:40At this point on the banks, is it a handful of banks or in the dozens?
07:44It's a handful of banks.
07:45OK.
07:45Did you have to get approval from the White House to give access to this model?
07:49We work through our trusted framework.
07:51And that involved getting signed off from the White House?
07:53That involves making sure that we follow all the regulatory requirements for access.
07:57OK.
07:57OK.
07:58Well, you continue to watch that space then.
08:01Advertising.
08:01You're piloting advertising.
08:03Yeah.
08:04In countries like Brazil, Canada, Mexico, here in the UK as well.
08:07Yeah.
08:08How significant will the revenue stream be for OpenAI from ad revenue two years from now?
08:14Yeah.
08:15I mean, I don't want to project out two years from now, but I can tell you that this is
08:17really an interesting area.
08:19The response from both advertisers and from users has been really positive.
08:23Great momentum, as you mentioned.
08:25We rolled it out in the UK last week.
08:27We don't have any results to share about that yet.
08:29But overall, we see this as a really important opportunity for us as we move forward.
08:33Does ad revenue end up being core, as a core revenue stream for OpenAI in the future?
08:38It'll be an important revenue stream for sure.
08:40OK.
08:41I want to ask you about Stargate and Stargate UK reports that you've paused in terms of a partnership on
08:49building out a data center in the UK.
08:51Energy costs, regulation being cited as two of the factors.
08:55Can you give us an update on that?
08:56Is that pause still in place?
08:57Are you rethinking your commitments in the UK around the infrastructure, around the data center store?
09:01Yeah.
09:02The way we're thinking about this is we need to have a global approach for that.
09:04So we look at it across the globe and think about how we prioritize our investments and continue to watch
09:08that.
09:10OK.
09:11But your commitments to the UK, when can we get an answer as to whether or not that's going ahead?
09:16Do you resolve that question this year in terms of whether or not you're still committed to that data center
09:20in the UK?
09:20You know, I can't give you a specific time when we'll resolve it, but I can tell you that we're
09:23actively looking at that from a global perspective.
09:25Is it mostly an energy question or a regulation question?
09:27I think it's all of the above.
09:28It's thinking about how we think about managing our investments, how we think about the areas that we need to
09:32grow.
09:34Particularly in London, we see a vibrant activity here.
09:37I mean, we just had a session with a thousand developers who are developing on Codex.
09:43The market, it's going to be our second largest research hub here.
09:46We're opening a new office.
09:48There's so much going on here.
09:49And so when we think about our global strategy, we'll think about the demand, the presence.
09:52What is the business case for building out and investing in data centers in Europe where energy costs are two
09:57to three times higher than in the US?
09:59You know, I think we look at the demand and certainly the global economy and where we think we need
10:02to be based on what customers need.
10:06Okay. How do you resolve that energy piece when it comes to Europe?
10:09Are you getting any signals from European governments and leaders that they are taking that into consideration,
10:15that they are concerned that they don't have that data center build out that equates to what we're seeing in
10:18the US?
10:18You know, what we see is there's demand everywhere.
10:20And so we're trying to meet that demand and meet all of the inbound requests that we have
10:24and certainly manage all the things that we have to consider when we make an investment.
10:28Often when we speak to investors who are cautious on what's happening in AI,
10:32there's no sense from the equity markets right now that there's that much caution,
10:35but they point to that commitment around infrastructure spend.
10:37And Sam has spoken about potentially a trillion dollars of infrastructure and data center spend
10:42to power the innovation, the developments and shipping the products that you're shipping,
10:45that there's a gap between that spending commitment and what you're seeing in terms of revenues.
10:49Is that a valid concern? How do you answer that?
10:51Well, listen, I have to give so much credit to Sam and Greg for really what looks like now being
10:55prescient
10:56to understand the demand and the compute advantage that will happen.
11:01So the more we invest in compute, the more intelligent and capable the models become.
11:06And that means that our customers can actually derive more value out of it
11:09and they can do more things and deliver more economic value.
11:12And that becomes a cycle for our business.
11:13And I really do think that we've been very forward looking in that.
11:16And then, as I mentioned, the demand is sort of a vertical wall right now
11:20as enterprises think about how they roll out, not just productivity,
11:23but a true AI platform in their organization.
11:26Is Codex and Agentec, is that a tipping point for revenues?
11:28Yeah, I think Codex is an absolute enabler, not just for the software development lifecycle,
11:33but for all of Agentec work.
11:35If you think about it, you can, I use Codex.
11:38Here's a great example of me using Codex just as a normal knowledge worker, if you will.
11:42I use all of our products.
11:44I use the model to summarize my day and what do I need to do next?
11:48When I used Codex with the 5.5 model, what I noticed was that Codex now understands
11:53not just what I'm asking, not just literally translate my day.
11:57When I say summarize my day, it's intuitive.
11:59It knows the shape of the problem.
12:01It knows to look in Slack.
12:02It knows to look in email.
12:03It knows to look in Salesforce.
12:05And actually, it not only helps me summarize the day, what decisions did we not make?
12:09What are my open commitments?
12:10And I think about this as, this is going to fundamentally change the way that we
12:15think about how we can work.
12:17And I think the opportunity and the obligation for us right now is to make sure that every
12:21human being, every company can understand the power of you can just build.
12:25If you can imagine it, you can do it as an average person at the work, which is fun.
12:30And that drives token use and token consumption.
12:33Yeah.
12:33And 5.5 is the most efficient model we've ever released.
12:36You can monetize that.
12:37Yes.
12:38Okay.
12:38And you can monetize it more effectively than previous models would be the argument.
12:41Does that, does that put you then on a, on a path to profitability within one to two
12:45years at open AI, or is it kind of an end of decade story?
12:48Well, I wouldn't comment on timeline relative to profitability, but what I can tell you is
12:51that based on the momentum I shared about Codex going from 2 million to 4 million weekly
12:56active users in 30 days, the 50% of the UK usage of Codex is not coding.
13:03You can see that this is just an opportunity for all companies and certainly for us to
13:08solve problems for our customers.
13:09Do you have a view on when the spending on data centers and infrastructure starts to slow
13:14down?
13:15I don't have a view.
13:16I mean, listen, I think that as we've said, being prescient about the compute investment
13:20and the infrastructure investment has really set us up to have a structural advantage to
13:25meet customers, to be reliable, scalable, and meeting this demand that is happening as
13:30enterprises move from just using capabilities for productivity to transforming a business.
13:34Where are we on the investment cycle?
13:36Is this the early and opening innings in terms of the investment cycle or halfway there?
13:41Where, where are we in that investment cycle?
13:43I think we'll continue to scale demand, scale the infrastructure and the investments as we
13:47see demand and certainly demand continues to grow.
13:50I can't really comment on how exactly we're going to invest.
Comments