There is a lot being done on the
training of prior generation models, future generation models.
There is a finite amount of talent. Give the open eye side of this equation.
How much of a premium are you having to pay out?
I don't know if it's salary or stock or engineering freedom to get the top
names. Well, thank you, Ed and Caroline, for
having me. It's great to be with you.
You know, look, this is a competitive market for sure.
It's an amazing place that we're at because there's very
few numb
er of people that can really make such a dramatic impact in this
field. But we try and pull people into our
mission. And I think that's historically what's
been our strength, is people want to build in the place where we're pushing
the frontier the furthest and also where we really think about the impact of the
technology, both on a societal level and then also on an applied and industry
level. How big is open air now?
How many employees and where are they? We're about 1200 people.
500? Yes, mos
tly in San Francisco, but also
in in London and Dublin and increasingly other places in the world.
Increasingly other places in the world. The reporting was that Japan is the next
frontier for you. What's the strategy there and where you
choose? I don't know whether you call it opening
a new market the way you choose to have operations.
Well, we're very fortunate. We have a very global base of demand.
So we want to show up where our customers are.
We think this is going to be a global transforma
tion.
We feel a lot of pull from from places like Japan and Asia broadly.
And so we'll show up where we feel like we need to be.
And let's talk about why show up from an industry perspective.
You're here in New York talking to partners.
We understand some and the teams have been in Hollywood discussing how Sora
can be used, but ultimately how you can have a working relationship with
companies and industries that think they're going to be upended.
Talk to us about how those conversations going, f
or example, with Hollywood and
publications. Yeah, you know, we take a very unique
perspective here. We really model Open Eye as a
partnerships company, and we think this transformation is going to be very broad
based. And so we're one company.
We can't do everything ourselves. Being able to find ways to partner with
industry, with countries, with with specific companies, that's really
important to our mission and the way that we think about bringing technology
to the world. By way of recent exa
mple, we work really
closely actually with the publishing industry right now on ways that we can
deploy AI into publishing to make the quality of journalism better, to make
the quality of reader experiences better.
We've heard nothing but really actually a lot of enthusiasm from that industry.
And so we're always looking at ways that we can partner with industry.
I expect a lot more from us in the future on this front and sort of the
great example of where we really are excited about the impact
that we can
make in media. It's interesting that media and now
companies over in Europe, for example, seem to have got with that quite
quickly. Axel Springer, a key deal that you sort
of birthed early days and Axel Springer therefore seems for businesses to
perhaps a little bit more blue links within a catchy beat.
New York Times on that discussion point right now.
How is that legal fight and indeed behind the scenes discussion going with
those sorts of companies? Well, I can't comment on The Ne
w York
Times case specifically other than to say we think it's without merit.
But the overwhelming majority of what we
hear from the publishing industry is actually quite positive.
There's a high level of excitement everywhere about what these tools can
do, both to unlock better journalism, to give readers more access to information,
to enhance the way that people engage with content.
So the way that we actually think about even the fundamental business of of
information, engagement and consumpt
ion, these things are all really possible now
with language models and increasingly with multimodal models too.
So we think we're just starting to scratch the surface on that opportunity.
We've got a lot more to share there soon, but stay tuned.
You're being positive, being optimistic. Of course you are.
What's interesting is I come away from meetings where Hollywood is deeply
anxious. People in my industry are deeply
anxious. Goodness, everyone with a job is trying
to work out how they adopt A.
I.. And also there's this question that is
ongoing of open versus closed. You're going for another legal fight at
the moment. Elon Musk has dragged you into it and
wants you to rename yourself. What do you sit on the closed versus
open and how much more open do you have to become?
Yeah, well, look, I think we really focus just on where we can deploy the
technology and what's the best way to do that.
And at the end of the day, we're going to focus on where our customers are and
the way that they
want to deploy it. We work with thousands of enterprises
across the world. We spend a lot of time working with
those enterprises on ways that we can configure the technology specific to
their use case, specific to their stack. And we think that the easiest way to do
that is the way that we serve it. But that's not to say that we wouldn't
look at other ways in the future. We really ultimately want to be where we
think the best place to deploy the technology is going to be problematic.
Television
and radio audience worldwide. We're in New York City speaking to open
air, a red light cap I want to learn more about.
Where open air sits in its growth. Its day to day operations.
You just said thousands of enterprises. We came to know you because of the the
explosive response to GPT. And that was principally in the context
of a consumer facing chatbot. But it sounds very much like growth now
is focused on that enterprise arrangement, licensing of technology.
Can you give me proportions, the ra
te of growth for those two pillars?
Well, we've seen tremendous growth in the enterprise.
You know, 24 is going to be the year of adoption for AI in the enterprise.
We felt really like 23 was the year that people just started to wrap their head
around what was possible with AI in the enterprise.
But increasingly, the market is pulling us toward real applications that are
delivering real business results and a very broad focus on AI enablement.
So how do we think about bringing AI both to our ope
rations but also to our
workforces and then also to new product experiences that we can deliver to
customers? And we work with plenty of companies
that are doing all three of those things at the same time.
We think that's very possible and it's having really dramatic bottom line
impact today. So this is what we think the pull is
going to be and we're ready to support. Brad, you're the CEO, so I'm assuming
you're responsible for also some of the purse strings involved in Open AI's
finances. Compe
te must be a really considerable
cost right now. And with that, the energy consideration
as well. Are you making enough revenue from this
enterprise business to to keep the lights on when you factor in the compute
costs involved? We are, yes.
We we think that we have well, first of all, a very diversified business.
And so we serve hundreds of millions of users with our product.
And many of those users are paying users.
Of course, we serve now over 600,000 individual users in the enterprise with
Chad Liberty Enterprise. And so we're just seeing tremendous
momentum. We're very early innings on this.
And so we look at it as, hey, this is this is the start of something.
And ultimately our priority is just to be able to deliver the best
intelligence, whether it's in the enterprise or individuals, that's the
most useful that drives the biggest the biggest business impact.
Kara I'm thinking back to the conversation we had the other day with
Daniel Amadi from Anthropic and what she explained,
Brad, was the relationship of
S and bedrock, that getting a lot of business through that you have a quite
similar and actually a pre-existing situation with Microsoft Azure.
Are you able to explain how much new business comes through that conduit of
the Microsoft relationship and and what were probably existing as your
enterprise customers? We're really fortunate to have Microsoft
as a partner. They've been an amazing partner with us
for a long time. We've increasingly looked at the world
as way
s that we can bring open AI technology to life together.
We're very fortunate that they're offering our models through their Azure
Open AI service, but we have an independent business.
So we serve we serve customers independently of Microsoft.
And is that one bigger than the Microsoft Avenue?
Yeah. Look, we you know, we're it's it's fluid
between the two, but we really are focused on what's best for for for the
companies we work with. And we're very fortunate to be able to
diversify the surfaces
in which we offer our products.
We think the demand for this technology is massive.
And so being a smaller company, obviously having a partner that can help
bring that technology to scale is really, really advantageous for us and
we're excited to support them in the future.
It's an extraordinary, very novel kind of partnership and in fact, the whole of
AI is birthing these kind of frenemy relationships.
I just think of also acquires deals that seem to be going at Microsoft has just
seemingly ac
quired the whole of inflection.
What did you make of that particular move, particularly when suddenly you've
got what the co-founder of DeepMind coming in to be the head of consumer
A.I. at Microsoft?
And ultimately you put out a response to that being like, Congratulations, But
how did that make you feel as the previous partner of choice for
Microsoft? Well, we're excited for them.
We congratulate Mustafa and Karen and the whole team, and we work very closely
with Microsoft. And so we think tha
t's going to be a
major advantage for our partnership as well.
We're focused on what we can, you know, what we can control in the opening AI
side, which for us right now is really thinking about the next wave of the
technology and thinking about ways that that wave is going to start to enable
businesses. And the more we can think about that
problem together, the better we'll all be.
A lot of. CEOs.
I speak to some of them. All your customers say they're
technology agnostic right now. In other wo
rds, they might use GPT three
or another foundation model. They might also be building their own in
parallel. And when we told our audience you were
coming on the show, a lot of people wanted to get your thoughts on the
Palantir relationship. It's an interesting case study.
They seem to be one of the beneficiaries of the enterprise investment in AI, or
at least the data analytics side. What's it like working with Alex Karp?
Yeah, well, we you know, we actually work with a bunch of partners that
help
us bring the technology to life. And there's so many ways to deploy the
technology in the enterprise. That's what's kind of amazing about this
technology is it's not just one function in the enterprise or one part of the
enterprise stack. It's so diverse and it's so broad based.
And so we work with with with with many partners to bring that to life.
And we'll continue to look for partners to to work with to help us do that.
But yeah, now we you know, look, we're we're excited with with with
the way
Palantir has been enabling their customers on top of our technology.
And we'll continue to support them going forward.
Brad, I every day it crosses my desk. You know, I've done quite a lot of
reporting on Sam's chip ambitions as well.
There's some relationship going on between the UAE and open AI fundraising
or technology partnerships. Can you clarify what's going on there
and why you'd need to raise funds from somewhere like that?
Yeah, well, look, I won't comment specifically on that.
We broadly, though, believe that there's just not enough total supply in the
world and that of accelerators, specifically the accelerators and
broadly that the supply chain will need to adapt to what we think is going to be
this highly inflected and nearly exponential demand in the next ten
years. And so we spend a lot of time thinking
about ways that we can make sure that that demand is met.
Part of it is just being able to bring our models to bear.
But upstream of that is obviously making sur
e that the actual underlying hardware
and infrastructure exists to be able to build the systems that we need to
ultimately serve that demand. And so, you know, we think a lot about
it. And, you know, our hope there is that
the supply chain starts to think about that the same way we do.
Brad, your your mantra, your manifesto, your focus is to build artificial
general intelligence for the good of humanity.
And I speak to a lot of humans who are terrified.
Can you encapsulate how you can without re
gulation being that in place yet and
the U.S sort of putting the gauntlet out there?
You're doing all of this without disrupting.
Too much too quickly. Too hot.
Yeah, well, look, central to our mission is to make sure that we bring the
technology in a way that's safe and that creates broad benefit.
It's really encoded into how open I was structured, and it is actually the thing
that we spend our time thinking about how to deliver in real world results
today. And so we're very optimistic and I ca
n
understand where the fear comes from if things are moving very fast, even for us
at the center of it. It's sometimes it's dizzying.
But look, our perspective is that this is an enabling technology at its core.
It's going to make businesses more efficient.
It can make people more productive. It can be dramatically inflecting in how
people learn, how people can understand information.
And so we think that that has a very broadly positive effect on the world and
the world that we wake up to ten y
ears from now will be one that we look back
on, you know, and really are are excited to be able to go build today.
And, you know, we're that's that's that's really what we wake up every day
and want to go do.
Comments