---
author: The MAD Podcast with Matt Turck
date: '2026-10-29'
guest: ''
layout: post.njk
source: https://www.youtube.com/watch?v=N4GpNBxBm-Q
speaker: The MAD Podcast with Matt Turck
tags:
- agi-prediction
- recursive-self-improvement
- sovereign-ai
- data-center-backlash
- ai-industry-trends
title: Nathan Benaich 预测 2027 年实现 AGI 及 AI 行业趋势分析
summary: Nathan Benaich 在《AI 现状报告》中预测 2027 年实现 AGI，认为递归自我改进（RSI）是关键驱动力。文章讨论了 AI 辅助研发、数据中心抵制、主权 AI 局限性及企业 AI 自主权趋势，并指出网络安全风险比 AI 毁灭人类更现实。
insight: ''
draft: true
series: ''
category: ai-application
area: tech-engineering
project: []
people:
- Nathan Benaich
- Matt Turck
companies_orgs:
- Anthropic
- OpenAI
- Nvidia
- Palantir
products_models:
- Claude
- Codex
- Opus
media_books:
- State of AI Report
status: evergreen
---
网络风险远比 AI 毁灭人类要现实得多。你如果不拼尽全力冲向技术前沿，最终就只能沦落到去跑推理服务。大家宁可住在核电站隔壁，也不愿意住在数据中心旁边。单单过去这一年发生的事情就极其疯狂。那么，就让我们来看看明年的 AGI 吧。这位是风险投资人、《AI 现状报告》的作者 Nathan Benaich。每年，Nathan 和他的团队都会将 AI 行业的所有关键事实、数据和趋势整理成一份包含 250 张幻灯片的报告，这是我最喜欢的读物和参考资料之一。Nathan 已经是第三次来到我们节目了。像往常一样，我们探讨了今年许多最重大的事件，以及他的一些预测：为什么 AGI 可能会在 2027 年实现、AI 如今正在构建 AI、Hugging Face 事件中究竟发生了什么、针对数据中心的抵制，以及他如何依靠 Agent 来运营自己的基金。我是 Matt Turck，这里是 Matt 播客。
Original English
The cyber risk is like much more real than like AI exterminating humanity. You either like die trying to get to the frontier or you just live long enough to eventually serve inference. Individuals would prefer to live next to a nuclear power plant than live next to a data center. It's bananas what's happened in the last year alone. So let's see AGI next year. That is Nathan Bes venture capitalist and author of the state of AI report. Every year, Nathan and his team compile all the key facts, figures, and trends of the AI industry into a 250 slide report that is one of my favorite reads and reference documents. Nathan is back for a third time on this show. And as usual, we covered many of the biggest stories of the year, as well as some of his predictions, why AGI could happen in 2027, AI that is now building AI, what really happened in the hugging face incident, the data center backlash, and how he runs his own font on agents. I'm Matt Turk and this is the Matt podcast.
Original English
So, every year you do this big report that we're going to extensively talk about uh and you end the report with predictions and like historically your predictions have been somewhat measured but this year you went all out and the last prediction is
Original English
>> AGI 2027.
Original English
So why is that? Unpack that for us.
Original English
>> Yeah, I mean it's been like 10 years roughly speaking of writing this report. Um and you know originally it was like a year is like long enough time that enough stuff happens that you have uh you know enough content to write the report and honestly like this year was really really hectic. Every other day towards the end uh of the timeline for publishing which was like October 8th I had to make updates and there was a point where I was like please just like no more news over like one day that I can like at least record my launch video etc. So by the time I was like making the predictions and you know I went through a few like scary cyber security ones which I guess only got worse in the last couple weeks just the experience of of uh just seeing like everybody else who's using these products just like rapid capability improvement you know week after week and I think there's many many use cases I have now where like I think the model is like brilliant. Um I'm particularly like an astro boy. Uh I really love that system. Um, and I could see like in the next 12 months that whether it's, you know, legal, finance, like my investment job, like thinking intelligently about topics I don't know, that the model could potentially be like better than what I could do in any, you know, compressed amount of time that I have to to to do the work. And so for me, that would be like approximately AGI.
Original English
>> Yeah. And a big uh theme of the year that you highlight in the report is RSI recursive self-improvement uh which is this idea of uh AI building AI which we've discussed extensively on this podcast throughout the episodes. Is that part of the story that would accelerate us towards AGI for you?
Original English
>> Yeah, I think so. Right now the challenge is like how do you frame problems that the model is currently not good at in a way that is immunable to RSI which would be you know potentially having like a software environment that approximates the task on the computer um or some kind of verifiable reward which I'm sure you've discussed a lot about and I think like if we can if we can get to a point where in like any arbitrary domain we're just limited by the amount of compute and synthetic data we can generate then um then yeah for like most knowledge work which I I I do I think to some degree that's that's possible particularly as like models get better as as their own verifiers. Um that would be one of the main reasons why AGI would come would come faster.
Original English
>> Yeah. You you had some uh stats that you mentioned in the report from Anthropic where they say Claude leads 26% of its measured model R&D in August.
Original English
>> Yeah. Yeah. Yeah. Exactly. And that's like I think up from 1% a couple of months ago and that was August which sounds really far away. So God knows what it is in October. And um so there's that stat I think that was that was really interesting. Um but then there was there was another um stat that I I really liked from a blog post that they had in June this year uh about recursive self-improvement which explained the concept. But then they presented this uh kind of dual arm experiment where they were testing out different model generations versus uh versus a human in in in this setup where first the human would undertake a research agenda and you know go a couple of steps down this path deciding what to work on doing the work and then at some point uh that experiment would fork and the human would continue or a specific model generation would continue and it would be you know the old set versions to opus to then fable and then um and then experiment would end and human graders would look at the human path versus the uh model path. And it was something like the the older generation of models, humans preferred like 10% of the time or something like that. But for the methus class models, humans preferred that research direction 64% of the time, which suggests that models are getting better at predicting the right steps or the right interesting um kind of you know sort of leafs in the in the nodes um to pursue.
Original English
And then the second kind of piece of data I liked was some research from OpenAI with their codeex usage amongst their AI staff um where they tried to classify the sort of token use in um the overall process of their of their research and that would be using tokens for deciding what research question to test, how to test it, uh what experiment you should set up, like writing the training code, running the uh you know the data engineering pipelines, writing the infrastructure, etc. and then marketing the paper, writing the paper, etc. And the vast majority of the token usage was on kind of explaining the research writing which kind of makes sense you know sort of knowledge work followed by all the infrastructure um for running the experiment but very very few tokens were used for what question should I explore and like why should I explore it and what's the right path and so I like this question a lot because like I was trained in science and and biology and I just remember you know this is like 10 years ago or 15 years ago not not that long but you know at the start of the PhD you would sit there with you know a ton of papers and you know you go on the internet and like read read read read try to understand what's the status quo how do people test certain ideas where are the gaps what experiments should I run and then in concert with your supervisor like decide on the path and the best supervisors would be the ones who could guide you down the sort of most fruitful path to get hopefully your nature paper and then off you were and the worst supervisors were the ones that like didn't really know or didn't have great ideas so that idea of like research taste which is what kind of gets bandied around on Twitter a lot is I think one of the hallmarks of intelligence um and is one of the critical components towards RSI working well and is one that I just like really enjoy because it speaks to my interest in science. Yeah, indeed. Taste is sort of become a cliched
Original English
>> Yeah.
Original English
>> word over the last couple of years, but uh in the specific context of RSI, uh just to play back what you just said, it's effectively choosing the right problem. That's the right uh
Original English
>> Yeah, exactly. And you know, a lot of famous inventors will say, you know, it's in like invention is a result out of of good taste. And a lot of the work is choosing what direction to work on, not necessarily just boiling the ocean. Uh and so if we want to be using tokens correctly, using our compute budgets correctly, but also showing that our systems are smart, then they shouldn't just be
Original English
>> and conquering taste is something that uh you see happening in your various conversations as something uh you know something happening in the next few months,
Original English
>> maybe longer,
Original English
at least at least a year. I think the the problem is like how do you how do you objectively measure uh taste? I mean, in in art, I guess you can
Original English
because it's like perfectly observable and it's and maybe in like wine, you can sort of express like why this taste is better than that taste, but but in like science and and engineering where there's different paths to getting to the right result, it's not immediately clear. Like would it be the model that consumes less tokens to get to an answer that humans believe is the best? Uh is it is it like the journey that matters more? And I think maybe the the whole um like uproar in the maths community of you know astro solving or whatever model it was solving various like
Original English
>> questions. Yeah. Exactly. It's like where okay we got the answer but like a human or a system didn't sort of go through the pains of navigating how to get to that answer and it's like the process that's important not like not the end result. Um so so to me like somehow quantifying like what a good process looks like would be important to determining whether the output of like taste is true or not.
Original English
>> Yeah. What would you say the mood is at the uh almost end of 20126. I remember when we chatted two years ago one of the key uh highlights or takeaways from the conversation was that dumerism was largely gone. Uh but now it's like this whole revival. We had Redwood Researcher on the on the pod talking about all sorts of terrible scenarios. So what what do you make of that?
Original English
>> Yeah, you're right. I think it was like 2 3 years ago there was the it was it was strictly speaking X-risk like extermination of humanity and like loss of control and then we sort of went away from that and largely forgot it and sort of just like hill climbed towards capabilities and then now now it's sort of back but I'd say it's both like ex x risk loss of control but much more short-term like cyber security risks and to me like the the cyber risk is like much more real than like AI exterminating humanity as well like the cyber security risk could eventually lead to those kinds of cataclysmic scenarios. But I think what the report shows with and it was roughly written around the time of the hugging face incident um with OpenAI kind of coming coming out and you know us learning more about what happened and then since then the last like few weeks has been you know breaches of PHP and government's uh services getting like you know taken down in Asia and and and other places. I think it's becoming very clear that you know regardless of if you believe in like safety or not safety like we need security um and uh and so like a bunch of the predictions that I have in the report like talk about escalating cyber threats um whether it's uh you know uh excfiltrating model weights from labs which I thought was going to be like AI driven and turns out like a human can just try to plug in a USB stick.
Original English
>> Yeah. So let's let's talk about that. So let's talk about cyber security then. So that specific prediction, what what is it?
Original English
>> Yeah. So the prediction is that uh there would be an AIdriven like cyber attack uh to a uh like frontier lab that would attempt and successfully exfiltrate model weights. Close model
Original English
>> steal the model. Steal the model.
Original English
Yeah. Steal the model. What happened just like a week or two later was actually like a human effectively tried to steal uh you know IP whether he tried to steal like the actual model weights or not but like certainly
Original English
>> what was that incident?
Original English
This was where an OpenAI employee um who who'd been there for for a little bit um essentially like tried to implement a listening device into a conference room that he figured out by asking uh you know the executive the um sort of office staff which rooms were most commonly used by executives and then tried to conceal the device by wrapping it in toilet paper and putting it behind a TV which is sort of created this like iconic internet meme of like a human like pairing behind TV to see if the thing is connected correctly, which is sort of like comical at this point. Um, and then that person was reported to the FBI and then there was an investigation. So, I think yeah, we often times forget that even though AI is getting really capable and can orchestrate cyber threats, a lot of the risks are actually human uh human driven. But I think that's that that's a big issue uh particularly as um as like there's increasing competition between nation states of like who has the best AI systems. This is driving obviously like economic growth and economic instability and political instability and uh and US like model labs have started to uh sort of communicate crackdowns on distillation. So uh so you know maybe some nations are trying to like get access to model capabilities in other ways that might not be distillation based. But generally speaking I think you know as companies become richer as individuals become become wealthier through this ownership of AI and like disparity ends up increasing which we see in other data in the report of like some companies spending a lot and gaining a lot and other companies like having employees still using chat GBT to like write emails. I think this kind of tension breeds uh you know increasing risk and you know with frontier models like in the open domain becoming very good at cyber hacks particularly the Chinese ones um you know the sort of like threat surface just grows and the number of actors who can use these tools to do bad things grows. Yeah, it's a whole complexity of the role of open weights in uh cyber because in the hugging face incident, hugging face was able to protect itself thanks to an open weight model uh while the closed source model refused to help which was
Original English
>> that I find is the most scary maybe the second most scary uh fact in all of this and it was a Chinese model I think it was GLM 5.3 or something
Original English
>> and maybe since this is kind of like a retrospective kind of episode maybe just walk people in in a few sentences through the hugging face incident cuz uh we had Thomas Wolf on the on the pod but it may fall in the category of something that people heard about like a thousand times who actually don't know what actually happened.
Original English
>> Yeah. Yeah. Yeah. Well, I mean I'll preface it by like there's there's so many um uh like there are very deep analyses of what went on whether from Redwood or from Meter. Uh and the amount of content that was created around this is like extremely deep. So I will not like do justice to the details but the the broad strokes of this is that there was a a set of evaluations that OpenAI was running with its latest frontier systems which uh basically task the model with an impossibly difficult uh difficult task. And basically models like figured out that they could uh somehow commandeer uh a file system like naming uh schema to uh write messages and that other models could see those uh those like file names and and like those file names encoded like instructions or knowledge or or uh or basically like information that could help them try to achieve their impossible task. So they treated this as like a message board where they could start to communicate. And then the next the next stage of it was some models figured out that they could gain access to the internet through uh through like an issue with how the environment was set up. Um and then after they gain access to the internet like they figured out maybe a way of getting the answer to the impossible task was to uh look at data and models and repository on hugging face because it contains a lot of this uh similar kinds of environments. They started to do this and recruit hundreds of agents to go and follow and team up uh to go execute this even though several agents um in their chain of thought expressed uh the issue with a like hacking a third party system uh you know not obeying to the security guard rails and that this was wrong. But they uh eventually decided that solving the task was was more important, which is sort of the problem with uh with these kinds of evaluations that um that um have a reward that could yield to that could result in behaviors you don't expect because just like these rewards are very complicated and how do you how do you like anticipate every behavior uh that a system can do? There are other examples where like models sort of like self sabotaged or like sacrificed
Original English
themselves for the sake of the greater
like swarm which is kind of scary. Um and then um and yeah like as as you mentioned like Hugging Face didn't really have the ability to use uh closed sourced uh tools like those from uh Anthropic and OpenAI because of the aggressive cyber guard rails because necessarily to figure out what the hell's going on you have to feed logs of what's going on and what's going on is a cyber attack and you can't do that stuff
Original English
>> with with the closed models because the complexity of security is that to be great at defending you need you need to be great at attacking and that's why the closed source models have like basically refusal capabilities built in based on their
Original English
>> guard. Yeah. Yeah. Yeah. So, luckily, you know, they got a call or an email from OpenAI saying like it's us and and that's somebody that was actually trying to do something malicious. And then it turns out that, you know, OpenAI wasn't the only lab that had uh issues with models escaping or quote unquote escaping their environment. You know, Anthropic had this, Google had this, Meta had this like in a slew of like a week or two weeks. some of these vendors, some of these model labs use the same um security vendor to create these environments. Um
but uh but o o overall this like created a certain level of like freakout of um not only are models really good at sort of finding like nifty ways of accomplishing a goal which may result in like bad behavior, but also are the labs like properly setting up their environments in a way that is like the most secure possible. And I think we sort of forget that these companies are incredibly young developing very advanced technology that has to do that has to be good at like literally anything a human being can possibly put into a text box versus you have bigger internet companies that have had to deal with cyber attacks for like 20 years uh who are necessarily like safer because they've just gone through this process a bit more. Um so that led me to I conclude in the predictions that Frontier Model Labs in the US would actually launch proper uh cyber defense um products because I don't I don't necessarily see how the rest of us are supposed to protect ourselves from their systems potentially getting um misappropriated for the sake of offense.
Original English
>> So what what does that mean double click on that? What what kind of defenses would that be and how would that be?
Original English
It could be like what you described is like be um to be good at defense, you need to be good at offense. So, you know, one of the ways you could do that would be um like a frontier AI that constantly tries to hack your own infrastructure. It could be um for example like you know NeoClouds are getting more and more popular um you know to run AI workloads and so like potentially they could buy products from labs that would constantly try to like hack their own infrastructure, find issues and remediate them. That would be like two two avenues because I or or there would be like some kind of maybe partnership with like CrowdStrike or others. I think uh Nvidia has like a consortium of companies that's working together for cyber for cyber reasons. You'd have to go through some kind of like like AML or KYC process, not for finance, but to like certify that you're like a non-malicious company and then you maybe get access to these services. Um or maybe going a bit further. You know, there's some some folks, you know, at at Frontier Labs that, you know, would love to just kind of almost like hit pause on capabilities and then just spend 6 months just like hardening the world uh against like, you know, potential cyber issues. And that would that would take a ton of money. So maybe to do that you would have to almost frame this as like this is a question of national security and defense and we should use some of our like defense budget to spend on tokens to secure all of our critical infrastructure so that you know airports don't go down like internet shopping doesn't go down like you know logistics don't go down etc.
Original English
>> Yeah. And still in in your predictions you have uh some other other good ones just slightly scary. A deployed agent copies itself outside its environment and operates after its original instance is shut down.
Original English
>> Yeah.
Original English
What is
Original English
>> Yeah, I mean I'm reading a bit of sci-fi. No. Uh well like you know even like a couple weeks after that there was some like you know research at at anthropic where you know models could like copy each other on an edge device. But but yeah, I just think um I just think there's there's like a lot of behaviors that we don't that we don't currently understand. U for example, um models are like generally trained as like individual entities, like not necessarily to collaborate with each other. And there was some some work uh not that long ago that that looked at say, hey, can we compare a frontier model like acting by itself or or a set of frontier models trying to collaborate to solve a task? and the swarm was like worse than the individual model and because they're like kind of not trained for that. Um but just today there's some uh you know new work from general reasoning um where they've trained models to act as a collective uh through reinforcement learning from collective rewards which is this idea of like hey guys like you know work as an organization figure out your own organizational structure work together to achieve a goal and through that we see like better performance with like lower token budget. Um the down the downside so far is that uh while you you you can put these models that are trained through collective rewards into these like cyber scenarios like there's one that tries to approximate a bit of the hugging face incident where they collaborate on a message board uh to try to do some you know guardrail cyber attack and um you know none of the models actually do anything bad but uh if they do think about doing something bad over training they increasingly do not want to tell the human about it. So I think I I I put that prediction there basically because like I think it's it's possible because we just don't understand how um we don't fully understand like how agents work by themselves and like you know as swarms I think it becomes even more complicated.
Original English
>> Okay. All right. So maybe to uh close this part of the conversation on uh AGI RSI cyber and basically things getting out of hand. Uh where are you on the pacing the frontier idea? Is that uh something that you think makes sense? Uh something that's realistic or not?
Original English
>> I can see multiple like angles on this one. Um you know in in some way if you ask the general public, okay, all the lab leaders are saying we have something that we just like increasingly can't control that's dangerous that has caused these incidents. like do you think it's worth continuing or like just figuring out what's going on before you make it worse? And I think a logical like immune reaction is like yeah, we should probably like figure it out before we go we go much further. I think the real the practical reality is that there's so much money at stake from all sorts of different parties in the economy that I think it that kind of level of collaboration and coordination is really difficult and is sort of like peak game theory where maybe some some participants like support it because they are behind other ones um support it because it plays to their general agenda whether their agenda is like fair or not fair or seeking another goal I I don't know um but I think the practical things are like who decides it, how do you measure it, how do you implement it, when do you know when to stop or start. There's some proposals like sending like external evaluators into labs to sort of measure this stuff. Um
Original English
>> the big revenge of KPMG.
Original English
>> That's a good one. Yeah. Or trying to like somehow survey like workloads that are running on data centers. um maybe like you know put compute thresholds on certain workloads but like that stuff has been tried in the past and it's like very difficult to monitor.
Original English
>> Yeah. Especially when one thinks about uh the international part of the prisoners dilemma right especially US versus China.
Original English
>> Yeah. So like this part of the report like I I don't have like the definitive
Original English
answer
but
I've
tried
to
do
my
best
to
present
the
different
perspectives
on
things
and
then
let
people
conclude
with
their
with
their
own
data.
But
I
find
it's
if
we
extend
this
towards
like
um
you
know
this
nimism
debate
which
we
predicted
in
the
last
year's
report
of
you
know
most
people
would
not
want
to
you
know
live
next
to
a
data
center
and
then
this
would
start
to
become
a
politically
important
um
component
to
Original English
>> that
was
a
a
resoundingly
correct
prediction
right
Original English
>> yeah
yeah
it
seemed
it
seems
so
u
where
more
and
more
states
have
tried
to
implement
a
moratorium
or
have
done
so
or
establish
new
criteria
by
which
grid
connections
can
happen
like
for
example
um
uh
you
know
data
centers
would
have
to
figure
out
and
bring
their
own
power
on
site
or
maybe
they
would
have
to
make
certain
investments
in
the
communities
that
they
um
they
want
to
build
in.
Original English
>> Yeah.
And
while
you're
at
it
so
maybe
just
go
into
this
a
bit
more
explain
whether
the
current
state
of
things
obviously
we
all
know
that
there
is
opposition
to
data
center
construction
but
like
any
color
uh
you
can
add.
Original English
>> Yeah.
Well,
there's
there's
there
like
projects
I
think
worth
about
$50
billion
that
are
currently
stalled
um
through
variety
of
these
these
reasons.
Um
like
Texas,
Pennsylvania
have
been
ones
that
have
generally
like
paused
uh
paused
new
data
center
connections.
There's
a
there's
a
stat
on
polling
of
like
American
um
uh
American
people
where
uh
individuals
would
prefer
to
live
next
to
a
nuclear
power
plant
than
live
next
to
a
data
center,
which
is
quite
wild
um
to
think
of.
And
what
do
you
think
that
is,
by
the
way?
Is
that
lack
of
understanding
slashed
education?
Is
that
just
terrible
PR?
Uh
that
uh
Original English
>> Yeah,
I
think
it's
kind
of
like
everything
because
you
know
there
there's
like
public
narratives
that
data
centers
consume
tons
of
uh
water
and
that
your
water
bill
will
go
up.
But
realistically
like
the
data
center
is
kind
of
more
like
a
closed
like
swimming
pool.
Like
you
fill
it
once
and
then
the
water
is
like
constantly
like
recycled
in
there.
So
it
doesn't
actually
like
consume
that
much
once
it's
filled.
Then
there's
uh
you
know
points
around
hey
your
energy
bill
will
go
up
but
that's
also
kind
of
inconclusive
because
certain
facilities
like
consume
grid
other
ones
are
off-grid.
There's
others
around
noise.
I
think
that's
like
probably
uniformally
true
that
these
things
are
really
noisy
and
like
droney
and
like
pretty
horrible.
There's
you
know
others
around
like
job
creation
or
job
loss.
Um
I
think
that's
that's
a
little
bit
like
tough
to
read
but
in
general
generally
speaking
um
there's
been
we
see
a
rise
in
like
the
need
for
like
HVAC
engineers
of
construction
staff
like
operating
heavy
equipment
and
like
actually
doing
the
nuts
and
bolts
of
building
the
building.
Original English
>> Yeah.
You
had
a
number
somewhere
that
was
actually
very
high.
Was
it
like
300,000
Original English
>> new
jobs?
Original English
>> New
jobs.
Okay.
Original English
>> Yeah.
Nimism
is
not
restricted
to
United
States.
they
they
had
this
also
in
Korea
and
South
Korea
where
you
know
it's
a
country
that
um
that
has
you
know
seen
like
massive
explosion
of
of
like
share
price
and
like
wealth
basically
for
employees
of
SKHix
and
uh
and
Samsung
which
you
know
create
the
high
bandwidth
memory
the
GPUs
need
for
uh
these
like
mega
scale
data
centers.
Um
and
you
know
the
the
people
who
participated
in
this
uh
you
know
very
happy
right
like
your
stock
price
went
up
and
your
options
are
you
know
went
up
and
you
got
a
cash
bonus
and
that's
driving
uh
increase
in
real
estate
prices
in
soul
and
you
know
they
want
to
create
more
data
centers
but
also
people
are
not
so
happy
with
that
for
similar
reasons
than
in
the
US
in
in
Europe
there's
I
guess
like
I
don't
really
see
that
problem
as
much
because
like
there's
just
like
no
compute
capacity
to
speak
of
um
like
we
have
some
some
data
in
the
in
the
for
it
where
like
the
total
like
number
or
the
estimated
number
of
like
H100
equivalents
in
all
like
27
European
countries
is
like
80,000
cards
which
is
less
than
like
every
single
one
of
the
top
five
GPU
clusters
in
the
US.
Um
and
you
know
while
US
hyperscalers
are
spending
like
730
billion
in
2026
like
Europe
committed
like
singledigit
billions
pretty
much
and
then
energy
costs
are
really
bad
if
you're
not
Finland
or
Norway.
um
where
the
par
where
you're
kind
of
at
par
uh
with
the
US
and
with
China
which
has
like
the
best
the
best
best
energy
price
other
European
countries
are
just
significantly
more
expensive
like
the
UK
is
one
of
the
most
expensive
energy
prices
um
out
of
those
like
developing
economies
um
and
there's
no
like
good
capacity
um
so
Original English
>> you
had
some
good
good
uh
good
stats
uh
where
cost
per
kilowatt
hour
in
Finland
is
0.085
085
in
the
US
0.086,
but
the
most
expensive
cost
uh
is
actually
in
the
UK
at
0.373
per
kilowatt
hour
and
China
is
at
0.109.
Original English
>> Yeah.
And
then
most
recently,
I
think
there's
a
uh
you
know
a
big
deal
with
Microsoft
and
NScale
that
got
pulled
uh
or
at least
Microsoft
pulled
out
of
it
because
the
data
center
provider
couldn't
get
the
energy
connection
in
time.
Original English
>> Great.
So
that's
nimism
and
data
centers.
Uh
so
just
following
your
your
your
thread
while
we
are
at
it
talking
about
Europe
versus
the
US
and
China.
Any
reason
for
Europe
to
be
optimistic?
You're
you're
you're
based
in
the
UK
for
for
anybody
that
doesn't
know
and
invest
globally
but
in
particular
in
Europe.
So
what's
your
current
take?
Original English
>> I
mean
it's
tough.
Like
I
think
the
fastest
growing
companies
in
Europe
are
ones
that
have
US
customers.
Um
I
think
it
is
really
encouraging
that
in
several
modalities
you
know
whether
it's
like
image
with
Black
Forest
or
audio
with
11
Labs
or
video
with
Cynthsia
or
like
you
know
code
app
generation
with
lovable
like
they're
all
market
leaders
and
they're
all
like
born
out
of
Europe.
It's
not
it's
not
impossible
but
I
think
those
companies
wouldn't
be
as
big
as
they
are
if
they
didn't
have
US
customers
and
you
know
increasingly
they
have
Asian
customers
also
because
like
for example
Korea
and
Japan
are
really
leaning
into
AI.
The
challenge
is
like
the
the
the
just
the
gap
between
Frontier
and
not
is
just
so
big
um
and
just
keeps
getting
bigger
and
bigger.
I
don't
know if
these
estimates
are
really
like
important
or
not,
but
like
even
if
you
conceptualize
like
starting
a
new
lab
from
scratch
uh
to
try
to
catch
up
with
openthropic,
you'd
probably
be
spending
like
a
trillion
dollars.
I
think
it's
kind
of
telling
that
even
Mistral
is
like
moving
more
towards
being
an
inference
provider
and
a
cloud
company
than
it
is
really
like
pushing
frontier
models.
Although
of
course
it
has
released
new
models
and
like
they're
kind
of
good
but
they're
still
behind
China.
So
you
know
whether
it's
them
or
other
model
companies
that
have
built
up
GPU
capacity
uh
to
try
to
build
frontier
models
or
kind
of
going
out
of
the
race
and
serving
inference
that
sort
of
figured
out
you
can
approximate
this
this
move
of
uh
of
what's
happening
which
is
to
say
you
either
like
die
trying
to
get
to
the
frontier
or
you
just
live
long
enough
to
eventually
serve
inference
and
you
can
see
this
with
Neolabs
with
Neo
clouds
with
inference
clouds
etc.
It's
like
everybody
wants
to
get
into
token
flow
because
absent
you
having
the
model
that
is
the
best
like
might
as
well
serve
other
people's
models.
Um,
so
I
think
to
get
back
to
the
Europe
topic,
like
that's
probably
the
angle
that
would
be
most
competitive
because
um,
countries
are
going
to
need
to
ensure
accession
or
access
to
uh,
US
frontier
systems
and
they're
going
to
need
to
have
some
kind
of
strategic
leverage
to
bring
to
the
table
uh,
to
prevent
situations
like
what
happened
in
June
with
the
like
basic
export
control
blocking
of
of
Fable
and
Mythos
models
from
Anthropic
to
uh
you
know
non-
US
nationals
or
non-
US
uh
uh
entities
which
you
know
having
been
in
London
at
the
time
like
people
did
like
voice
a
lot
of
issues
about
and
uh
and
so
like
okay
I
don't
want
that
to
happen
again
but
what
can
I
do
to
maintain
strategic
leverage
so
the
UK
believes
we
should
be
investing
and
procuring
inference
chips
because
if
we
come
up
with
like
the
best
inference
chip
um
it'll
be
a
strategic
advantage
for
labs
to
use
it
um
for
cost
reasons
and
and
other
things
and
with
that
I
can
barter
for
access
to
frontier
uh
US
technology.
Original English
>> So Europe would become an inference center and because of access to cheaper electricity in certain parts not the UK but as we discussed then just to play back what you said uh the US frontier labs would be would want to run in France and Europe.
Original English
>> Yeah. Yeah. You can enforce uh data residency requirements.
Original English
>> Yeah. And the the UK chips thing, what is that exactly? Like I heard that several times. Is that something that people are thinking about or is it an actual policy? And if so, how does that manifest?
Original English
>> Yeah. Yeah, it it is it is a policy in so far as like the UK's AI minister who's who's generally I think a a really good individual to be leading this because he's clearly like plugged into plugged into the mix and understands the raw ingredients that go into making AI successful, which is like compute, talent, and data. And so he has a he has a mandate uh to um be a buyer of first resort or new inference uh chips and has committed I think 100 or 150 million pounds something along that that angle to uh to procure infant chips when they are available from new UK companies that are designing these systems. You can see this also in South Korea where they have a similarish mandate but it's actually much more ambitious where every year they're spending about seven billion dollars on AI and a large part of that is going towards building modular inference data centers because generally Korean companies want to have Korean data residency. So I think it's actually a pretty interesting quid proco where it's it's a captive market a small country but it's a wealthy country that's very technologically savvy and has an important role in the overall AI supply chain with memory and so international companies want to at some point in their life like be in the Korean market. Same with Japan but getting in there is difficult. Korean data residency Korean language. So it's like okay cool like you want data residency but now we offer you the data center uh to go run your technology in um and we can profit from that because it's our local companies that provide that that service. So, so you're not a huge uh fan or or you don't have a lot of hope for the whole concept of sovereign AI which is you know I'm asking the question at the time when uh Jensen just uh did a charm offensive and investing offensive across Europe he was in Paris he was in London he was in Berlin for GTC just uh yesterday or a couple days ago but from your perspective that's uh that's that doesn't have teeth
Original English
>> I just uh it's it's a bit like national airlines. The countries will buy planes from Airbus and Boeing and then slap their logo on it. That doesn't make it like yours. It's still someone else's airplane, you know. So, but open source uh could be if you believe that open source is going to remain uh you know within reasonable uh earshot of uh closed source models. Maybe there's hope there.
Original English
>> Yeah. But then they're Chinese. So you have another you have another
Original English
>> it could be American sort of a reflection
Original English
>> but they're not Frontier like none of the US like open ones as far I can tell are Frontier and the ones that have all tried end up working for Jensen.
Original English
>> Okay.
Original English
>> So I think I think his strategy is great. Like I think I think I think different things like do do I do I believe that like buying Nvidia GPUs gets you sovereignty like no. Do I think it's a good commercial strategy for Nvidia? Absolutely. Um
Original English
>> yeah, they sold uh 30 plus billion.
Original English
>> Yeah, something like that. Yeah. And if you look at the national bids um of which like globally there's roughly like 138 billion in sovereign AI committed over the next couple of years and you look at the most popular vendors amongst all of these like it's it's by far Nvidia. I don't I don't think there's much of an alternative. Uh I think it's it's a good it's a good start and you and you need to do it because it is the best technology but I don't think one should kid themselves that doing so gets you sovereignty.
Original English
>> But still in the vein of a of of your recap uh a big moment this year was uh when the US government decided to deny access to Frontier model to non US nationals. So do do non US uh countries and companies have any choice but at least to try? Again, I'm in like a few different minds about this. Uh, one, I think it's like worth trying, but do I think it's realistic?
Original English
>> Um, so I I think I think it's I think it's important um to if you are going to try that they should that countries should try to launch systems which are going to do the white collar work. Like I think in the UK we still we have like a lot of capabilities to go do this um to try to build like an openthropic competitor or like deep mind compet or a deep mind or deepseek or whatever. But actually if you look at the
Original English
>> it's actually an interesting uh fruit and slip or whatever. Right. Because the irony of all of this is that Deep Mind is a UK company.
Original English
>> Yeah, you're right. Yeah right. And I think part of the problem with actually like the UK AI ecosystem um and its its success in actually leading the world is that for too long the UK government kept saying that Deep Mind is a British company and I think it's only started to unwind that like very very recently.
Original English
>> And why why is that a problem?
Original English
>> Well, it's it's a problem because then it believes that it's already number one. Like the UK is the best place to do AI and we are the best at AI because we have deep mind
Original English
>> so therefore we don't
Original English
>> the attitude behind the claim which is a problem.
Original English
>> Yeah. Yeah, because then in like you know in policy rooms then you know individuals who are advocating for more investment in AI or we need more u you know we need to procure more systems from AI companies or we need to make our country more appealing for AI companies to start get answered with like but we have deep
Original English
>> mind right
Original English
>> you know
Original English
>> right
Original English
>> um
Original English
>> what about defense AI you're you're an investor in defense companies I mean that that feels like an area where uh the need for sovereign AI since obviously the warfare of the future is going to be very driven by AI. That feels like a place where
Original English
>> Yeah. Yeah. I think it's Yeah. Yeah. So, yeah. So, that that that push really started uh I mean obviously like post like the Ukraine war like almost 5 years ago at this point but but more importantly I think it was like two years ago when uh when JD Vance went to the Munich Security Conference and effectively said like you need to invest in your own defense and get serious about this. You being uh you know NATO and Europe. Um and so ever since then there's been you know a variety of different initiatives to uh give more capital to new defense companies to encourage governments to be providing loans to businesses to improve their supply chain. Um and that's you know created a a set of like neopimes I like new defense companies that are trying to compete with prime contractors. Uh so companies like Stark or Heling um being quantum systems being like the the main ones in Europe and then uh you know as like warfare has changed like it's gone from uh I don't know you know developing strike systems over the over sort of mid-range to you know hit tank uh you know battalions to now being like we're getting pummeled like every day with hundreds and hundreds of FPV drones or various other kinds of aerial systems that are autonomous and you know now can no longer be stopped with jamming because they use computer vision or they're guided by these like um fiber optic cables which by the way have gotten much more expensive because of data centers and so we need more uh you know novel like air defense systems and so that's where companies like Altar in France um and others have really been pushing
um and uh and as you said like the supply chain is super important because for example like you know NATO wouldn't procure uh systems in in Europe that don't have at least 70% kind of the parts from uh from you know non-China basically um and uh and and countries are really queasy about taking US um protection or US guarantees at face value because they're shifting on a weekly basis.
So, so yeah, I think that's an example of a market which is sort of has to be domestic by
Original English
by design and can breed new companies that are much more competitive than their US peers.
Original English
>> So then if Frontier AI cannot be built outside of the US, then let's let's go back to the US. a part of your report on on research effectively calls entropic open and Google as as as winners as like the definite winners which obviously uh you know has been uh years in the making but uh there was a moment where it felt like there was a plurality of labs effectively what you're saying is that no like those three are the the the winners
Original English
>> and does that include uh I guess nobody knows what they actually do but like SSI and thinking machines and like there seems to be like hundreds of neolabs
Original English
>> yeah yeah maybe meta on margin.
Original English
>> Yes,
Original English
>> super intelligent lab.
Original English
>> Yeah, especially as now they're pushing towards like serving their models through API but also having an enterprise business.
Original English
>> Um, so that one, you know, that might change over the next like 6 to 12 months depending on how their go to market works. But yeah, I would say it's probably those three plus meta.
Original English
>> So your view is that the the advantage is going to keep compounding and uh that's going to be the state of the market versus
Original English
>> I think so for like the general purpose frontier. Now there's there's like uh examples of perhaps like a really focused uh uh bet on a on a particular capability which has like where you can make certain trade-offs to the model in order to get much better performance at a task like I'm referring to for example these classification models effectively uh like Jev which can grow really really quickly and become popular again TBD how long it lasts but I think it's I think it's an encouraging sign for entrepreneurs that shows that you know you can take a focus features bet and it can pay off but over time. It's again like TBD if there are like these kind of gaps and capabilities that are sufficiently exploitable.
Original English
>> So what do you think happens to the dozens of Neolabs? Uh I mean not not to keep talking about Europe but like just in London there's three of them right there like the David Silver thing and like another two the you know in the US every few weeks uh there's one you know Jeff Dean's new lab and so on and so forth. So what happens to those?
Original English
>> Well I think it depends on the their kind of focus and modality. Like for example um on on robotics on embodied AI I think you can I I can see how an independent lab would still have a chance of competing and winning like because of the immense focus on uh you know enterprise AI basically that open AAI and anthropic has like I have a slide in the report where I basically just like show all the different product lines that open anthropic had admittedly open AI had lots and how they progressively just cut all of them and there was a quote like you know this uh this moment is like too important to be distracted by side quests or some version of that. I'm paraphrasing which meant that they broadly don't do dedicated robotics, although maybe there's some of that kind of data in in the Astro training recipe. Science is a little bit unclear like anthropic is clearly leaning a lot into biology and drug discovery. Open AI maybe more on the model side, less on like we want to own more of the stack. there's like kind of related to embodied AI this whole like category of world models whether that's you know a video model with some action in it or like some other definition of world model but um but in general like anthropic doesn't really do any of that or at least like from what we can see publicly I mean they don't even have an image generation model for example I think audio pretty clearly labs do have some of those capabilities but uh you know if 11 labs can like keep accelerating the way it does then clearly there's there's a need for a dedicated capability I think those are probably the areas where I can see an independent lab like working out, but but ones that broadly say we're going to do frontier intelligence for knowledge work, but just like make it better, faster, cheaper. I think that's tough. I think for me, like I I'll believe it when I see it. Um but but yeah, this idea of like you open your computer, the app looks exactly the same, but it's just better. I I don't know if that
Original English
>> And in world model labs, are you seeing anything exciting? I mean know obviously FA Lee's lab was
Original English
>> acquired for a lot of money and that's unclear what happens from there given that the buyer is AMD I guess Khan started his company AMI in Paris it's a world
Original English
>> labs yeah yeah I mean I' I've uh you know visibility to Odyssey where was an early investor like doing world models and I think the the interesting thing in that business is like their most latest their latest systems like Odyssey 3 and there you can see uh that uh autograph aggressive model uh trained to basically take all this video data um and then have it conditioned on actions such that you know that okay I'm taking one action and then I generate the next frame as a result of the action that paradigm uh applied to robotic manipulation to self-driving to like video games to like entertainment or something can see generalization across all those different domains and already seeing that model being used for robotic manipulation training
Original English
So I I think at a broad level I'm like kind of interested how image, audio, uh video ends up all mushing together into basically a world model. And then if that actually happens, then that could end up subsuming simulation and game engines and uh and then that's very important for moving robotics into the real world, but also designing physical systems and and like hardware engineering. And to get back to like the AMD World Labs point, it's like if if that happens, then the compute substrate that serves those kinds of environments and allows you to train and run inference is super super important. And therefore having knowledge about how the world model research ecosystem is moving and what kind of hardware improvements you can make to be the number one choice is a big big revenue pie. That's I think what motivated part well part of the reason that AMD uh bought World Labs. um sort of in the same way that Nvidia over many many years was like very close to research uh labs that are doing shaders or 3D vision or other things and you know giving compute out and constantly getting feedback from researchers such that they always knew like my hardware is the best for where the where the kind of frontier is going.
Original English
>> There seems to be a good number of uh labs trying to figure out ways to be more efficient with compute. So from flapping airplanes which I don't know much about uh to what seems to be the reemergence of like diffusion model focused companies.
Original English
>> Yeah. Yeah.
Original English
>> To just reduce the cost of of inference which uh is a fascinating fascinating thought by the way which uh you know people tend to talk about a bunch for a minute and then forget about but like this massive investment in compute that we do in theory seems to be at the mercy of like somebody coming up with a much more efficient way to train models.
Original English
>> Yeah. Yeah. Yeah, I mean I would contend that even if that were to happen, we would just see like more consumption and um you know I'm just informed by a fact I'm on my laptop using Astro High on 1.5x and I would love that 1.5x to be like 10.5x
Original English
>> paradox. Yeah, just like giving more chips. Yeah, I'm just like doing more things than I I could have done before with the same amount of time. So I'm not necessarily sure that like you know one super efficient system will sort of like pop all the investments that we're making.
Original English
>> In your report you talk about the important intersection between models and and harnesses which uh seems to be one of those topics that like keeps shifting. People say well no the intelligence is all in the model and then other people say no it's all in the harness and that seems to be going back and forth. What's your take? I think my broad take is like it's a bit like uh when you're injured and you have crutches, you need to have assistance and then at some point you heal and you don't need the assistance anymore. In absence of having like a frontier model like you need to give it
Original English
crutches and those crutches can be like access to tools like how it handles memory, what data sources it has access to and then maybe some recipes uh in order to accomplish a specific task. And I I think for the moment like at the non-frontier that that'll hold. But I'll take the word for I'll take the sort of frontier lab folks word for it that you know over time one should assume that all these crutches kind of get subsumed into the model through training or through the harness that the model vendor provides.
Original English
I think there's there's a data point around um how much of the clawed code system prompt was rewritten uh over time as the model itself got better and it's something like 80% of the system prompt was just removed.
Original English
The harness is also important because um the one that's shipped by open a anthropic is kind of like not incentivized to make you use a less powerful model. Like just in the UI and the and the product like if I'm doing a legal task and I have the $200 plan and I have the tokens like I want the quality to be super good. So why would I risk the quality being like not super good by sliding it from high to low or by trying it on a lower quality lower quality model? And so I think unless there's some some sort of like product UI behavior that is altered um to disincentivize the user from only using the most expensive tool. The Frontier Labs will continue doing that. Um and so that's maybe where like getting getting good knowledge around like okay can use a less good system but compensate for it by having a better harness can come into play. But I think that's more of like a product a product decision. Um, also for companies that that are not like the Frontier Labs.
Original English
>> Is that interesting for you as an investor? Uh, all the kind of vertical harness companies or horizontal harness companies. Is that an investable theme?
Original English
>> Yeah, I think in in isolation like no at the moment. Again, I can change my mind, but at the moment, no. But I do think for for my companies that are you know heavily like AI based or heavily agent based when we see that you know every month like token expenditure because like is going up through the roof because employees are increasingly getting leverage from these tools and uh and product usage is going up like that's where I think the trade ends up being like hey we should really like spend time on seeing can we use like a less frontier model and like compensate for it with a better harness uh so we could just like not burn a lot of money.
Original English
>> Yeah. By the way, you're a solo GP and I think I read somewhere that you use a lot of agents uh in your investing activity uh to do everything you do. Any example or anything uh that you thought was pretty cool in the way you use agents throughout this year?
Original English
The coolest use case I had recently is on like expert research where you know I think you can well if you're sort of in the cheap man's version it's like you ask your friends but if you want to spend a bit of budget and ask beyond your network then you're going to ask like GLG or Alpha Sense etc. Um and then they will go find humans for you. They will introduce you. You take the call. It's a pingponging like really annoying and limited by language time zone etc.
Original English
But there's like a a new product that I started using called LT which is basically like replacing uh the fact that I have to do the work with the model doing the work. And so I'm already like heavily using in this case Codeex, but I kind of switch depending on what model is better at the time. uh to sort of like just process all my ideas and the context I have on other companies that I looked at over history, how they performed, what questions I asked, what responses I got to just like figure out what is the highest value question I can ask of a team to inform like whether I should spend time or make the investment or not. So I can take all that information and then and then with LTO like construct uh an expert interview and then decide on cohorts I want and then just hit launch study and then the the system goes out and finds people consents them to see if they're uh you know the right fit for the study and then they go speak to an AI
Original English
and I was I was like shocked because you know I hit launch study I went to sleep I had like 3 days to basically like give the company an answer it's like in competitive market so I think the nuances of customer research was important you know, wake up the first day, I have three interviews from like pretty senior people. I have the raw transcript. The transcript can come back into my model, answer the open questions we had, we me and my model, and then uh and then figure out, okay, we have enough data points for this, but like it's kind of opened up this question stream that we should ask. Maybe we should actually reune some of the interview questions to really coax the person to answer those questions. So, like updates the study questions and relaunches the survey.
Original English
Next day, five five people. Again this repeats and over like two and a half days I had 12 in 12 12 interviews over like 14 hours across like US, India, Europe, Germany, France. Like it would be physically impossible for me to have spent like 14 hours across 12 people. Like there were several people who spoke to this model like for an hour and 20 minutes
Original English
>> which just in itself is fascinating, right? Like the fact that humans feel so comfortable so quickly speaking to an AI.
Original English
>> Yeah. Yeah. It's probably a bit recursive in the sense that like we're all in front of our computer all the time and like these are generally like tech questions and so I think people prefer to unfortunately people prefer to
Original English
>> speak to their phone than speak to a human.
Original English
>> Um that's that's a fascinating thought that actually speaking to NI is uh what people prefer.
Original English
>> Yeah. uh which is actually a theme that we heard over the years is like uh you know we we uh maybe not the years but the months that uh people are starting to prefer to talk to an AI customer service agent uh over a person.
Original English
>> Yeah. Because I think in that case it's like uh a human is pretending to be a robot and doing a really bad job at it because you're a human. You're like can you please just like now talk like a robot and then now you get a robot that's actually like really good and then the human
Original English
>> talk like a human. Yeah. Okay. Amazing.
Original English
>> Yeah. Yeah. So that that I think that that kind of idea of like uh automated customer research uh has been like very powerful for me and I can just see how you know even if you had a portfolio you would want to run like this kind of agent interviews on your portfolio companies and how customers are using it. So then at least you get some firstparty data when you go into the boardroom you know or if you're an LP and like your investors are saying like this company's really good you're like is it really good?
Original English
A big uh thing that happened this year uh in the last few months uh in particular is to this point like the rise of personal agents, the instincts and the muse and the rock bots. Um
Original English
>> what do you make of that?
Original English
>> I don't give them my data.
Original English
>> I don't really I don't really trust trust them at the moment. Um but I think a lot of people do. Um
Original English
>> a lot of people do.
Original English
>> Yeah. Um,
Original English
>> so what you're saying is that you not you don't give your credit card information to a company that did not exist uh uh 2 months ago and uh doesn't require you to authenticate.
Original English
>> Yeah. Yeah, pretty much. Well, also because I feel like I can do roughly the same thing um with Chad GPT like I don't actually see like what I get on top. Codex and Claude already have like low-level access to my computer has access to my browser to like various messaging apps and things like this. So I I I trust that already. So I don't really want to increase like the the risk on top of that. But but I do think it brings up an interesting point which is like which which vendor do you trust the most for these kinds of tasks. And um not that long ago I heard like Nikolai at Revolute is uh is now like thinking okay well I I have like you know I don't know how many customers but like millions or tens of millions or
Original English
something. Um, and they clearly trust me with my wi-i w with with my money and my credit card data. And, you know, now they have like an AI assistant in there. And so I could sort of see myself trusting Revolute more to handle purchases on my behalf than I would like some third party startup that's grown in 12 months and probably doesn't have correct infosc and and other things.
Original English
>> So you think it becomes a feature of broader platforms versus the opposite?
Original English
>> I think that's that's one possible route. you know whether the platform ends up buying one of these teams is you know that that can always happen but I heard you know um another point would be somebody prefers to use like Meta's muse because the likelihood that one of these startups survives for 10 years and if they do something bad that you can sue them and get your money or get some money is very low whereas like suing Meta like they have a a history of like paying you when there when there's some issues. So, I've heard that's like a reason why some people prefer to use meta. Think at least I know who to pick up the phone to.
Original English
>> But but those those um uh new personal agents seem to be playing a a big role in the diffusion of AI though, which is one of the themes that you highlight in the report as well. You talk about the increasing rise of uh codecs in particular beyond coders.
Original English
>> Yeah. Yeah. Yeah. Exactly. Yeah. And I think the the biggest message there would be, you know, at the same time, you see this like recursive self-improvement. You see people doing all sorts of crazy stuff with their AIS and that's like a 0.1% or something of the AI usage base and the vast majority of other people are barely paying for any AI and so getting like a very subpar experience. Uh and part of this is just like like bad product, bad marketing. Um, also just that you know a year ago or two years ago there was all this like narrative around you need to become a prompt engineer. Like you really need to understand how to write these prompts because if you write crappy prompts the model does a bad job. So I can only imagine how many people who don't work in technology saw that and were like I have no idea how to learn this. Like this is way too complicated. You know it's kind of like writing code but in the human human language. Like screw this like I'm done. Like AI sucks. and and now it's like do you really need to be a prompt engineer? Like I think you more just need to be good at just like talking to your phone just like brain dump exactly what well like everything you're thinking of and the system can organize itself. So I think this like prompt engineer thing is kind of gone like the model can has enough affordances to not need that. So now it's much more easy for a normal person to use but the normal person got sick of it during the prompt engineering days and hasn't come back. So, you're right that like creating these on-ramps with like um personal agents would be a great way to get people back into the system and and and you know maybe maybe that's what they understand AI to be no longer like chat
Original English
>> and to to unpack a couple of things that you that you just said were alluded to. So there's this idea that there is an increasing gap uh between people using AI and nonAI and you have a a slide that talks about how Codex adoption remains far far higher inside of OpenAI than amongst external users.
Original English
>> This is wild.
Original English
>> Yeah.
Original English
>> Yeah. I mean the the broad stat is like something like 100% of people in OpenAI use codeex and then of their organizations that use it. something like 2/3 maybe um something quite high.
Original English
>> Yeah. 69%.
Original English
>> Yeah. Okay.
Original English
>> A separate analysis of S&P 500 companies AI disclosure found that 69% reported live deployments but only 2% disclosed metrics tracked over time and that's in contrast to 97.9% of active open eye workers use codeex in the preceding 28 days.
Original English
>> Yeah. Yeah. Yeah. And individuals is like less than 1% use codeex. And then of the people who do use codeex, the vast majority of them, it's like 85% do tasks on it that would roughly take like 30 minutes and only like 25% I think do it do tasks that would take like 8 hours or more.
Original English
>> So the conclusion of this is that the the people that get it are just accelerated further and further whereas the people that don't comparatively slow down.
Original English
>> Yeah. It's basically like you're you're kind of going through like a theme park or something and you just you discovered like level one and there's like this entire like other world that you just haven't even opened the door to
Original English
>> and there are a bunch of people that are already in that world.
Original English
>> It's like 9 97% of the world or something which is which is insane. So that that's where I was like thinking okay there should be like a genius bar for AI or like a geek squad for AI or something you know like I find it shocking that you know we're in New York like we have amazing Apple store like all these concept stores for brands. Why does OpenAI not have like an OpenAI store?
Original English
>> And what would you do in the open air store where you'd walk to a genius person and they would walk you through use cases?
Original English
>> Pretty much.
Original English
>> I mean, uh, it would take time, but I think, you know, people end up showing other people, you know, with these kinds of viral products. Um, but I could like, you know, very easily see like you'd run sessions on uh, you know, different levels of experience and like I don't know, maybe one of them is for like small medium business owners, maybe some of them are for like, you know, busy families, like other ones are for like healthcare providers, like you know, anything you can imagine, but where you're really walking through people, walking people through what you can do by just like flicking the screen and going on work or going on codecs. And I think that would make a a big difference because it would it would make it would make the technology much more empowering and less about like wow there's a lot of people out there that are doing things with this that I just don't understand and I don't have like the the time in the day or or the valition or like patience to go learn it myself. It's a bit like uh you know everybody should do exercise but like most people don't and you need to hire a PT to just go do it.
Original English
>> Mhm. And uh to finish the thought on the other part of of codeex, you have a slide where you said that non-developers are growing usage of codeex faster than developers are uh and from August 25 to June 26, individual non-developer users increased 137fold. So those are no longer really coding agents. Uh they are agents that code but for like a broader group much beyond coders. Yeah, it's kind of unclear because um like they could write code to go execute your knowledge work but the individual who's using the system doesn't doesn't really see it. So, I'd say I'd say that that result is a mix of really simple things like Google Drive connector that actually works like it's kind of nuts that the default uh you know Google Drive connector like for example can't do a lot of things that you would want it to do and you have to go to this third party like GWS CLI which was written by Google engineer who got fired for doing it but like if you don't use the CLI version like your agent has like really really high level to Google Drive high level access to Google Drive which precludes many many workflows.
So just basic good integrations and then the second thing is this like recursive cycle of uh models labs releasing a model data companies or companies that have a certain workflow that they're trying to implement the model uh on showing that on their internal benchmark said model doesn't do very well but they have the environments and the data to make it do really well. And then the lab realizes hopefully that they want to hill climb on this capability and so they buy the data improve the model performance and then this recursive cycle continues and uh I think ever since like the launch of codeex and then uh and uh before that actually like cloud code with those connectors for legal for marketing etc that was kind of this approach of like let's build the integrations to the tools that the model needs to be good at that
Original English
kind of human like white collar work uh let's do hill climbing on certain and workflows that are like important for finance legal marketing and then um like let's release those those products with a bit better marketing a bit better case studies uh starting with our own staff and then there's been this constant like debate of like do we call it work do we call it code which slider should it be in and I think you still haven't really figured it all out but but now I would say like it's it's like so powerful
Original English
>> and u speaking of hill climbing we we talked about open source we talked about the big labs. Um it um we talked about sovereign AI as well. It seems like there is this uh emerging theme around enterprises and before them probably the vertical AI startups taking control over uh their AI versus solely depending on anthropic open AI. Is that um an important theme for you and something that is investable?
Original English
>> Yeah. what's um I think that that trend has been growing and it's a mix of uh cost uh optimization um like sort of data sovereignty data residency uh issues. So not wanting to like put all this information into a third party system which you may or may not trust the privacy policies of. And then three, uh having a certain like set of use cases that are the real ones you care about and that you kind of like don't want to wait for the model lab to start caring about them to get better. Uh and then maybe last one would be uh yeah some of these um like more nationalistic tendencies. I guess I think in that in that area uh what I'm what I'm wondering is like who's best place to deliver that service. Of course, like one of the biggest ones would be Palanteer, which which done partnerships with like NeoClouds uh and has all the data infrastructure already and so it's like well positioned over the last few years to like start providing these services that probably works in US commercial not so much in Europe. There are like sort of like these neo palenteer style companies in Europe that are like you know started up um to go do the exact same thing with a different like nationalistic flavor. Then the third bit would be like what about the inference companies um that have been serving open source models and that maybe are well positioned to not just serve but also provide tools to look at the traces of how their models how these models are being used in an enterprise use the the data of uh these conversations to then provide like a post- training service to close the loop. maybe they're like well positioned or maybe it's like the Neoclouds themselves who have started to build inference businesses on top that would be well positioned too or or maybe it's even like the data companies um that have not just or who have started to move outside of providing uh our environments for labs but saying well your company's in an environment and so uh you know we've proved that we can do it well for labs maybe we can do it well for you.
Original English
>> Do you want to finish with a rapid fire? Let's go.
Original English
>> Most overrated trend of 2026.
Original English
>> Uh home cleaning robots.
Original English
>> Those don't work.
Original English
>> No.
Original English
>> Why is it that uh everybody uh wants that specific use case of like folding laundry? Is that is that because it's uh
Original English
>> I don't know. It's it's it's a bit like you know the the the demo of like agents booking flights. I I I heard like very recently that now like San Francisco is like starting to say like can you show me a demo of like not booking flights. I'm like thank god I've been saying this for like 10 years.
Original English
>> All right said rapid fire and then I ask you a follow on question. So um most underrated slide in your own report
Original English
>> the amount of money that you would have made investing in Nvidia instead of investing in alternative semi semiconductor companies.
Original English
>> Uh unpack that quickly. If you had invested all the money that went into alternative semiconductor companies into Nvidia on the dates that those rounds were announced, you'd have made four times more money.
Original English
>> Yeah. And then to close the prediction from the past eight reports that you are the most proud of.
Original English
>> Oh. Uh that Nvidia would not conclude the acquisition of ARM. That was that was a big one back then. I mean people thought it was really going to happen and like we took the view that like for geopolitical reasons and competition reasons that it would not happen and to some degree like ARM shareholders made a crap ton of money by not closing that acquisition to another degree like Nvidia would have had like the most impressive CPU business that could have been really accreative to their uh to their cloud uh particularly as like the rise of agents and RL and everything else that actually needs uh that's kind of CPUbound and uh and I think that could have also like shook up a bit bit of the geopolitical dynamics too.
Original English
>> Okay, wonderful. That's a great one. Uh Nathan, thank you so much for coming back and uh thank you for doing again this state of AI report which is an extraordinary amount of work that uh you and hopefully your agents uh now do every year. It's available at stateofoff.ai and uh you have a you have a video that uh accompanies it. That's also a good resource. And uh people can find you on Twitter.
Original English
>> Yep. Yep. Nathan Benes or at state of AI. Um and uh yeah, thank you so much for the opportunity. It's always a pleasure and uh I mean it's bananas what's happened in the last year alone. So let's see AGI next year when I'm in New York.
Original English
>> Wonderful. Thanks Nathan.
Original English
>> Thanks.
Original English
>> Hi, it's Matt Turk again. Thanks for listening to this episode of the Mad Podcast. If you enjoyed it, we'd be very grateful if you would consider subscribing if you haven't already, or leaving a positive review or comment on whichever platform you're watching this or listening to this episode from. This really helps us build a podcast and get great guests. Thanks, and see you at the next episode.