And it has to be culturally specific. If
you just take war, the German way of
fighting was different than the French
way, which is different than British
way.
Which is different. And you see that
nowadays. But
the the the There there are technologies
that are linear. This is not a linear
technology situation. And if you take a,
you know, and and so it's just crucial
that everybody
has a private space where you're like,
is this working? Why is it working?
What's not working? And then in every
country you have part of the stack you
can realistically power. I mean, China's
a little different. I mean, I think that
part will be mostly powered by Chinese
companies. But every other country will
have to make a decision
of
what part of the stack will be powered
by American companies, what part of the
stack will be powered by indigenous
companies, local companies. Who are
those local companies? Who are the
experts?
Um
And that expertise
is very very rare.
Um and it it this is also one of the
weird things about this industry is um
there's so many things you It's like
your industry. Like you're a master in
your finance industry. There's so many
things you know about finance
that it would be hard to replicate. Like
say we switch positions. I I am certain
I would It's like And that kind of
knowledge is um just really important to
make sure you have a sovereign ownership
up in your country. And if we can help,
we will. But ultimately, you'll probably
have to do it yourself.
>> So
Can you help us understand the
intersection between Palantir
and the AI frontier models? And how how
does that work? Like how do you bring it
together? How should they understand
sort of cuz they're trying to understand
the ecosystems of the AI world. And
that's one of the discussions today.
Like what's that ecosystem? So maybe you
could sort of lay that out.
>> terms, there you have compute, so I
think Jensen was here.
>> Yeah.
>> World's expert, talk to Jensen. Um
and then you have basically then the
caricature of it is you would have
frontier models and it real in real in
reality you're going to have frontier
and non-frontier models and non-frontier
open models that you can make
frontier-like in specific use cases. Um
and
and so, what you're trying to do, let's
say you're trying to solve a problem,
like how do I optimize the supply chain
of my country in a way where I know
every single dependency on a country I
may not trust. Okay, that's a that's
literally an issue every single every
single person in this room has. Um and
and or may not trust tomorrow or if if
this part left, what would be the
replacement? Replacing the parts are
very difficult because a part isn't a
part. Part made by this company is
different than even if the same spec.
Every single every single manufacturing
company in the world has this. Um and
how do I do it while while not exporting
my way of manufacturing um
to another country. Let's just say I am
a French company, I have a way of
building a nuclear reactor. I'm pretty
interested in nuclear reactors are very
complicated. I need to know every single
part of my supply chain that is not
French or not European and I also need
to know if I had how would I have to
rebuild the supply chain if I had to buy
a part from someone else and what would
be the cost. These are So, that's a no
business every business in the world has
this problem. Okay, so um
you
if you want to make a frontier model be
able to work on that problem both
passively, meaning you know the answer
when you ask the question, actively,
meaning you know the answer when you ask
question, but also passively. You get
You guys get in a fight
and you're like,
I mean, you're tough guy.
So, that's just and you're like, okay,
we got to we got to rebuild the supply
chain. Okay, um
how do you get pre-
these models without an application
layer? Well, A give you an answer, but
that answer even in the the active state
is not 100% accurate. It is a estimation
of the answer and it also degrades over
time. So, first problem is how do you
make sure that answer is correct and you
can decomp it. So, how do you take apart
the question so you can see every single
assumption in the question in the answer
so that if you're a country leader and
you have to decide that that that what
you did in order to do that you will
need an application layer. You will also
need an application layer, which is what
we provide. I'm not just but you'll need
to you you someone will have to provide.
You also need to make sure that your way
of building the supply chain is not
being migrated to a large lab to a lab
model company because that's why you
have a company. So, okay, in reality
what you're going to do is you're going
to bring someone in who understands what
part of the data has to be captured in
this model. And to do that you're going
to have to train either post train a
closed model
or as likely currently you're going to
train a an open model on the precise
questions you need to ask and answer to
and you're going to do it in a closed
environment. And that's how you're going
to solve this problem because otherwise
you are
you have to worry about let's just say
you tell me cross country skiing is evil
and I pull your problem. None of these
things actually happen in practice. And
so, where what does that mean for you?
It means you need someone who
understands the strengths and weaknesses
of every model.
Uh also the legal vulnerabilities that
you're putting yourself into by using
certain models. You need someone who
technically understands the interplay
between your proprietary data and a
model as it exists and a model as you
can train it and can also set up an
architecture that will allow you to
bring in more expensive and cheaper
models, but also more secure and less
secure models because
there are also then you may not care if
a model from America that you don't
trust is running on your ad campaign in
Argentina, which is your biggest market.
You will care if it's running on how you
actually make your nuclear reactor work.
And so that part of the stack, we are
the application layer north, we will put
our an application and south to some
extent the integration layer, we will
put our our ontology in the middle and
allow you to train these models so that
you can control them and there's no
exfiltration.
But whether you used our product or not,
you will need this to make these things
A work and make them secure. And and if
you and one of like just to my biggest
criticism of regulation outside of
America is
please get technical experts to do it.
Like currently the regulation is meaning
it's a meaning the application of it is
you have companies that are essentially
there to deploy large language models.
Those companies are exfiltrating, not
intentionally, your alpha to the company
that provides the model.
It is like the opposite of what you're
trying to It is literally the opposite
of what you're trying to do.
And
if this is not sovereignty, this is
anti-sovereignty. And whoever's telling
you this is works needs to be fired
because they are hurting your country.
They're not hurting me, but this is
literally the way to export every one of
your secrets to some other company. And
it's only because the people doing the
regulation don't actually understand the
technical material. And this is very
hard. Every country has this problem.
Like how you know, I think you've been
very effective in kind of tweaking at
the edges when people have gotten out of
control and and in a way that you know
was very effective.
Uh and um
but we have the problem like who's going
to monitor the development of AI? It's a
very hard problem because you need to be
an expert. And how would you why would
you trust the expert? That's a really
hard question. One simple way of asking
and answering the question is
does the expert is the expert on the
payroll?
Cuz a lot of these experts are actually
wrapping a company. I am, too. But I'm
not wrapping buying this product. And
part of the problem is, you know, we
have a very very you know, one of the
things about the development of nuclear
energy and maybe a fortunate
unfortunately war was the people who did
it actually weren't doing it for the
money. And we have a problem now and I I
mean I I I'm kind of more left. I think
it would be fair to say then, you know,
I believe in capitalism as the best
system because everything else is worse.
I think you believe in it as just the
best system. There's a slight
distinction there. Uh
I'm still somewhat of a German social
democrat or a which means things have to
work. I am skeptical of migration.
I think it was a huge mistake. Um I mean
not controlled migration, but what we
got. Um
uh but uh which was the opinion of
Helmut Schmidt.
That was Helmut Schmidt's opinion.
Maybe we could listen to him. Uh
but any case uh
he was a titan. Um for those of you who
never heard of him. Uh
but any case uh
Um it it's uh it it it's very very hard
to do these things correctly. And And
I'll tell you one of the most
interesting things about regulation
which is on you know, someone who says
one of the things I think is happening
is when we had CDMA and Europe had GSM,
these are the wireless technologies. I
think most most people thought CDMA was
better, but most Europeans thought the
regulation in Europe was better. And so
you ended up with a I think Europeans
would universally say they have a better
product. You can debate it. But I think
people are pretty And this is just a
very different revolution where the
technology is like a breakout
technology. And um and and and and you
just can't intuit what would be good for
your country. You have to really have a
deep understanding of like where where
is the where where is the IP and alpha
actually being monetized? It doesn't
inherently mean because a local person
is doing it, you're safe. It's actually
not true at all.
So,
is sort of a core
teaching moment for the group uh that
they should deeply think about making
sure that their data remains there. You
know, that that what they're training
the model on remains in their control.
That you need an application There's two
basic things. One, I mean, there's a
technical version. You will need an
application layer which prevents which
makes the models valuable, which helps
train them, and which makes them
specific. That will also allow you to
select weaker models that might be
indigenous and control them. So, like
let's say as a matter of theory, you
have I don't know, a pair You have a
model from Chile. It's weaker than the
model in America. You still might need
one to use it. And and and and open
models are we're like we we just we have
a very important client in America for
lots of reasons uh
they wanted to use open models. The open
model is being done by
is actually on certain vectors. Now,
again, that you can't generalize. On
certain vectors, that model is better
than a frontier model. Now, it's only
certain vectors, but it's the vector
they care about. Then, everybody has the
same issue as the US government. Every
single one of you has If you're like, I
don't know, we have a big presence in
there. Do you really want to call large
language model and provider and
negotiate the parameters of what you are
doing on, I don't know, your security
apparatus which in every country is
different. Um, so like, you know, the
security the bounds of what you would
do, for example, just a neutral example.
In Europe, in America, civil
libertarians believe that um
you should not purge data because by
purging the data, it's hard to see what
the government has done with the data.
Core to European concept of your of of
civil liberties is that you do purge the
data. I'm kind of a little bit on the
European side because
maybe you have an affair. Maybe
something like that happens. Do you need
everyone to know that? So, I kind of
like the purge the data thing, but I
think it's hard to say one or the other
is actually better. If you If you purge
the data, you can't really see what the
police did. If you don't purge uh if you
don't purge the data, you can, but it's
harder to have what might be viewed as a
normal personal life. Uh and most people
do have a normal personal life at some
time in their life. Um
of course no one in this room, but um
I may have had one. But uh um
and uh uh
so, those are but if you have to
negotiate the parameters of the model or
take
I mean, I I like I you know, we have
it's commonly known we have a lot of big
presence in French internal security. Um
and um but we don't have one in other
places.
Um Do French the the French internal
security are very effective at what they
do.
Do they really need to negotiate when
they use a model with an American
frontier lab of the parameters that you
they're using the model? By the way,
they might have a way of doing something
that allows them to catch people no
one's ever thought of. If you exportate
that, you no longer have a way of doing
it. And so, the the the those things are
I would also say then the second point
is not that related. Every single one of
you has a a national in this country who
is a titan in industry.
And I I told Merkel this. I've got I've
got someone I
I I talked I told a lot of you have to
get them to move home.
Whatever it I don't know how you do it,
but you got to get them to come and say
you are now going to do this for my
country and you're going to set policy
in my country. And those technical
experts are with the exception really of
China, they are sitting here.
And they're sitting here for lots of
reasons. We, you know, like again, the
reason why I think it's important to
remember how crucial the time is, we
don't have time to debate why people are
coming here. You can debate it. I can
tell you why they're coming here, but it
doesn't matter. They are here.
Almost everybody who knows this is here
and
they you need to get them to come home
to your country and sit next to you and
explain what regulation should look in
your terms. What should regulation look
like? What should the stack look like?
What model should we use? What software
we should use? What chips we should buy?
And what are and what is the cost of
not, you know, it may not be possible to
buy the best stack. Like I'm I'm
negotiating this with a lot of times of
with of countries abroad that quite
frankly are you know, like they're like
we don't want to be dependent on the
Palantir stack completely.
Which is, you know, it's a heavy lift
because
there are certain things that just we
have especially on the battlefield. But
I'm helping them to build a plan B.
For because of course they don't want to
be dependent on me. I mean, in this case
I'm wildly pro-European despite
I feel like being slightly maligned. But
I'm I'm supplying things to Europe that
are not even in my economic interest.
What if I get hit by a car?
How are you going to like dealing with
the next person running the company
who's like, "Are you not paying me as
much as the gas station?" That's what's
going to happen. So, like
developing these things and a technical
expert
which you have, can really really drill
into these issues in hours. A
non-technical expert is very likely
working for a company that wants you to
spend money.
And then last lesson, you don't need
money. You need knowledge. I
One of the things I won't pick on
countries here, but a country I care a
lot about is always telling me how much
money they're going to spend.
I don't want to hear how much money
you're going to spend. Who's going to
get the money?
Like the money doesn't matter as who as
much as who gets it.
And that is crucial. The person who is
in charge of spending, investing in tech
needs to be someone who really is
thinking, "Why am I doing this? I could
be making even more money in America."
And you remind them, "Because your
country is great."
And that's how you do it.
>> One of the
earlier conversations today
said that
having access to data centers
>> Mhm.
>> is non-negotiable. Whether you built it
or you rented it from someone else, that
the ability to have compute for your
country
is non-negotiable. Do you agree? Do you
buy that? What What's your take?
>> Um I I I just want to say for the record
your palantir I get on my so crazy I get
on my line. There's this insane guy
saying we're building data centers like
in Florida. We don't even build data
centers. So just in case anyone's
watching this, um I It's like, you know,
yeah, look, you you are going to have to
have You need access to compute, you
need access to your own data. So the
stack is compute,
data, ability to manage data, ability to
control the models, ability to train the
models, and ability to compete against
your competitors and adversaries. And
one crucial part of that is compute. And
countries are going to figure out
different ways to do it. I
I I would My go to We're partnered with
Jensen and and
Nvidia. I would I think he's the world's
best at this. I would take I I also So
you know, like there's different ways
we're doing some pretty interesting
things that are are not yet public um
around how you can build compute in in
kind of very sovereign architectures,
both commercially and in government. Um
so I that I think that is a starting
point. You also need electricity.
Um, and uh, if you look at, you know,
just like
again, I I would say, um, you know, the
move in France to build nuclear and
stick with nuclear is probably what's
powering like it's just that like we
have to have a non-politically
uh, it like Luddite conversation about
these things. Like you are going to need
a lot of electricity. How are you going
to get it? And again, I where like where
I I tend to think all these things come
with negative externalities. So I don't
think there's like I don't think the AI
revolution is 100% like good. But but
there's a huge dangers. But and with all
these technologies, but you but but then
being left behind is probably the
biggest danger now. The and and then we
all have and you know, some countries
here I know pretty well. We all have
political insanity in our countries on
both sides. By which I do not mean you.
>> Thank you.
>> I I I I but like how do you negotiate
with wide swaths of your populace that
you know, don't Again, I think they're
legitimate. I mean, I personally and I'm
I'm speaking for on my own behalf and
not in I think a big problem we have in
this country is
I don't think people are down for me
being 20 times wealthier.
It it it's just it it leaves a bad taste
in people's mouth. And I come from an
academic family. I'm not even sure how I
ended up with all this money. And so
like I do think that's a that's like
that's like a real issue for our
societies. And I do think like some
societies in this room have less of that
issue because it's like very much like
we're looking out for you. I think the
Arab Middle East here has a huge
advantage.
Um, because it's like uh, um, we are
like on the political front, we are
under enormous pressure because the big
technology is not egalitarian. And then
it's I think easier to organize if you
are a a country with a one
with a very understanding of a order
that everyone in the country that
belongs is being looked after. You just
have a lot less of these issues. And
there's no getting around these issues.
And so we have to figure out a ways to
bring people along and explain they'll
participate, but
but but I actually believe this, but I'm
not the right person to explain it.
And so it's like, "Guys, it's going to
be great for everybody." And I'm not
That's actually not my message. I think
it There are going to be some disruptive
things. I don't believe in the doom
message. Now, part of the problem we
have in this country is a certain
company was doing a doom message every
day. And that doom message is really
really scared people and and then
they're and it scares people because
there are elements of truth in it.
But the elements of truth have been
exaggerated to the point where it causes
huge political problems and I I I think
you will
you you see these across the Western
world. Uh maybe less actually uh
definitely less in Japan, Korea, China
than than in other places, but there I
think also um you just have you have a
very different setup. I mean, all those
countries are very different from one
another, but in the end I think um we'll
have political advantages on the
adoption of AI. Uh whereas the American
advantage is the production of AI. So,
that's our We have an enormous advantage
in the production. The dissemination of
AI and getting people to adopt without
political insanity, that's where we and
I would I would say without overreach
European countries have a pretty extreme
disadvantage. And so, how do you how do
you get that to work?
Is a real Maybe we should see if they
have any questionings. I don't know, but
I love your questions. We can keep going
with yours.
Well, I I think
uh I'd like to sort of end it with are
you
tell me about optimism. Is this uh an
optimistic time? You know, there are
always risks with change, but isn't this
an incredibly optimistic time? The
opportunity that they all have for their
citizens to be able to
uh extend education, knowledge,
uh democratize knowledge, the capacity
to change things. I'm interested in your
take. Is it Is it worrisome to you or is
it optimistic?
>> I ain't ready to put it.
I mean, like in our like my I I am
optimistic under the condition that
everyone goes home and says, "Okay,
guys, let's get real. It we're not
really taking this seriously yet."
Uh
I think
it I would say it's very similar to
battlefield context. You Under the
condition that everyone goes home and
says, "Okay, we didn't really take this
seriously, and there were lots of
reasons not to take it seriously, and
some of it is a joke." And I mean, as
I'm speaking as someone who's lived
abroad a lot of my life, many times
people think something is American
marketing.
Like, you know, it's a thing. People
like think we're the best at this, and
maybe it's, you know, that. This isn't
that.
And uh and so that's where it under the
condition that people like you're very
important people in the world, and it
going to your
very justifiably important, historically
important, in the present important uh
culture and history-shaping countries,
of which you all represent in one form
or another here,
uh and saying, "This is game time.
How serious are we?" And and I do think
you should get access to all the
classified data. What has worked on the
battlefield? Is a very important You
know, historically, the test of
technology has been what has worked on
the battlefield. Get access to the class
your most your actually classified data.
What has worked, what hasn't worked. And
then build something around we are going
to make this work and find actual
patriots in your country. Then I'm very
optimistic.
Um and then I'm also in then in the
American context,
you know, finding ways to educate people
about where things will work and where
they won't have a more honest
conversation.
I would be even more optimistic. But the
net net, I'm of course optimistic.
Um and you know, I guess I'm still an
academic, so I can't be 100% academic as
as cheery and lively as you are, but I I
I'm aspiring to it.
I'm aspiring to it.
>> Thank you. Okay.
>> Thank you very much.
>> Good afternoon, everyone.
Um
part of our objective
uh
was to bring about a variety of views
and ideas
um without filter.
Right? AI is without filter. And so I
didn't sit aside and plan what my