00:00:00
who's your number one Who's your number
00:00:01
two if you had to could only bet on two
00:00:03
here to achieve super intelligence AGI
00:00:06
let's just say win the AI re win the AI
00:00:10
uh big prize the big prize super
00:00:12
intelligence AGI you know in the midterm
00:00:15
five years five years from now we're
00:00:16
sitting here Thomas give me your number
00:00:18
one give me your number two look I I
00:00:20
think to me number one I I still think
00:00:22
Nvidia right I don't see the GPU kind of
00:00:24
getting displaced I see additional
00:00:26
architectures kind of coming on board
00:00:28
right and growing the market but um at
00:00:31
the end of the day all roads still lead
00:00:32
to the GPU for all of these models So I
00:00:35
would kind of still put um kind of
00:00:37
Nvidia on that My number two more of a
00:00:41
dark horse but I I would pick Tesla
00:00:44
I do think it has the most potential for
00:00:46
vertical integration right From all the
00:00:49
way the silicon to the model to actually
00:00:52
the hardware right That might become
00:00:54
super important not just in cars but in
00:00:55
Optimus So Nvidia 1 Tesla is my dark
00:00:59
horse Wow stunning Chimath who's your
00:01:02
number one and number two In the midterm
00:01:04
5 years from now we're sitting here on
00:01:05
AllIn episode 700 Tesla's one and
00:01:09
Google's two And the reason is because
00:01:12
they are the closest to having that
00:01:15
vertically integrated stack that I spoke
00:01:17
about I think that Tesla has the best
00:01:20
vision models Now with XAI they'll have
00:01:23
one of the best LLMs and reasoning
00:01:26
models and they'll be able to eventually
00:01:29
stick that on Dojo And then all of that
00:01:32
will be in all of the physical AI that
00:01:35
you will interact with in your daily
00:01:37
life whether it's a robot or whether
00:01:39
it's a car or whether it's a robo taxi
00:01:42
So that's number one And then number two
00:01:44
for many of the same reasons I think
00:01:46
Google because you'll have the Gemini
00:01:49
family of models which just absolutely
00:01:51
kickass like VO3 which we haven't really
00:01:53
spoke about is going to destroy
00:01:55
Hollywood like in the next year Like
00:01:58
Hollywood is done I think but they're
00:02:01
landing model after model They have the
00:02:04
TPU and the next generation TPU I think
00:02:06
is exceptional They're baking quantum
00:02:09
and then they have an entire funnel of
00:02:11
billions of people that they can direct
00:02:13
experiences to So Tesla one Google 2
00:02:16
Chimath quick followup on that I'm
00:02:18
curious on Google This is the because I
00:02:21
I oscillate a lot on this particular
00:02:23
name Can Google win if search declines
00:02:26
Yes And I think that what probably has
00:02:29
to happen is bear with me when I say
00:02:31
this but if you had to boil down
00:02:34
Google's economic northstar metric right
00:02:37
not the value northstar the economic
00:02:40
northstar metric would be price per
00:02:42
click
00:02:44
And I do think that Google is extremely
00:02:46
well positioned to pivot that to price
00:02:48
per token And I think that they have
00:02:51
some emerging classes of physical AI but
00:02:54
they have the largest pool of people
00:02:56
where they can generate a price per
00:02:58
token value framework through YouTube
00:03:01
through Gmail through Workspace I think
00:03:04
through search but probably it's a
00:03:06
different kind of model It just requires
00:03:08
them to rip the band-aid off at some
00:03:10
point But yeah I think Google can do it
00:03:12
I'm going to go with you Chimoth I'm one
00:03:15
uh my one and two are either Google uh
00:03:19
or Elon And I I'll just say Elon because
00:03:21
I uh like you I spent a day up at um XAI
00:03:24
and I saw what a magnet for talent he is
00:03:27
I got to sit in some meetings and just
00:03:29
he was interviewing people and he was
00:03:31
working with that talent 8:00 at night
00:03:33
There's a lot of people there on a
00:03:34
Saturday grinding it out It was nuts I
00:03:37
first went to XAI
00:03:39
In the 15 minutes that I was in the
00:03:40
parking lot finishing a call the kinds
00:03:42
of people that were walking in and out
00:03:43
of there you could tell they were big
00:03:46
brains Yeah I don't know how you know
00:03:48
what I mean Like from every walk of life
00:03:50
they all just looked much smarter than
00:03:52
the rest of us Yeah It some of them were
00:03:54
like chain smoking cigarettes and just
00:03:56
like stressed out It was crazy I hit a
00:03:58
couple of zins I'll be totally honest Um
00:04:00
but the reason I say Elon versus Google
00:04:03
is I think Elon's in a unique position I
00:04:05
don't have any insider information here
00:04:06
and and I haven't talked about this or
00:04:08
I'm not back channeling from Elon lest
00:04:10
anybody aggregate this I think what
00:04:13
Colossus has done and what Tesla has
00:04:16
done both of these things Tesla with
00:04:17
their own stack of hardware to your
00:04:19
point Shimoth hardware plus software
00:04:21
plus the user application of FSD and
00:04:23
Optimus Then you put that together with
00:04:25
the data the real-time data of X
00:04:28
formerly known as Twitter plus um you
00:04:31
know what he's building with XAI and
00:04:32
obviously those two companies merged I
00:04:34
think Tesla board XAI board have to get
00:04:36
together put those two companies
00:04:37
together One's worth a trillion one's
00:04:39
worth 100 billion Put them and just have
00:04:42
all that brain power going in one
00:04:44
direction as opposed to Elon T switching
00:04:46
between the two You do that I think he
00:04:48
wins number one you don't do that I
00:04:50
think he either gets one or two and then
00:04:52
I think Google um is going to have a
00:04:55
better search product Thomas I think
00:04:57
it's a really important point Do they
00:04:58
lose search share Doesn't matter What I
00:05:01
think matters is are their ads more
00:05:02
effective Is their ad network more
00:05:04
effective And I think based on what they
00:05:05
know on you from your chat searches and
00:05:08
your discussions and what they analyze
00:05:10
in your email just analyzing your Gmail
00:05:13
and your surfing behavior and Chrome if
00:05:14
they get to keep it your Android phone
00:05:16
if you use it your YouTube list and what
00:05:19
you how when you drop off allin and when
00:05:21
you start listening to another podcast
00:05:22
whatever it is all that data all that
00:05:24
data is going to lead to an ad network
00:05:26
that performs so much better that even
00:05:28
if they lose search hair their ad
00:05:30
network is going to continue to grow and
00:05:31
I think it will increase in velocity So
00:05:33
those are the my top two Freeberg I'm
00:05:35
curious from your position
00:05:38
which one you think is number one and
00:05:40
number two I saved you for last because
00:05:41
you know what we do here We save the
00:05:42
best for last Free part go ahead I think
00:05:45
there's a difference in how I would kind
00:05:48
of lump them I I think that Tesla
00:05:50
probably has the it is the best place to
00:05:55
invest if you want to have a shot at a
00:05:58
massive new industry So they've got a
00:06:01
baseline business in in obviously the
00:06:03
automobiles but I think this humanoid
00:06:06
robot opportunity is absolutely
00:06:10
mind-blowingly ginormous And I don't
00:06:12
think that there's a better company on
00:06:14
Earth positioned to execute against this
00:06:17
humanoid robotics opportunity than Tesla
00:06:20
So you know it's sort of like I would
00:06:22
call it a low probability high upside
00:06:25
sort of call option embedded within that
00:06:28
business And obviously you're paying a
00:06:29
premium for that because it is still a
00:06:30
very healthy premium you pay for that
00:06:32
business I think Nvidia to Thomas's
00:06:35
point I think the common thesis is it is
00:06:38
the most protected The durability of the
00:06:41
business is there But I would argue that
00:06:43
there's actually a low probability but
00:06:44
very high severity risk to Nvidia in
00:06:47
China There was just a demonstration
00:06:48
last month of a 1 nanometer
00:06:51
semiconductor manufacturing process out
00:06:53
of China I think the more that we
00:06:56
continue to try and isolate China from a
00:06:58
policy perspective the more we are
00:07:01
emboldening investment in China meaning
00:07:04
from the government from private
00:07:06
industry into China to create
00:07:07
alternatives to the chip stack where the
00:07:10
United States companies particularly
00:07:12
Nvidia have emote today So I do think
00:07:15
that there's going to be an emergent
00:07:16
competitive threat coming out of China
00:07:18
to Nvidia And just like we were knocked
00:07:20
over by DeepSeek I think we will be
00:07:22
knocked over by some semiconductor
00:07:24
manufacturing processes um coming out of
00:07:27
China in the near term But the overall
00:07:29
kind of by the way Dave just on that
00:07:31
point I think Sax's work on the
00:07:34
diffusion rule just generally I don't
00:07:36
think has kind of gotten enough
00:07:37
attention in the rescending of the
00:07:39
diffusion rule which essentially
00:07:42
handicapped our ability to even arm our
00:07:44
allies right with our sem with our
00:07:48
semiconductor technology um in my
00:07:50
opinion was kind of a milestone and very
00:07:53
important moment um to to try and offset
00:07:56
exactly what you were just describing
00:07:58
That's exactly right I mean there there
00:07:59
there was a report a few months ago and
00:08:01
I mentioned it on the show or maybe I
00:08:03
didn't or maybe I sent it to Sax and we
00:08:05
talked about it offline I I can't
00:08:06
remember but it was about a $40 billion
00:08:08
investment being made in developing
00:08:10
competitive semiconductor manufacturing
00:08:12
full stack solutions out of China So I I
00:08:16
do think that the lithography IP moat is
00:08:19
being crossed in China I do think that
00:08:21
China is developing actually new
00:08:23
technology for uh DUV and EUV systems I
00:08:27
I do think that there's a risk uh to
00:08:29
Nvidia's core Now look Nvidia is such a
00:08:31
durable business There's great modes
00:08:33
great advantages but we're creating
00:08:34
every incentive for an alternative to
00:08:36
Nvidia to emerge from China And then my
00:08:39
my third kind of categorization would be
00:08:41
what's the portfolio uh solution I think
00:08:44
that's Google I think that there's a
00:08:46
diversification of high beta bets inside
00:08:49
of Google of any one of which could have
00:08:52
call it a trillion dollar market cap
00:08:54
outcome ranging from Whimo to quantum
00:08:57
computing to the biologics work that uh
00:09:00
Demis is working on out of um isomorphic
00:09:04
Uh there's a number of things that do
00:09:06
not get a lot of attention at Google So
00:09:08
yes there's a there's a core business
00:09:10
that that may be at risk Thomas but I
00:09:12
think that there's a a portfolio of
00:09:14
options you get at Google and you just
00:09:17
need any one of them to hit to kind of
00:09:19
make up for the loss But I do think also
00:09:21
Sundar in my interview with him which we
00:09:23
put out a couple of weeks ago is very
00:09:25
thoughtful about where search evolves to
00:09:27
and he is being I think reasonably
00:09:29
aggressive in in trying to evolve the
00:09:31
search product architecture to meet the
00:09:34
market to meet the consumer I do give
00:09:36
him credit for that So Google would be
00:09:38
in a good place for me as an overall
00:09:40
kind of pick in that set of options So
00:09:42
just to be clear Nvidia one Google 2 or
00:09:43
Nvidia Tesla Like I said I think in
00:09:46
terms of like having the right sharp
00:09:47
ratio is how I would think about it The
00:09:49
alpha and the beta adjusted returns I
00:09:51
would put Google number one I would
00:09:53
probably put Tesla Uh Tesla's valuation
00:09:57
I think already has a premium associated
00:09:59
with those options So I don't know Yeah
00:10:00
So I don't know if I would really pay
00:10:01
that premium I think um aside from the
00:10:04
valuations let's take valuations out of
00:10:06
it Just the the game here is who wins
00:10:09
the AI prize 5 years That's how I
00:10:11
understood it as well Yeah So valuation
00:10:14
irrelevant Valuation irrelevant Who wins
00:10:16
the AI prize One you're saying Google
00:10:18
Two you're saying Tesla I think Google's
00:10:20
in such a position I I mean look Deus uh
00:10:22
Demis I think has been fairly koi about
00:10:26
where they are They obviously promote
00:10:27
Gemini 2.5 but there's a lot still
00:10:30
coming And it's and and as Chimath
00:10:33
pointed out it's not just LLMs There's a
00:10:36
pretty sizable family of models
00:10:39
including a a lot of these um
00:10:41
graph-based models that are being used
00:10:42
in really novel applications that no one
00:10:45
else is even close to no one's spending
00:10:47
time on I mean some of the weather
00:10:49
forecasting it might seem small and
00:10:51
trivial but it's a demonstration of
00:10:53
Google's competency in in core model
00:10:55
development that shows an understanding
00:10:58
and a depth of research and work that
00:10:59
goes well beyond LLM So I'm pretty
00:11:02
bullish on the depth of talent the full
00:11:04
stack Yeah And whatever they learn there
00:11:06
could apply to Gmail could apply to
00:11:08
search could apply to ads could apply to
00:11:09
YouTube algorithm right It's just goes
00:11:11
up and down Yeah Yeah From a product
00:11:13
perspective I do think you see this kind
00:11:15
of multi-model emergence that that we're
00:11:18
now seeing that no one talks about the
00:11:20
single model that sits behind the
00:11:22
application There are multiple models
00:11:24
that work together And obviously this
00:11:27
agentic architecture unlocks another
00:11:30
layer of not just kind of solutions to
00:11:33
complexity Sure And so there's there's
00:11:35
quite a lot I think that's emergent here
00:11:37
um that Google will start to kind of
00:11:39
benefit from uh in the year ahead
