00:00:00
How do I sound?
00:00:00
>> You sound perfect.
00:00:01
>> You sound great.
00:00:02
>> How do I look?
00:00:03
>> Yeah, you sound great. Better than you
00:00:04
look. Have
00:00:05
>> a face made for radio.
00:00:08
>> You don't look as tired as you have in
00:00:09
recent weeks.
00:00:10
>> That's true.
00:00:11
>> Yeah.
00:00:11
>> Oh, yeah. Somebody was slagging me for
00:00:13
the bags and for my eyes. I mean, this
00:00:16
audience is brutal.
00:00:17
>> They're brutal.
00:00:18
>> They're brutal.
00:00:19
>> It's a good thing I'm rich.
00:00:24
>> Let your winners ride.
00:00:27
Rainman David
00:00:31
>> and it said we open sourced it to the
00:00:33
fans and they've just gone crazy with
00:00:35
it.
00:00:35
>> Love you guys.
00:00:39
>> All right everybody, welcome back to the
00:00:41
number one podcast in the world. It's
00:00:43
the Allin podcast. With us today,
00:00:46
Chimoth Polyatia, David Saxs, and our
00:00:49
fifth bestie, Mr. Brad Gersonner is
00:00:52
here. I think uh David Freeberg is
00:00:55
suffering from some socialist related
00:00:58
flu. He's very sick of reading about
00:01:00
socialists, but he'll be back next week
00:01:02
with two incredible incredible
00:01:04
interviews.
00:01:05
>> You guys see those Spencer Pratt ads?
00:01:07
>> Wow.
00:01:08
>> It's one of the best political ads I've
00:01:09
ever ever seen.
00:01:11
>> Oh, there's like three or four of them.
00:01:12
>> There multiples.
00:01:13
>> Yeah.
00:01:13
>> Whoever that social media team is is on
00:01:15
fire. If you get a good social media
00:01:17
team and you get a good ad production
00:01:18
team, I think it's nextgen because these
00:01:20
things go crazy. And Spencer Pratt, if
00:01:22
he wins this election, which I think
00:01:24
he's going to in Los Angeles, the reason
00:01:26
is what Brad said. Those ads are
00:01:29
incredible.
00:01:30
>> Well, he's also quite a good debater.
00:01:31
Did you see clips from this debate?
00:01:34
>> Incredible. He's so funny. He's so
00:01:37
chill.
00:01:38
>> Yeah. Well, he's up against Karen Bass,
00:01:39
who's the mayor, who is basically
00:01:42
extremely leftwing. And then there's
00:01:43
someone who's a city council woman who's
00:01:45
even further to the left of Karen Bass.
00:01:48
I mean, she's
00:01:49
>> often like Fidel Castro territory. She's
00:01:51
an Indian Fidel Castro
00:01:53
>> Ramen. So, she was basically, I guess,
00:01:56
criticizing the mayor for the
00:01:58
homelessness problem. And then Pratt
00:02:00
pointed out that this council woman is
00:02:03
actually in charge of all these homeless
00:02:05
programs already.
00:02:07
>> He eviscerated her.
00:02:08
>> Eviscerated her. And he basically made
00:02:11
the key point, which is, look, the
00:02:12
problem here is not lack of housing.
00:02:14
It's an addiction issue, and it's a
00:02:16
mental illness issue. And he said,
00:02:17
"Look, if you
00:02:18
>> He said, "If she went to the street,
00:02:20
she'd get stabbed in the neck."
00:02:22
>> Yeah.
00:02:23
>> Which is pretty accurate if you've been
00:02:24
to Skidro. I mean, you would not want to
00:02:26
walk through there. It It was like the
00:02:29
Spider-Man photo.
00:02:30
>> Wait, Nick's got the clip. Play the
00:02:31
clip, Nick.
00:02:32
>> Oh god, this clip is brutal.
00:02:33
>> This is a different clip, but this went
00:02:34
super viral. It reminded me of Trump a
00:02:36
little bit.
00:02:37
>> Let's see.
00:02:38
>> I'm not sure how to respond to that
00:02:39
vision of Los Angeles. This is a MAGA
00:02:42
Republicans idea of what Los Angeles
00:02:44
looks like. This is This is really not
00:02:49
>> unbelievable. For those of you
00:02:51
listening, he put his hands up and
00:02:53
wiggled his head like, "Oh my god."
00:02:55
>> Hey, hey, Sachs. You know, Steven Pratt
00:02:58
wins mayor.
00:02:59
>> Spencer Pratt the, you know, the ballot
00:03:01
initiative, the retirement protection
00:03:03
and savings act, right? It's going to
00:03:05
pass. Going to pass with big numbers.
00:03:06
This is the, you know, referendum that
00:03:08
effectively is going to knock out the
00:03:10
wealth tax. Can you imagine if
00:03:11
California effectively passes a
00:03:14
constitutional amendment protecting
00:03:16
retirement savings and personal assets
00:03:18
and banning the wealth tax and crack
00:03:20
gets elected? The message that would
00:03:22
send to the country. That's that's a
00:03:24
very non-conensus view that I'm becoming
00:03:27
increasingly optimistic about.
00:03:29
>> Well, from your lips to God's ears, but
00:03:32
until that message actually is sent, I
00:03:34
think I'm going to be uh
00:03:36
>> in Texas. in Texas.
00:03:38
>> Yeehaw. Well, I mean, this is also in
00:03:41
the face of I don't know, just a
00:03:42
follow-up story here, but Mandami did
00:03:45
like an attack video on Ken Griffin's
00:03:46
house. We talked about it on the pot a
00:03:48
couple weeks ago, and like literally
00:03:50
stood in front of his house, pointed at
00:03:52
it, and this is in the face of like a
00:03:54
CEO getting shot for ideological
00:03:56
reasons, Sam Alman's house being
00:03:58
targeted. This is like a really
00:03:59
dangerous thing for Mandami to do. And
00:04:02
Ken Griffin came out today and uh or
00:04:04
yesterday in an interview and said,
00:04:05
"Hey, listen. I'm out. We're going to be
00:04:07
putting our efforts into um Florida. And
00:04:11
this is the same thing that happened to
00:04:12
Chicago. And he basically said like I
00:04:14
really felt offended and I was, you
00:04:16
know, nervous about this because of my
00:04:19
personal safety. And he called him out.
00:04:21
Mondami came out with like a a
00:04:23
mealymouthed response that didn't even
00:04:25
apologize for what he did. Just double
00:04:28
down on it essentially. All right, let's
00:04:30
get to the dot. I don't know if you guys
00:04:31
saw it or not, but
00:04:32
>> who cares? New York is becoming a
00:04:34
flyover city. It's interesting way to
00:04:36
put it. Uh, don't disagree. All right,
00:04:38
first story. Elon just leased all of
00:04:40
Colossus 1, his data center.
00:04:42
>> He did.
00:04:45
>> What? WHAT? YES. SHOCKING to Daario and
00:04:48
Anthropic Chamath on last week's pod.
00:04:51
Uh, go ahead and give yourself a pat on
00:04:52
the back. You said Elon and Dario should
00:04:54
do a deal tomorrow. It didn't happen the
00:04:56
next day. It happened 5 days later. So,
00:04:58
you came close, Jimoth. Uh, but no cigar
00:05:00
because of Anthropic's obvious compute
00:05:03
constraints. Anthropic just added over
00:05:05
220,000 Nvidia GPUs, over 300 megawatts
00:05:09
of energy. The deal is already having an
00:05:12
impact. As we've discussed here, Claude
00:05:15
users have been experiencing rate
00:05:16
limits. Well, Claude has now doubled the
00:05:19
Claude code rate limits, removed peak
00:05:20
usage caps for paid users, and increased
00:05:22
API volumes for Opus models. XAI is now
00:05:26
trading their models at Colossus 2. So,
00:05:29
they have more than enough compute. Elon
00:05:31
made a great bet on compute and built up
00:05:34
those data centers really fast and that
00:05:36
is now paying off. We had the cursor
00:05:38
deal we talked about last week. Let's
00:05:40
talk about the emergence of Elon Web
00:05:43
Services EWS Chim. He is now in the
00:05:46
hyperscaler competing against Google
00:05:49
Cloud, Amazon Web Services and Azure.
00:05:51
And uh I don't know if you had inside
00:05:53
information or just a brilliant uh
00:05:55
epiphany but uh take us behind the call
00:05:57
and what do you think about the deal
00:05:59
itself?
00:06:00
I think the deal is fantastic. I'll say
00:06:03
maybe three quick things.
00:06:06
The first is, as I mentioned a couple
00:06:08
weeks ago, Anthropic and OpenAI's
00:06:13
revenue performance has nothing to do
00:06:16
with demand. Zero. It is entirely to do
00:06:20
with the supply constraints that exist
00:06:23
in data centers and specifically in
00:06:26
power. If they had infinite power, I
00:06:29
think that their revenues would probably
00:06:30
be even more parabolic. And so all the
00:06:33
breathlessness about either exceeding or
00:06:37
underperforming a forecast, in my
00:06:39
opinion, mean nothing. I think the
00:06:41
five-year view for those two companies
00:06:44
is quite robust. The thing that they
00:06:48
really need is more compute and more
00:06:51
power. That's the first thing. The
00:06:53
second thing is while they need that, we
00:06:56
have a very big problem which is we
00:06:59
unfortunately have very poor leadership
00:07:02
at the head of most of these AI firms.
00:07:07
I think they are coming off as
00:07:09
untrustworthy or too self-interested.
00:07:13
The political reaction now is starting
00:07:16
to turn negative. The community reaction
00:07:20
is negative. You have about nine
00:07:23
gigawatts that are supposed to come
00:07:25
online this year. Almost 50% of it now
00:07:28
is being protested. More than likely, if
00:07:31
if history holds,
00:07:33
most of that will get turned off. So,
00:07:35
they will get even more supply
00:07:36
constraint. So, that's the setup. So,
00:07:39
what's the opportunity I think for Elon?
00:07:42
If you look inside of how people try to
00:07:45
nitpick the SpaceX valuation case, or
00:07:47
let's not even let's give them sorry,
00:07:48
let's be more generous. When people try
00:07:51
to paint the bare case or they tried to
00:07:52
red team the valuation, the biggest
00:07:55
element is the onthe-c value around the
00:07:59
orbital data centers. And by actually
00:08:02
landing a bunch of terrestrial capacity,
00:08:05
I think you start to blunt that because
00:08:07
you can now start to say that even if
00:08:09
the orbital data centers get delayed by
00:08:11
a few months or a few quarters, even if
00:08:13
the technological
00:08:15
derisking of it takes longer, he now has
00:08:18
this structural core business that will
00:08:20
effectively subsidize
00:08:23
his ability to train Grock, which I
00:08:25
think is a really important and under
00:08:26
reportported theme. So you have all this
00:08:29
infrastructure. He somehow saw the tea
00:08:31
leaves before most people. He built to a
00:08:35
level of scale and secured power before
00:08:37
most people. It has now become the
00:08:39
critical asset. And now he's kind of
00:08:41
kingmaking.
00:08:43
And I think that that's a really
00:08:44
interesting valuation
00:08:47
reinforcement as SpaceX goes through
00:08:49
testing the waters and the and the road
00:08:51
show.
00:08:52
>> Brad, your take?
00:08:53
>> Yeah. No, I think it's well said. I
00:08:55
mean, first we we know that there's
00:08:57
nobody better on planet Earth than Elon
00:08:59
at converting electrons to tokens. It's
00:09:01
a critically important evolution to the
00:09:04
story. You know, I think our friend
00:09:06
Shawn Magcguire, he he he sent out a
00:09:07
tweet that summed it up well and he said
00:09:09
SpaceX has this five layer cake, launch,
00:09:12
connectivity, compute, hyperscaler,
00:09:15
space data centers, and then
00:09:17
applications and models and then other
00:09:19
bets. Right? The question on the road
00:09:22
show has been but X.AI AI doesn't isn't
00:09:25
on the revenue trajectory of open AI and
00:09:28
and anthropic and yet there are huge
00:09:30
commitments and now we see the ace card
00:09:34
that Elon's playing. He said he was
00:09:36
building AWS all along or EWS all along.
00:09:40
And so I estimate that this is going to
00:09:42
generate in this year an incremental 4
00:09:44
to5 billion of revenue on top of what I
00:09:47
I've seen analysts estimates in the
00:09:49
mid20s. That's a material amount of
00:09:51
incremental revenue to offset the cost
00:09:53
of the investments that he's made here.
00:09:55
And that will subsidize to Chamas's
00:09:58
point all that he's investing to build
00:10:00
the next generation of Groth. Remember
00:10:03
too that he has three facilities,
00:10:05
Colossus, Macro hard, and Macro harder.
00:10:08
1.2 gawatts in Macro hard and macro
00:10:11
harder in Blackwell. So he's given the
00:10:13
one that's kind of less connected.
00:10:15
H100's great for inference to anthropic.
00:10:19
He's monetizing it in a big way. It's
00:10:21
terrific for Anthropic and it solves
00:10:24
what I think was the biggest question in
00:10:26
the valuation story which is what if he
00:10:30
spends ahead of X.AI's revenue. It takes
00:10:33
the pressure Chimath off X.AI delivering
00:10:36
immediate revenue. Now he becomes uh an
00:10:39
immediate competitor in the hyperscaler.
00:10:41
I don't think this is the last
00:10:42
announcement. I think he's going to make
00:10:44
a lot more you know moves in this
00:10:47
direction. I think it will be a material
00:10:49
part of their story and their revenue
00:10:51
projections uh as they come together and
00:10:54
I would just say finally you know again
00:10:56
everybody has talked about how we don't
00:10:58
have enough power how we don't have
00:10:59
enough compute how the revenues would
00:11:01
not show up this year you know but the
00:11:03
chaos that is American capitalism
00:11:05
somehow finds a way okay and there's
00:11:08
tremendous demand for anthropic and we
00:11:11
find a way I was so happy to see kind of
00:11:15
the dant and the kind exchange
00:11:18
between the team of Anthropic and Elon
00:11:20
because we need all of this in order to
00:11:23
produce uh American frontier models to
00:11:25
stay at the frontier. And then finally,
00:11:28
I just say, you know, Chimath, you
00:11:29
referenced these activists that are
00:11:30
protesting delaying these data centers
00:11:33
in in these localities. One thing I want
00:11:35
to dispel this myth, these this is not
00:11:37
like organic hyper local protests by
00:11:40
people in a community that aren't being
00:11:42
spurred on. This is highly organized
00:11:45
activists that are moving across the
00:11:47
country to stir up trouble in the exact
00:11:50
same way they did to stop all fision
00:11:52
reactors being built 30 years ago in
00:11:55
America. Now we have no nuclear reactors
00:11:57
being built. China's got a hundred of
00:11:59
them. Who was funding those activists? I
00:12:02
think we need to really look into who's
00:12:04
funding the activists now. I'm not
00:12:06
saying that there aren't any concerns,
00:12:07
but the misinformation about water, the
00:12:10
misinformation about electricity bills,
00:12:12
electricity bills are going up in the
00:12:14
places that are not building data
00:12:16
centers, New York and California,
00:12:18
because they haven't built any supply on
00:12:21
the grid. In Texas, where you're
00:12:22
building the most data centers in the
00:12:24
country, electricity costs are going
00:12:26
down. So, um, I'm I I think that's a
00:12:28
boogeyman that we got to take on.
00:12:31
>> Sure. Johnson.
00:12:33
>> Well, look, the deal is highly
00:12:35
complimentary. As Chamath and Brad
00:12:37
pointed out, SpaceX has a profitable, I
00:12:41
think very profitable space and
00:12:43
telecommunications Starlink business,
00:12:45
the satellite business, but the XAI
00:12:48
business had huge losses. The reasons
00:12:50
pretty straightforward. You need these
00:12:52
super large training clusters, but they
00:12:53
cost a lot of money. And until you have
00:12:55
a model that's capable of competing at
00:12:57
the frontier, you're not making any
00:12:58
revenue. And that problem is compounded
00:13:00
by the fact that right now all the
00:13:01
revenue is in enterprise, which is to
00:13:03
say coding. We know that XAI just did
00:13:06
that deal with Cursor to try and catch
00:13:08
up, but they don't have a coding product
00:13:10
yet. So they're not participating in the
00:13:12
revenue, but they're participating in
00:13:13
all the cost. So this deal fixes that
00:13:16
problem. Elon's now able to have a
00:13:19
frontier model company, but he's able to
00:13:22
now not have these massive unpaid for
00:13:25
capex commitments, right? because he's
00:13:27
able to kind of lease that capacity. So
00:13:30
I think it solves a major problem for
00:13:31
them and their balance sheet. And then
00:13:33
you have to say that philanthropic this
00:13:35
is a really great thing because they
00:13:38
were compute constraint. And just to
00:13:40
build on that point, I mean I guess let
00:13:42
me be the first to congratulate Dario on
00:13:44
winning the AI race.
00:13:46
>> And you've been, let's be honest, Saxs,
00:13:49
you have been on this podcast, you've
00:13:51
been moderately critical of that company
00:13:53
and Daario himself for being um, you
00:13:57
know, a little P Doomer 110. And on your
00:14:00
ex account, you've been even a little
00:14:02
spicier. So now that there's peace in
00:14:04
the Middle East of uh of the AI
00:14:07
business, what's your take here?
00:14:09
>> My take is look, let's just honestly and
00:14:11
accurately assess where the state of
00:14:14
this AI market is at right now and
00:14:16
Anthropic's place within it. So for the
00:14:18
last 3 years, Anthropic has been growing
00:14:21
at a rate of 10x a year. I think going
00:14:24
into this year, probably the
00:14:26
conventional wisdom was that there'd be
00:14:27
no way to sustain that kind of rate of
00:14:30
growth at this level of scale. And what
00:14:32
happened in the first four months of the
00:14:34
year? First, we find out that from
00:14:36
January 1st to March 31st, they grew
00:14:38
from roughly 10 billion of ARR to 30
00:14:40
billion. So, it tripled. And then in
00:14:43
April, if anything, the rate of increase
00:14:45
seemed to accelerate. They went from 30
00:14:47
to 44 billion of ARR. Nobody in Silicon
00:14:51
Valley has ever seen anything like it.
00:14:52
Forget about the rest of the country. I
00:14:54
mean, all we do in Silicon Valley is
00:14:56
deal with exponentials. And still,
00:14:57
people have never seen that kind of
00:14:59
growth at that level of scale. The only
00:15:02
thing holding them back in the future
00:15:03
was compute. Now they've made this deal.
00:15:06
They've made other deals as well to get
00:15:07
that compute. I think it's pretty much a
00:15:09
foregone conclusion that they will hit
00:15:11
that forecast of 10x this year exiting
00:15:14
the year. Call it roughly 100 billion of
00:15:15
ARR. And now the only question is
00:15:18
whether they hit a trillion in 2027.
00:15:22
And we can debate
00:15:24
>> getting on board. We can we can debate
00:15:26
whether that's true or not. But look, if
00:15:28
they do that, I think they'll easily be
00:15:31
the most valuable tech company in
00:15:33
history. In fact, they might even be
00:15:34
more valuable than the rest of the Mag 7
00:15:37
put together. Just to give people some
00:15:38
basis for comparison here, you know, the
00:15:40
biggest tech companies, Apple,
00:15:42
>> Nvidia,
00:15:43
>> Nvidia, Google. I think they kind of do
00:15:45
around 4 to 500 billion a year right now
00:15:48
>> of of revenue. I guess Nvidia is in a
00:15:50
little bit of a different category, but
00:15:51
you look at
00:15:52
>> you look at the hyperscalers, the three
00:15:53
hybrid scalers. Yeah. I mean, Google is
00:15:56
doing what, like 120 billion a quarter,
00:15:58
something like that. 100 billion a
00:16:00
quarter.
00:16:00
>> Correct.
00:16:01
>> But growing at what, 20% year-over-year?
00:16:03
Not 100%. It's certainly not a
00:16:05
thousand%.
00:16:06
So, the fact that anthropic could be on
00:16:10
track. In fact, let me correct.
00:16:11
>> You see them going to the track. It'll
00:16:13
be a Maggie. I'm saying something else
00:16:15
which is that unless something about
00:16:19
their current trajectory changes
00:16:22
anthropic will be the most powerful
00:16:24
monopoly ever created in human history.
00:16:27
>> Oh,
00:16:28
>> again it will be, you know, a trillion
00:16:30
dollars of ARR growing at some
00:16:32
exponential.
00:16:34
>> Interesting.
00:16:34
>> Dario calls it AGI. I call it the
00:16:36
biggest monopoly in human history.
00:16:38
>> Interesting to hear that word monopoly.
00:16:40
Sachs, very interesting placement.
00:16:42
Chimath go ahead and then I'll move
00:16:45
in 2025 was 420 billion Microsoft was
00:16:50
300 billion Alphabet was 390 billion
00:16:53
Amazon 700 billion Nvidia 190 billion
00:16:56
Meta 185 billion Tesla 110 billion total
00:17:00
about 2.3 to 2.35 trillion so if Sax is
00:17:03
right and Anthropic you know can tack on
00:17:08
a trillion it won't be the mag seven
00:17:10
it'll be the mag
00:17:12
Just to put it in perspective though,
00:17:13
Daario on Dwarkkish said he thought the
00:17:16
combined AI revenue of the market
00:17:19
leaders would be about a trillion in 29.
00:17:21
I love what you're saying. Sachs, I
00:17:23
think there is unlimited TAM. We may be
00:17:25
over our skis a little bit in terms of,
00:17:27
you know, the forecast. If you back your
00:17:29
way from compute, right, they're they
00:17:31
expect to have 5 gigs by the end of this
00:17:33
year, 10 gigs by the end of next year.
00:17:34
It's kind of hard to get to those
00:17:36
numbers for a single company, but I do
00:17:38
believe that the, you know, trajectory
00:17:40
that they're on, I totally agree with
00:17:42
you, is on an exponential um that not
00:17:44
many people believed in 4 months ago.
00:17:46
>> Right. So then the question is, okay, I
00:17:48
think we all agree they're on an
00:17:49
exponential curve and that the TAM is
00:17:51
big enough to support that. Just one
00:17:53
data point on TAM. My understanding of
00:17:56
the total market size just on coding is
00:18:00
one trillion meaning that a trillion
00:18:02
dollars a year roughly is spent on
00:18:04
software developers and all things
00:18:06
related to the creation of software.
00:18:08
Now, I'm not saying that they eat that
00:18:10
entire market, but I can easily see the
00:18:12
market for software doubling.
00:18:14
>> Well, hold on. Doubling from a trillion
00:18:16
to two trillion given that coding tokens
00:18:20
basically 10xes or 100xes the value of
00:18:24
that market and the ability to generate
00:18:26
code. So, I think we all agree that the
00:18:28
TAM here is large enough to support a
00:18:31
trillion dollars of revenue. Brad, I
00:18:32
think you bring up a couple of really
00:18:34
important constraints. First, there may
00:18:36
not be enough comput and there's not
00:18:37
enough energy. I'd say the second big
00:18:38
one is what's the competitive reaction
00:18:40
going to be? Totally.
00:18:41
>> Because I would say at the beginning of
00:18:43
this year, all these frontier labs were
00:18:45
playing around with a lot of different
00:18:46
things. I mean, Anthropic was the
00:18:49
porcupine. They believed in one thing.
00:18:51
All these other companies were kind of
00:18:53
acting like the fox who thinks they're
00:18:54
good at a lot of different things. They
00:18:56
were doing Nano Banana. They were doing
00:18:58
Sora. They were doing, you know, they
00:18:59
were doing image generation. They were
00:19:01
doing fantasy character chatbots. In
00:19:04
hindsight, they were doing a lot of
00:19:06
things that appear to be kind of a waste
00:19:07
of time. The whole market appears now to
00:19:10
be coding and the things that will be
00:19:12
built on coding tokens like co-work like
00:19:14
agents. And so there is going to be a
00:19:16
competitive response here where all the
00:19:18
other guys realize, oh wait a second, we
00:19:20
were misfocused. They're going to get
00:19:22
focused. I just don't know how much
00:19:24
share they're going to be able to take.
00:19:26
It does look like OpenAI has already
00:19:28
made the pivot. We hear very good things
00:19:30
about Codeex now based on GPT 5.5. 5.5
00:19:34
is based on a new base model called
00:19:36
Spud. I think they're very optimistic
00:19:37
about continuing improvements. Their
00:19:40
rate of growth appears to be
00:19:41
accelerating now because of uh 5.5.
00:19:45
So look, there's reason to believe that
00:19:47
OpenAI can take some share here. I'm
00:19:49
sure that Google won't be asleep at the
00:19:52
wheel. They're very very good at coding.
00:19:53
They've got a really good team and Elon
00:19:55
just tied up with Cursor. So there is
00:19:57
going to be more competition but still
00:19:59
what you have to say and I think all of
00:20:01
us know this from Silicon Valley is you
00:20:03
always want to be the company in the
00:20:05
lead that's on that trajectory where all
00:20:08
you have to do is maintain inertia
00:20:10
whereas the other people have to change
00:20:12
something
00:20:13
>> in order to put themselves back in the
00:20:14
race. So this is when I say somewhat
00:20:17
sort of faciciously congratulations
00:20:19
Dario on winning the AI race. I don't
00:20:22
mean that he's won it but he is winning
00:20:24
it right now. Well, here's the the
00:20:26
brilliance of what Elon's doing. If you
00:20:28
look at the existing business, which is
00:20:30
Starlink and basically the launch
00:20:31
services at SpaceX, incredible business.
00:20:34
Obviously,
00:20:36
20 billion this year, I think is the
00:20:37
estimate. But if you look at the
00:20:39
footprint of Amazon Web Services, Azure
00:20:41
and GCP, you're looking at, you know,
00:20:44
300 billion dollars in revenue and a
00:20:49
market cap of combined 5 trillion, 4
00:20:53
trillion if these were independent
00:20:54
companies. And if you look at what is
00:20:57
Elon's core competency at Tesla, it's
00:21:00
building factories. And if you look at
00:21:02
the footprint of these factories,
00:21:04
they're huge. What are data centers?
00:21:05
They're basically big giant factories.
00:21:07
And then if you look at energy, what
00:21:10
else is Elon extremely good at? This is
00:21:12
the battery deployment and he's also got
00:21:14
solar deployment from the Solar City
00:21:16
often criticized acquisition he did
00:21:18
years ago. So you put this all together.
00:21:21
If this is 5 billion, as I think you
00:21:23
referenced, Brad, if it's $5 billion in
00:21:25
incremental
00:21:27
Elon Web Services business and he's a
00:21:29
Neocloud, what could he build on planet
00:21:33
Earth? What could he build inside of
00:21:35
Teslas in terms of extra compute? What
00:21:37
could he build inside the power wall?
00:21:38
What if the power walls had his new fabs
00:21:41
in them and you built a distributed
00:21:42
system from home to home? The power wall
00:21:45
has compute in it. The cars have comput
00:21:48
in it. And of course, the ultimate
00:21:50
manifestation of this where nobody can
00:21:52
complain is you go right out into space.
00:21:55
And that's what he's going to do. And
00:21:57
the the sneaky small part of this
00:21:59
announcement from Elon and uh from
00:22:01
Anthropic and Daario was they're also
00:22:04
interested in space. So look for the
00:22:07
race to go from factories and data
00:22:10
centers to homes. The power wall with
00:22:12
comput in it. It's already online,
00:22:14
right? And Starlink also gives him the
00:22:16
ability to do distributed comput's
00:22:19
homes. Again, you could be paying people
00:22:21
to put power walls with computing in it.
00:22:23
That's going to be the next shoe to
00:22:24
drop, I believe. Did you guys see the
00:22:26
deal that was announced
00:22:28
yesterday between PY Homes, which is a
00:22:31
huge builder
00:22:33
>> and SPAN?
00:22:34
>> Yes.
00:22:35
>> Nick, just throw this up here. It's
00:22:36
super cool. What's happening is that
00:22:38
these guys
00:22:39
>> are putting many data centers with
00:22:42
Nvidia GPU clusters
00:22:44
beside every home and then allowing
00:22:46
people to actually run those things. And
00:22:47
that's just incredible. I thought that
00:22:49
was so cool.
00:22:50
>> It's a great pivot. What this company
00:22:51
did originally, Chimamoth, was they did
00:22:53
the power panel. They made smart power
00:22:55
panels. So, you know, when you flip your
00:22:56
breakers, all those breakers are in an
00:22:59
app. I looked at it for my house, but I
00:23:00
guess they pivoted to add this and I
00:23:02
think base power, Brad, you're an
00:23:03
investor in it. They're going to do the
00:23:04
same thing.
00:23:05
>> Zack Dell is doing that. One of the
00:23:07
things I just say in response, Jason, to
00:23:08
what you just said about Elon, right?
00:23:10
This is why the SpaceX IPO is going to
00:23:13
trade at 40 to 50 times revenue. Okay.
00:23:17
So next year if they do 40 to 50 billion
00:23:20
and this thing goes out at 2 trillion
00:23:22
right that they're going to trade at a
00:23:24
really high revenue multiple compared to
00:23:26
the mag five that are trading at like 25
00:23:29
times earnings and there's only one
00:23:31
person on the planet who has a future
00:23:34
pipeline of innovation and the largest
00:23:37
TAM in the world because he's playing in
00:23:38
all these different spaces that can
00:23:40
command that multiple and it's Elon and
00:23:43
it's deserved and it's great for the
00:23:44
country
00:23:45
>> has that same Tesla has that same Elon
00:23:48
uh variable in it as well which is
00:23:50
people value his companies at I would
00:23:52
say two times market three times market
00:23:55
four times market because of the future
00:23:57
pipeline and they devalue Apple because
00:23:59
they don't have somebody like Elon or
00:24:01
Steve Jobs there who is giving them the
00:24:03
future
00:24:03
>> I don't think it's devalued I think
00:24:04
>> or properly valued if you don't have an
00:24:06
Elon and you have somebody like I think
00:24:08
that's exactly what it is we talked
00:24:10
about this last week but explain why you
00:24:11
think it's different
00:24:12
>> I think all of these companies are
00:24:13
actually very fairly valued and then
00:24:16
Elon World gets a premium
00:24:18
>> totally
00:24:19
>> and that premium is because of what you
00:24:21
guys said that I agree with. The big
00:24:23
message that I take away from this which
00:24:25
the markets and retail are telling you
00:24:28
is you guys have stopped innovating.
00:24:32
There's a lot of incrementalism
00:24:34
and we as a society aren't benefiting
00:24:37
broadly the way that you told us we
00:24:39
would be. And so maybe this is the best
00:24:42
way for them to get this message, which
00:24:44
is to whack their valuation. And by the
00:24:46
way, I'll just say it again, when Tesla
00:24:48
and SpaceX merge and we have all things
00:24:51
Elon and Elon Corp, okay, which will
00:24:53
happen probably by the end of the year.
00:24:55
Maybe it'll happen in the middle of next
00:24:57
year.
00:24:59
It's going to then break everybody's
00:25:00
brains again because you'll have this
00:25:02
one asset, as you guys said, that will
00:25:05
trade at a valuation premium. And some
00:25:08
people will say it's unexplainable. And
00:25:10
I think it's logically explainable,
00:25:11
which is everybody else has stopped
00:25:14
innovating. People know how to draw more
00:25:17
blood from the stone, how to target
00:25:19
better ads.
00:25:21
That does nothing for society anymore.
00:25:23
>> That's it. Literally, in fact, it does
00:25:25
the opposite. It there is no good left.
00:25:27
That was literally the exact point I was
00:25:29
making when you cut me off. If you look,
00:25:31
Tim Cook's greatest innovation Tim
00:25:33
Cook's greatest innovation before you
00:25:35
cut me off was Apple TV. Not even the
00:25:38
hardware product. It was just spending
00:25:39
money and making a Netflix knockoff.
00:25:41
There's been no other product in
00:25:43
>> Hold on. Let me finish again before you
00:25:45
interrupt me making my
00:25:46
>> You don't like that? Oh, you like me?
00:25:47
>> Oh, no. Oh, yeah. Well, okay. Okay. Go
00:25:49
back to
00:25:50
>> meat kettle.
00:25:52
>> If you look at their track record and I
00:25:54
think this is why we had a change there
00:25:57
is they have not done anything
00:25:59
innovative and in fact the things they
00:26:00
were doing that were innovating in AI or
00:26:03
self-driving cars, they shut down. They
00:26:05
won't take any swings for the bat. So
00:26:06
they are getting penalized in their
00:26:08
valuation penalized. They're just not
00:26:10
getting a premium. They're not getting
00:26:11
penalized.
00:26:12
>> I think they're getting penalized.
00:26:13
>> Every metric they're trading at
00:26:14
incredible valuations. Just look at
00:26:16
them.
00:26:17
>> Oh no. I don't I mean if you compare the
00:26:19
two valuations, I think they're being
00:26:20
penalized. Anyway, let's Anybody else
00:26:22
want to get in on this before we move on
00:26:24
to the next one? Yeah.
00:26:25
>> There is no world in which Google and
00:26:27
Meta and Apple and Amazon could be
00:26:32
viewed as being penalized in valuation.
00:26:35
There is very clearly a world where Elon
00:26:37
gets a massive premium because he's
00:26:39
innovating.
00:26:41
>> You're saying the same things. You're
00:26:42
saying the same thing the same thing.
00:26:44
It's not the same.
00:26:45
>> Listen, we're I think we're we're
00:26:46
debating semantics here. I'm not letting
00:26:48
you off the hook, Saxy Poo. When Sax is
00:26:52
very deliberate in how he speaks, they
00:26:54
said he's the captain of the debate club
00:26:55
in his 20,000word article this week and
00:26:57
that he's a master debater. He's a
00:26:59
masturbator. And you slipped in. You
00:27:03
slipped it in. Are you saying that the
00:27:06
FTC or whoever should be going in and
00:27:08
looking at anthrop Oh, Brad's book is
00:27:11
getting attacked headwinds. You said
00:27:13
they're a monopoly or they're heading to
00:27:15
monopoly tactics s is that what you're
00:27:18
saying?
00:27:19
>> Well, look, I mean, we know that tech
00:27:22
markets have a history of consolidating
00:27:25
down and turning into either monopolies
00:27:27
or duopolies. And if you just look at
00:27:29
the revenue right now, there's only two
00:27:31
companies making substantial revenue on
00:27:35
AI. It's Enthropic and Open AI. We know
00:27:38
that OpenAI is growing at 3 to 4x, which
00:27:40
is incredible at the level of scale
00:27:42
they're at. Enthropic though, we said,
00:27:43
is growing at an exponential 10x a year.
00:27:46
And if they just do that for 18 more
00:27:49
months, they'll be by far the most
00:27:51
valuable company in human history. And
00:27:53
they'll have unprecedented control over
00:27:55
the most important technology of our
00:27:57
time. So, I don't know what you call
00:27:59
that, but it is something to think
00:28:02
about. And I guess I do have a thought
00:28:03
experiment for you guys, which is I just
00:28:07
want you to think for a second about the
00:28:10
case of of John D. Rockefeller, who I
00:28:13
think is known as probably the most
00:28:16
successful, most ruthless
00:28:18
>> monopolist in in American history. But
00:28:21
he wasn't very good at PR. He was
00:28:22
terrible at PR. Everyone sort of
00:28:24
recognized how ruthless he is. We've
00:28:26
seen movies like There Will Be Blood,
00:28:28
which is basically about him. In any
00:28:29
event, imagine if John D. Rockefeller
00:28:32
was way better at public relations, and
00:28:35
instead of calling his company Standard
00:28:37
Oil, he called it safe oil. Okay, let's
00:28:40
just let's just play this thought
00:28:41
experiment.
00:28:42
>> Clean, beautiful coal. Yes,
00:28:43
>> safe oil. He called it safe oil because,
00:28:46
as we know, kerosene is dangerous. Their
00:28:48
first big product was kerosene. And
00:28:51
kerosene can light your house or it can
00:28:53
burn it down. and in the wrong hands it
00:28:55
can torch a city or you can use it to
00:28:57
make a bomb. So John D, let's say should
00:29:02
have called for the creation of a new
00:29:04
government agency to regulate the safety
00:29:06
of his product and they could have done
00:29:08
rigorous testing, licensing, common
00:29:10
sense regulation. There would have been
00:29:12
a very intense debate over safety
00:29:14
standards, you know, what should the
00:29:15
proper wick thickness be and should we
00:29:18
allow all those dangerous independent
00:29:21
refiners, right? And I think people
00:29:23
would have gotten so wrapped up in this
00:29:25
debate over what constituted safe oil or
00:29:29
safe kerosene that they would have
00:29:31
missed what was really going on, which
00:29:33
is that Rockefeller was building the
00:29:34
richest, most powerful monopoly of all
00:29:36
time. In fact, people might even have
00:29:38
called Rockefeller an effective altruist
00:29:41
because of course he was so concerned
00:29:43
about the safety of his product.
00:29:46
>> I love it. Shout out to David Sax's
00:29:48
writers. Great great writers. Newman.
00:29:51
Newman wrote this.
00:29:52
>> No, I wrote it.
00:29:53
>> Emmy award for best writing in a
00:29:55
dramatic monologue goes to Newman. Wow.
00:29:58
Sax writing. He landed it. Very good.
00:30:00
Sax,
00:30:01
>> I thought after the Elon anthropic dant
00:30:04
where Elon said, you know, complimented
00:30:06
anthropic that and David started off
00:30:08
with a bit of a compliment. I thought we
00:30:10
maybe were past this first. It's
00:30:11
ridiculous to think of this as a
00:30:13
monopoly. You know, we're talking about
00:30:15
annual run rate revenues, David, but on
00:30:18
a gap basis, they're doing about the
00:30:19
same revenue as OpenAI in the month of
00:30:21
March. Okay? So, we're way ahead of
00:30:23
ourselves. By the way, 5 months ago,
00:30:25
everybody thought Open AI was going to
00:30:27
run away with this. Google's revenues
00:30:29
are very substantial in AI. And by the
00:30:32
way, Google, Amazon, etc., these
00:30:34
companies are producing hundred billion
00:30:36
dollars of free cash flow to justify
00:30:38
their incremental investment. At the
00:30:39
same time, you have these two startups
00:30:41
that are still fledgling, that are still
00:30:43
fragile in the scheme of things. You of
00:30:46
all people should know we've got the
00:30:47
best competition in AI on the planet,
00:30:50
which is why we're at the frontier and
00:30:52
kicking the tail of everybody else on
00:30:54
the planet. So, I just want to see these
00:30:56
companies compete. I want to see DC stay
00:30:58
out of the way. The last thing I want to
00:31:00
be doing is is is, you know, seeing
00:31:03
people talk about this and throwing
00:31:04
roadblocks into the way of the
00:31:05
competition. Um, I think I agree.
00:31:08
>> Well, let me let me hold on. Let me
00:31:10
translate Brad for you. Don't [ __ ] with
00:31:12
my paper is what he's saying. He's got
00:31:14
bets on these. So, Sachs, Washington,
00:31:16
DC, don't [ __ ] with Brad's paper.
00:31:19
>> Saxs, uh, do you want to get into the,
00:31:21
uh, regulation stuff right now as a
00:31:23
segue or
00:31:24
>> Let me respond to Brad and also
00:31:25
translate what I'm saying satirically.
00:31:28
Okay.
00:31:29
>> First of all, nobody wants to see these
00:31:32
companies compete vigorously more than
00:31:35
me. That was the whole premise
00:31:38
>> of the action plan that we worked on
00:31:40
last year is we want to bring out the
00:31:41
best in everyone. This is how America's
00:31:43
going to win the AI race. We have five
00:31:45
major labs vigorously competing and as
00:31:48
long as that competition is taking place
00:31:50
that I think that's a good thing.
00:31:51
Doesn't mean we can't have guardrails
00:31:52
and the rest of it, but basically
00:31:54
competition should be our northstar. All
00:31:57
of that being said, okay, what I am
00:32:00
pointing out and I I think it's
00:32:01
historically true that people in
00:32:03
Washington have woken up to monopolies
00:32:07
on the late side, not early, right?
00:32:10
Because I mean, once a company has won
00:32:12
80% of the market, that's when they wake
00:32:14
up and say, "Oh, we have a monopoly
00:32:15
here." And I'm not saying that they have
00:32:18
a monopoly yet. But if the trajectory
00:32:20
continues for just 18 more months,
00:32:23
>> then I think it will be in this
00:32:25
unprecedently powerful position. I mean
00:32:29
>> and and hold on and I don't think people
00:32:31
should be distracted from that fact by
00:32:34
this rhetoric around safety because
00:32:36
someone like Rockefeller could have used
00:32:38
it too. And I do think I mean just like
00:32:41
one one last point on this. I do think
00:32:43
that if you actually look at what a lot
00:32:46
of the the safetiest policies are
00:32:49
calling for, they're basically calling
00:32:51
for a form of regulatory capture and
00:32:53
they're calling for things that would
00:32:54
create a stronger moat around this
00:32:57
monopoly or duopoly that's in the
00:32:59
process of being created and it would
00:33:01
get in the way of competition. So again,
00:33:04
I think that people might not have such
00:33:07
a charitable view of all this safety
00:33:09
rhetoric if they understood that what
00:33:12
was being created here is the biggest
00:33:14
monopoly in human history. And I think
00:33:16
we should just be a little bit more
00:33:17
skeptical about some of these altruistic
00:33:20
claims. I can't believe that David is
00:33:24
like, you know, talking monopolies when
00:33:27
we haven't even left the starting gate
00:33:29
of AI. I I I I think this is a uh I to
00:33:34
me
00:33:34
>> there's only two companies with revenue.
00:33:36
>> The last thing I want is DC trying to
00:33:39
preemptively preemptively, which would
00:33:41
be like a disastrous consequence, get in
00:33:44
the game of picking winners and losers
00:33:46
at the starting line of AI. That would
00:33:48
be a disaster.
00:33:49
>> Brad, did you just put another soap box
00:33:51
on top of the soap box you were standing
00:33:52
on?
00:33:54
>> Look, Brad, like I said, my northstar is
00:33:57
competition. As long as there's
00:33:58
competition going on, I support it.
00:34:00
However, hold on, hold on. We know
00:34:03
>> that monopouists want to stop
00:34:06
competition and they use regulatory
00:34:07
capture to do it. And furthermore, they
00:34:09
do things like ban their competitors
00:34:11
from using their product. What
00:34:12
conceivable reason did Anthropic have
00:34:15
for banning OpenClaw using its models?
00:34:18
That is anti-competitive, is it not?
00:34:21
>> I I I would double click on it. I would
00:34:23
double click on it. I might not, you
00:34:24
know, file, but I would double click.
00:34:26
Okay, listen. Chimoth, the girls are
00:34:28
fighting. Let's keep moving through the
00:34:30
docket. We're going to be here all day
00:34:31
with these two. And uh one thing that
00:34:34
you're going to need to act on very
00:34:35
quickly is the all-in summit. It's
00:34:37
selling out fast. Don't miss it.
00:34:38
Speakers are top tier again. Freeberg
00:34:40
busy working on some amazing speakers.
00:34:43
Saxs will be there. Uh he's flying in
00:34:44
and out every day for four hours. And
00:34:46
then we're we're going to have a lot of
00:34:49
networking stuff going down. We're
00:34:50
building some networking software. So
00:34:52
when you come to our events, you get to
00:34:53
meet people. That's what we always say.
00:34:55
my playbook for events. If you learn
00:34:57
something from the speakers every day,
00:34:59
one or two things, if you meet somebody
00:35:01
new and you eat some great food and have
00:35:03
some fun, you get two or three of those
00:35:05
things. Ah, man. Even if you get one,
00:35:06
you're going to come back to the event,
00:35:07
you're going to get all three all day
00:35:09
long. allin.com/events. Los Angeles,
00:35:12
September 13th, 14th, and 15th.
00:35:15
Apologies to everybody asking, but
00:35:16
liquidity is sold out and we've shut
00:35:18
down the wait list. There's just no more
00:35:20
room. All right. The White House
00:35:22
allegedly, possibly is considering,
00:35:25
according to reports, an FDA for AI that
00:35:29
would vet, you heard that correct,
00:35:31
folks, that would vet new models for
00:35:33
safety. The thing we've been talking
00:35:35
about not doing here, the thing David
00:35:37
Saxs has spent the last year on, the
00:35:39
White House is considering. New York
00:35:41
Times reported Trump is considering an
00:35:43
executive order to create an quote AI
00:35:46
working group. This group would include
00:35:47
tech exeacts and government officials
00:35:49
who would quote examine potential
00:35:52
oversight procedures
00:35:54
including quote a review process for new
00:35:57
AI models. Oi according to the report
00:36:00
the catalyst was wait for it anthropics
00:36:03
mythos model which reportedly scared
00:36:06
spooked made people really nervous at
00:36:09
the White House. quote, "The White House
00:36:11
wants to avoid any political
00:36:13
repercussions if a devastating AI
00:36:16
enabled cyber attack were to occur. They
00:36:18
want a CYA, according to the New York
00:36:20
Times." Kevin Hassid, that guy, the
00:36:23
director of the National Economic
00:36:25
Council, confirmed the report on Fox
00:36:27
Business. Here's your 15-second clip.
00:36:28
We're studying possibly an executive
00:36:31
order uh to give a clear roadmap to
00:36:33
everybody about how this is going to go
00:36:35
and how future AIs that also potentially
00:36:38
create vulnerabilities uh should go
00:36:40
through a process so that you know
00:36:42
they're released to the wild after
00:36:43
they've been proven safe just like an
00:36:45
FDA drug.
00:36:46
>> Additionally,
00:36:48
friend of the pod Scott Bessen had
00:36:51
something to say. What we had in the
00:36:52
past month was a step change in the
00:36:55
power of one large language model, but
00:36:58
we're going to see it from the other a
00:37:00
AI companies. What we are determined to
00:37:02
do is work with our AI companies to
00:37:05
allow them to continue innovate. But our
00:37:08
charge of the US government is
00:37:09
maintaining safety. And there there is a
00:37:12
very important calculus here between
00:37:15
innovation and safety. And at the the US
00:37:18
government, we're going to make sure
00:37:19
that things stay safe. There you go.
00:37:21
Kevin Hasset and Bessant. Slightly
00:37:25
different positions here. Brad, what do
00:37:26
you think?
00:37:27
>> Actually, I don't think they're slightly
00:37:28
different positions, but I I I would
00:37:30
agree that Kevin bringing up the FDA
00:37:32
kind of muddied the waters. I talked to
00:37:33
Kevin last night after that clip ran,
00:37:37
you know, and I asked him I I just said,
00:37:38
"Do you think FDA is the right analog
00:37:41
here?" And he said, you know, I was I
00:37:42
was only bringing it up to say that they
00:37:44
we want them to show us the models so
00:37:46
that we can coordinate them. Obviously,
00:37:48
our job is to make sure that the
00:37:50
government is prepared, that we harden
00:37:52
our systems, that our intelligence
00:37:54
agencies are up to speed. But he does
00:37:56
not think, and I can't find anybody on
00:37:59
the right, you know, uh, that believes
00:38:01
that we're going to move to an approval
00:38:03
regime, right? The approval regime, this
00:38:06
idea that you're going to have to share
00:38:08
every model with an FDA in Washington
00:38:10
and they're going to have to pre-approve
00:38:12
the model is a disaster. Sachs has been
00:38:14
effectively fighting against this
00:38:16
correctly over the course of the last
00:38:18
year. It would just it would lead to
00:38:20
three bad things. Number one, we do not
00:38:22
want to put the wa Washington in in the
00:38:24
position of picking winners and losers
00:38:26
when it comes to these models. We're
00:38:27
winning. We're on the winning horse in
00:38:29
America. We're out in front of the rest
00:38:31
of the world. There's no reason to
00:38:33
change horses and regimes at this point.
00:38:35
And we don't want to burden this with
00:38:36
more democracy. But at the same time
00:38:39
obviously I call these pre-agi or agi
00:38:41
models mistral spud etc. I see a lot of
00:38:45
coordination going on between the
00:38:46
industry and government. I think we can
00:38:48
do an even better job of evolving that
00:38:50
framework so that everybody in
00:38:52
government is on the same page. We need
00:38:54
to build more capacity in government to
00:38:56
quickly be able to do the cyber review
00:38:58
on these models. Right now it takes too
00:39:01
long when the coordination does occur.
00:39:03
So we need to have a finite amount of
00:39:04
time that they get government feedback
00:39:07
etc. But the last thing that we want is
00:39:09
an FDA of models sitting in Washington.
00:39:12
Kevin understands that. Scott Besson
00:39:14
understands that. So I expect that we
00:39:16
will continue down the path that we've
00:39:18
been on. Chimoth obviously I think we
00:39:21
all agree we don't need an FDA for AI
00:39:23
but there are things that reasonably
00:39:27
people would want to have guard rails
00:39:30
around a AI. I'm sure you would agree it
00:39:33
shouldn't be a total free-for-all. So,
00:39:35
what's your take on this? Is it just
00:39:37
somebody gave a bad analogy here or
00:39:39
maybe some people were weasling their
00:39:41
way into the White House to try to shift
00:39:42
things when Sax was back at home or
00:39:44
something? What What's going on here?
00:39:46
Give us the uh cuz that's what that's
00:39:48
what people say. They say the last
00:39:49
person to talk to Trump kind of has his
00:39:52
ear and that things can bend a certain
00:39:54
way.
00:39:55
>> I don't think it's that. I think that
00:39:56
there's a pretty profound vibe shift
00:39:59
with respect to tech, tech oligarchs,
00:40:03
Silicon Valley, and particularly the AI.
00:40:05
That vibe shift has already happened on
00:40:08
Main Street, and I think that that's
00:40:11
starting to seep into Washington. I
00:40:13
think that regulations are coming. I
00:40:16
think they'll be worse under a
00:40:18
Democratic regime, but I think that some
00:40:20
form of oversight is going to exist
00:40:22
under a Republican regime.
00:40:25
The question that I think is worth
00:40:26
asking is why.
00:40:29
And if you listen to everybody's tone,
00:40:33
it's all around the negatives of AI.
00:40:36
So I think we suffer from two things.
00:40:38
Number one is we have horrible
00:40:39
messaging. Nobody spends the time and
00:40:43
the money to articulate the positive
00:40:46
upside case so that there's broad-based
00:40:49
support. And two, the idea that there's
00:40:52
going to be, as Sach said earlier, a few
00:40:55
winners and many, many, many potential
00:40:58
losers, I think is really disconcerting
00:41:00
to everybody. And the response from the
00:41:03
tech community again should be the
00:41:06
leadership of the tech world coming
00:41:09
together and actually reinvesting in
00:41:12
America at large. They're not doing that
00:41:14
in enough of a scale that blunts this.
00:41:17
So what you're seeing is the buildup of
00:41:19
antibodies. Is it avoidable? Yes. Are we
00:41:23
doing a good job of avoiding it?
00:41:25
Absolutely not. We're doing a horrible
00:41:26
job. I'd give the community, the tech
00:41:28
leaders a Dminus, trending to an F. The
00:41:33
response is what we're seeing. So I
00:41:34
think the question, Jason, isn't
00:41:37
regulation, no regulation. It's why did
00:41:39
we get here? And I think we got here
00:41:42
because the other version, the glass
00:41:45
half full version, the demonstrated
00:41:47
investment, the broad-based uplifting of
00:41:50
American society hasn't happened. And if
00:41:52
it has, it's been very poorly
00:41:54
communicated. And so the response is,
00:41:56
hey, hold on. We're going to give three
00:41:58
guys trillion dollar net worths and
00:42:01
we're going to allow them to control the
00:42:02
keys. That's why this is happening.
00:42:04
>> Exactly. Correct. And it's very easy,
00:42:07
Sachs, to imagine all the bad things
00:42:09
that can happen. Our minds are
00:42:11
constructed to do that. We're vigilant.
00:42:13
We look out for the tiger or, you know,
00:42:15
the tornado to keep ourselves safe.
00:42:18
Humans have a bias towards safety and
00:42:21
they're going to think about, you know,
00:42:22
deep fakes. They're going to think about
00:42:24
robotics. They're going to think about
00:42:25
self-driving cars, taking people's jobs.
00:42:27
They're going to think about, you know,
00:42:28
all the dark things that could happen,
00:42:30
boweapons, etc. And we don't have
00:42:32
anybody out there really talking about
00:42:34
all the positives that could happen.
00:42:37
What's your take on the palace intrigue
00:42:40
we all have here? What's going on in the
00:42:43
palace in the 47th administration around
00:42:46
this debate? Who's leading Trump down
00:42:49
the path of regulation and creating this
00:42:52
AI FDA? We know you're part of the camp
00:42:55
that wants to keep this train moving and
00:42:58
not overregulated, not have regulatory
00:43:00
capture. Who are the people trying to
00:43:01
slow this down?
00:43:02
>> Well, look, I I think there's several
00:43:04
things going on here. The first one is
00:43:06
there's a lot of fake news. This whole
00:43:08
idea of an FDA for AI, I don't think any
00:43:11
senior official supports it. Just like
00:43:13
Brad was saying, I spoke to Hassid as
00:43:15
well. That's not where his head is at.
00:43:17
So, I don't think anybody in the
00:43:19
administration is saying they want an
00:43:21
FDA for AI. Certainly, I don't think
00:43:24
that's the way the president thinks
00:43:25
about these issues. He's the most pro-
00:43:27
innovation president we've ever had. And
00:43:29
the White House chief of staff, Susie
00:43:30
Wall, just put out a statement last
00:43:32
night that I think pretty much shoots
00:43:33
this down. So I think there's a big fake
00:43:35
news component. Remember it was not
00:43:37
really the White House who was saying
00:43:38
it. It was the New York Times and there
00:43:40
and really I think actually Andrew Ross
00:43:42
Sto I'm not criticizing him but he's a
00:43:45
commentator and he's the one who said
00:43:47
this first and then somehow that spin or
00:43:50
that gloss somehow took on a life of its
00:43:53
own. And I think Silicon Valley reacted
00:43:57
accordingly. There's a very visceral
00:43:58
negative reaction here because we know
00:44:00
how damaging that would be to
00:44:01
innovation. But look, I think the good
00:44:03
news is that that was fake news. Second,
00:44:06
I think that there's another thing going
00:44:08
on, which is a straw manning of what the
00:44:12
Trump administration did on AI in its
00:44:15
first year. And in the same way that
00:44:18
they want to spin this FDA for AI,
00:44:20
they're also trying to spin what we did
00:44:22
as this completely lazy fair attitude
00:44:25
where there'd be no regulations
00:44:26
whatsoever, no guardrails. It's a way of
00:44:29
criticizing what we did. They're trying
00:44:31
to portray it as unsafe. In fact, if you
00:44:33
look on March 20th, the White House
00:44:35
released a national AI regulatory
00:44:38
framework that I worked on in which we
00:44:41
put out a four-page bulleted list of
00:44:43
legislation that we would support if
00:44:46
Congress wants to pass it. So, we have
00:44:49
not been against every conceivable
00:44:52
regulation or every conceivable law. We
00:44:54
just believe that there should be
00:44:56
specific solutions to specific problems
00:44:59
as opposed to a giant power grab by
00:45:01
Washington that would squash innovation.
00:45:03
So I think that's point number two.
00:45:06
Point number three is there is a
00:45:08
legitimate thing happening here with
00:45:10
let's call it mythos or cyber. Okay, we
00:45:13
know that it's not just mythos. Open AI
00:45:17
now has a model that's just as cyber
00:45:19
capable as mythos. And within 3 to six
00:45:22
months, all the major frontier labs and
00:45:24
including Chinese models will have cyber
00:45:27
capabilities.
00:45:28
In response to that, we do need there to
00:45:31
be a hardening of systems and we do need
00:45:35
there to be a scanning of code bases to
00:45:37
find these vulnerabilities and patch
00:45:40
them before the hackers do it because
00:45:42
the hackers will have these capabilities
00:45:44
in a matter of months. That's a
00:45:46
certainty because the same capabilities
00:45:48
you use for cyber defense can also be
00:45:50
used for cyber offense. It's the same
00:45:52
tool set and the open source models will
00:45:55
have these capabilities anyway.
00:45:56
>> They already have it to a certain
00:45:57
extent. Let's be honest, they have 80%
00:45:59
of it.
00:46:00
>> It's simply the case that AI will be
00:46:02
good at cyber and so we do need a
00:46:04
response to that. Now my view on what
00:46:07
should that response be should be first
00:46:10
of all we should want the government and
00:46:13
the private sector to work cooperatively
00:46:15
and I think they are we have a giant
00:46:18
cyber security industry in the United
00:46:20
States whose sole job it is to protect
00:46:23
systems and protect against breaches.
00:46:26
>> We have the best companies in the world
00:46:28
at doing that. We have Crowd Strike. We
00:46:30
have Palo Alto Networks. We talked about
00:46:32
that before. We have the best defense.
00:46:34
>> Right. Exactly. And so what we should be
00:46:36
doing I think is getting these tools
00:46:38
mythos and then the open AI model and
00:46:41
and others like it in the hands of our
00:46:43
cyber security industry. And by the way
00:46:45
not just the public companies like Palo
00:46:47
Alto Networks and Crowd Strike although
00:46:49
certainly they're two of the most
00:46:50
noteworthy but there's also some
00:46:52
incredibly strong startups on the way up
00:46:55
there that are at the cutting edge of
00:46:57
doing AI powered pen testing and all the
00:47:00
rest of it. We need to get these tools
00:47:02
into their hands as quickly as possible
00:47:03
because they're a force multiplier for
00:47:05
all the companies out there that aren't
00:47:07
that good at cyber security or maybe
00:47:10
they've got IT departments. They can use
00:47:12
these companies as vendors. So I think
00:47:14
that there is a role for
00:47:15
>> Can I ask you a question? Yeah.
00:47:16
>> Do you think that the models should
00:47:20
have a KYC rapper going forward?
00:47:22
>> KYC for the audience is know your
00:47:25
customer.
00:47:25
>> Yeah. So really what it would mean is
00:47:27
that before you can use mythos you have
00:47:30
to identify yourself so that we can try
00:47:32
to know that you're not a state
00:47:35
sponsored actor or you know a bad guy.
00:47:38
>> I think that's the type of thing that we
00:47:39
should be thinking about. So first of
00:47:41
all I want to say that both anthropic
00:47:43
and open AI acted responsibly here. No
00:47:46
one was trying to release these super
00:47:48
powerful models. So, in a way, all the
00:47:51
people who are saying that we need
00:47:52
pre-release approvals for models,
00:47:54
they're trying to solve a problem that
00:47:55
didn't exist regulated.
00:47:57
>> Yeah, they probably wasn't trying to
00:47:58
release. I wasn't trying to release
00:48:00
this. They all understood the power and
00:48:01
they were all acting responsibly.
00:48:03
>> They understood the ramifications. They
00:48:05
would have been sued. So, they there is
00:48:06
a self-p policing going on here, which
00:48:08
is the ultimate way to do this.
00:48:10
>> Yes. And but to your point, Jimoth,
00:48:12
yeah, look, I think that before giving
00:48:15
your API for a super powerful model, you
00:48:17
should not give that to a company or an
00:48:19
actor. You don't know who they are. So,
00:48:21
yeah, some basic KYC makes sense. They
00:48:23
should know who they're giving these
00:48:24
tools to. And I guess my view on the
00:48:27
mythos preview and whatever the
00:48:28
equivalent is of what OpenAI is doing is
00:48:32
that we very rapidly need to get these
00:48:34
tools into the hands of more good guys.
00:48:36
You need to know who those good guys
00:48:38
are. You need to know who they are. So
00:48:40
yeah, KYC is like a predicate for that,
00:48:43
right? You got a new
00:48:44
>> just to be clear, we'd all agree that if
00:48:46
you did have identity for those frontier
00:48:49
models, which they're probably doing
00:48:50
anyway right now, and you logged what
00:48:52
people were doing with them to look for
00:48:53
security breaches, that wouldn't
00:48:54
necessarily happen when you released it
00:48:56
to the public because of privacy issues.
00:48:59
Here's your Poly Market for Trump
00:49:01
ordering a federal review of AI models
00:49:03
by May 31st. 21% chance I think uh to
00:49:07
our partner at Poly Market. Man, I I got
00:49:09
to get in here. I don't Do I have inside
00:49:11
information here being uh the world's
00:49:13
greatest moderator on this podcast or I
00:49:15
can Can I collect this money? Chimoth,
00:49:16
what am I going to do here?
00:49:17
>> I would not do not do not place a bet.
00:49:19
Jal,
00:49:20
>> don't place a bet.
00:49:21
>> Don't place a bet. But anyway, Chim's
00:49:23
point, I mean, look, I think we're kind
00:49:25
of workshopping this in real time.
00:49:26
>> We are.
00:49:27
>> I think that I think that for the
00:49:28
preview period, we should definitely
00:49:30
have KYC.
00:49:31
>> Maybe.
00:49:31
>> What about logging? What about logging?
00:49:33
Well, look, once you're past the preview
00:49:35
period and it's in general release, I'm
00:49:38
not sure if the KYC matters as much
00:49:39
because so many people are going to have
00:49:40
it. But during the preview period, there
00:49:42
should be KYC.
00:49:43
>> Let me let me just say one thing. All
00:49:45
the labs are already tracking API use.
00:49:48
Okay? And anything suspicious because
00:49:50
they there there are major
00:49:52
anti-distillation efforts going on by
00:49:54
all the labs. There's a ton of
00:49:56
coordination going on with the
00:49:57
government. there's way more happening I
00:50:00
think in terms about API and API use um
00:50:03
and anything suspicious is being flagged
00:50:06
and being shared with the government. So
00:50:07
the idea that we have no idea who's
00:50:09
doing it I think is not the case and in
00:50:12
fact in some cases we may want to allow
00:50:15
people to use it so that we can see
00:50:17
exactly the types of things that they
00:50:19
are extracting. So I would just I would
00:50:21
say we're already down that path but
00:50:23
better coordination
00:50:25
may may in fact be called for.
00:50:27
>> Yeah. And just one last point in this
00:50:28
whole thing is I just want to build on
00:50:30
my point that pre-release approvals is
00:50:32
solving a problem that didn't really
00:50:33
exist because again anthropic and open
00:50:35
AI weren't trying to release these
00:50:37
models yet is that there is a
00:50:39
substantial faction of let's say AI
00:50:42
idologues or doomers who
00:50:45
are basically employing the classic
00:50:47
never let a crisis go to waste strategy
00:50:50
right that yes we do have this cyber
00:50:53
issue that is real you know everyone
00:50:55
needs to harden their systems now
00:50:56
>> totally over the next 3 to 6 months.
00:50:58
That is a real issue. But that is a
00:51:01
problem that we will solve over the next
00:51:02
6 months. We have to. But what they're
00:51:04
trying to do is use that issue to try
00:51:08
and create a permanent new
00:51:09
infrastructure in Washington. Again, I
00:51:11
don't think the admin that's not the
00:51:12
administration's intention. That's not
00:51:14
the administration's agenda. But you saw
00:51:16
a lot of people on social media, a lot
00:51:19
of the think tanks and even Bernie
00:51:21
Sanders weighed in and he said for the
00:51:23
first time, I like something that the
00:51:25
Trump administration wants to do.
00:51:26
>> The administration understands the 1% of
00:51:29
the 1% of Sachs and everybody understand
00:51:31
that this is out of control. The AI is
00:51:33
going to take the jobs. They're going to
00:51:34
take my summer home. It's going to be
00:51:36
terrible.
00:51:37
>> So, there are people who have this
00:51:38
agenda. Look, Bernie Sanders just wants
00:51:40
to stop the progress. I mean, he's
00:51:43
>> he wants to ban data centers. He's put
00:51:44
out a bunch of
00:51:46
He he basically has bought into the
00:51:47
whole doomer narrative. So look, that's
00:51:50
why he likes the FDA idea is because it
00:51:54
would put the kibos on innovation.
00:51:56
>> It's enough already. Let's give let's
00:51:57
have a go back to paper and pen. It was
00:51:58
a better society. Saxs
00:52:00
>> Jason, what do you think?
00:52:01
>> I think there's two really interesting
00:52:03
things I want to build on here. The
00:52:04
first is your point Shamath around how
00:52:07
do we turn around the sort of bad vibes
00:52:11
around AI? I think we have to have two
00:52:14
strategies here. One is giving what
00:52:16
you've been working on Brad with your
00:52:18
project we should see more people
00:52:19
giving. There's no reason why Nvidia
00:52:22
SpaceX when they go public
00:52:25
anthropic when they go public open AAI
00:52:27
if and when they go public or if they
00:52:29
become stay as nonprofit there's no
00:52:31
reason those folks in an IPO couldn't
00:52:33
give a portion of the IPO to every
00:52:35
American citizen. So IPO K, IPO for
00:52:38
kids, they all take, you know, whatever
00:52:40
it is, 5%, 1%, whatever they choose, and
00:52:43
they put it into the Invest America
00:52:45
accounts and we should see some major
00:52:48
giving from the people who are becoming
00:52:50
trillionaires, 100 billionaires,
00:52:52
whatever it happens to be. There's no
00:52:53
reason not to. But those people haven't
00:52:55
been doing that. We had this giving
00:52:56
pledge which was a little bit of virtue
00:52:58
signaling and it wasn't real. It was
00:53:00
just, you know, at the end of your life,
00:53:01
you promise to give away half your
00:53:02
money. So let's have something real.
00:53:04
Let's have something where you know
00:53:05
people say I'm going to give away 1% of
00:53:07
my stock over the next 20 years of my
00:53:09
life every year 1% will go into invest
00:53:11
America whatever it is it won't cost
00:53:13
anybody anything you can't spend this
00:53:15
money whether it's Bezos or whoever
00:53:17
second in that same thing in terms of
00:53:19
giving back we have not talked about how
00:53:22
massive this could be for health and
00:53:24
extending people's life and reducing
00:53:26
suffering we need to work on that that's
00:53:28
where contributions to basic science
00:53:29
could come in and obviously education
00:53:32
and lowering the cost of education And
00:53:34
if you look at what Americans on the
00:53:36
bottom half, you were talking about the,
00:53:38
you know, cup half empty, there's really
00:53:40
two or three things they really feel
00:53:42
anxiety about. One of it is income and
00:53:45
the second is healthare and on the
00:53:48
margins, housing and their kids, you
00:53:50
know, their kids' education, the cost of
00:53:52
those things. We should really take a
00:53:53
look deeply at, and I know this is very
00:53:56
unpopular amongst capitalists, including
00:53:58
myself. We should really look at the
00:53:59
minimum wage and study what happened in
00:54:01
New Zealand, Sweden, Switzerland, uh,
00:54:04
Australia when they raised it. What
00:54:06
actually happened when they raised it,
00:54:07
and there was a lot of hand ringing
00:54:08
about it, but when they slowly raised
00:54:09
it, what they found was those consumers
00:54:11
don't save money. They spend it. They're
00:54:13
always behind the eight-ball in terms of
00:54:15
their spending. We should opt in to
00:54:17
trying to raise the minimum wage company
00:54:20
by company by company. and just give
00:54:22
people who are at the end of the
00:54:24
spectrum that understanding that, hey,
00:54:26
year-over-year, whether it's Amazon or
00:54:29
Target, etc., restaurants, we're all
00:54:32
collectively going to add a little bit
00:54:34
to that minimum wage and try to lift the
00:54:36
bottom third of society. That's the
00:54:38
stuff we're not talking about. We don't
00:54:40
talk about it here on this podcast. We
00:54:41
don't talk about universal healthcare.
00:54:43
We don't talk about the minimum wage,
00:54:44
but that's what capitalists should be
00:54:46
talking about. And if we did that, if we
00:54:48
increased the minimum wage, and I'm not
00:54:50
a socialist. I'm a capitalist who think
00:54:51
this is good for capitalism. If we
00:54:53
increased the minimum wage just modestly
00:54:55
each year and we opted into doing that
00:54:58
and we figured out a way to give
00:54:59
universal healthcare, companies wouldn't
00:55:00
have to deal with universal healthcare
00:55:02
and we would have customers and we're a
00:55:04
customer-driven economy. Like 60 70 80%
00:55:07
of what happens in this country is
00:55:08
driven by the consumer. We need consumer
00:55:10
spending. It's great for companies if we
00:55:13
had more people being able to buy
00:55:15
Netflix or order [ __ ] on Amazon. Anyway,
00:55:18
that's my that's my TED talk. Thanks for
00:55:19
coming. How do we get from AI to the
00:55:22
minimum wage? I'm still
00:55:23
>> No, because of the black eye. No, the
00:55:25
black eye we have in this country with
00:55:27
polarization of wealth and people scared
00:55:29
of losing their jobs. We We should look
00:55:31
at why are they scared, David. And I've
00:55:33
talked to you privately and you said to
00:55:35
me privately, you can strike this if you
00:55:36
want, but you said to me privately, you
00:55:38
wouldn't be against necessarily figuring
00:55:41
out a way to un do universal healthcare
00:55:43
if there was a way to do it. You want to
00:55:45
see every human have healthcare. Yes.
00:55:48
>> Sure. Sure. I mean the issue is not the
00:55:50
desiraability of it. It's the cost. I
00:55:52
mean
00:55:53
>> and so you're a great entrepreneur of
00:55:55
our time. How would you do it? Have you
00:55:56
given any cycle to it?
00:55:57
>> I just I haven't studied that issue. So
00:55:59
I don't know. I just know.
00:56:01
>> How about you study the country's
00:56:03
>> I remember what PJ Oor once said which
00:56:05
is if you think healthcare is expensive
00:56:07
now just wait until you make it free.
00:56:10
>> Yeah.
00:56:11
>> So you take away all the incentives and
00:56:13
you have an even bigger problem.
00:56:15
>> Minimum wage. Yeah. Go ahead. Well, let
00:56:16
me just please get back to AI. Listen,
00:56:19
you guys are right about the
00:56:20
unpopularity of AI. We've all seen those
00:56:22
polls, but I want to just put up this
00:56:24
additional poll that came out about the
00:56:27
salience of this issue, which is how
00:56:29
important do people think it is? And AI
00:56:32
ranked 29 out of 39. So although AI is
00:56:35
not very popular, it is certainly not
00:56:37
top of mind for voters. It's not in the
00:56:39
top 10 issues. It's not in the top 20
00:56:41
issues. What is top of mind for voters?
00:56:43
Number one, cost of living. Number two,
00:56:45
the economy. And we know that AI is
00:56:48
deflationary. It helps with the cost of
00:56:50
living and it's creating an economic
00:56:53
boom right now. Okay? It's 75% of GDP
00:56:57
growth in Q1. By the way, that that
00:56:59
economic growth is not just limited to
00:57:01
startups in Silicon Valley. We're seeing
00:57:02
a construction boom. We're seeing a blue
00:57:04
collar boom. We're seeing 25 to 30% wage
00:57:07
increases for uh construction workers.
00:57:09
>> Absolutely.
00:57:10
>> And so on down the line. So,
00:57:12
>> and Brad, if you if you look at that
00:57:13
chart, there's healthcare in there, too.
00:57:15
>> So, look, my point is that AI may not be
00:57:17
popular, but the effects of it actually
00:57:20
are popular if the media would honestly
00:57:22
report what was happening, which is AI
00:57:25
is creating an economic boom right now.
00:57:27
>> It couldn't be better said, David. You
00:57:29
know, Bernie Sanders calling for a
00:57:31
moratorium, shutting out all data data
00:57:33
centers. We'd have negative GDP growth
00:57:35
this year. The stock market would be
00:57:37
down 15 to 20%. Unemployment would be on
00:57:40
the rise. You know there is a
00:57:42
consequence to the government
00:57:43
controlling through command to control
00:57:45
the economy. In 1929 we had 4%
00:57:48
unemployment. 3 years later it was
00:57:50
unemployment was 23%.
00:57:52
Because government got involved in
00:57:54
regulating everything and shutting down
00:57:56
you know what was working. That is the
00:57:58
greatest threat we have here. AI is
00:58:00
delivering huge net benefits today in
00:58:02
terms of unemployment rate in terms of
00:58:04
economic growth and productivity growth.
00:58:06
We need to tell the story. But to Chamas
00:58:08
earlier point, we also need to deliver
00:58:10
net benefits. Jason, thanks for the
00:58:12
shout out. Yes, every American having an
00:58:14
investment account that compounds with
00:58:16
the upside of AI. We're going to do
00:58:18
that. Going to deliver that and that's
00:58:20
going to be massive. But I also think
00:58:21
ideas like if we're going to put a data
00:58:24
center in Abalene, Texas, let's make
00:58:25
electricity in Abalene free for the
00:58:27
households in Abalene, Texas. Right?
00:58:29
There are ideas that can deliver net
00:58:31
benefits. We got to deliver those. I
00:58:33
think we're optimism will be on the
00:58:35
march. I think we're in the trough right
00:58:37
now.
00:58:37
>> What's your take on minimum wage and
00:58:39
universal healthcare? Brad, how do you
00:58:41
think about it as a capitalist, as an
00:58:43
innovator, and in the face of AI, which
00:58:45
could have a dramatic impact on these
00:58:46
issues?
00:58:48
>> You know, like Zach, I, you know, to me,
00:58:50
I haven't spent a ton of time thinking
00:58:52
about those except that as a society,
00:58:54
we're $38 trillion in debt. We haven't
00:58:56
been able to afford to deliver those
00:58:58
things. I generally think the market
00:58:59
works out those issues better than the
00:59:02
government top down trying to you know
00:59:04
the government gets more and more
00:59:05
involved in healthcare and the only
00:59:06
thing that happens it gets more
00:59:07
expensive. So we've all seen the charts
00:59:09
of the most expensive categories where
00:59:10
we've had inflation education healthcare
00:59:13
etc. It's where the government's
00:59:14
involved. I actually think if you just
00:59:16
let the markets work entering an into an
00:59:18
age of abundance a lot of these problems
00:59:20
are going to be solved. People are going
00:59:21
to have a lot of AI coaches in
00:59:24
healthcare and education etc. Let the
00:59:26
market work. government stay at bay,
00:59:28
keep things safe. We're on a good march.
00:59:31
>> I think this proves my point perfectly.
00:59:33
If you talk to any founder, they're not
00:59:35
thinking about housing. They're not
00:59:37
thinking about higher education costs.
00:59:39
They're not thinking about minimum wage,
00:59:41
you know, and they're not thinking about
00:59:42
healthcare all that often. Some some do,
00:59:44
though. There's some innovation there.
00:59:45
And it's because it's so regulated that
00:59:48
entrepreneurs and VCs are just like,
00:59:49
"That's kryptonite. The government has
00:59:51
poisoned the well. We can't participate
00:59:53
in that." And that's the roadblock. And
00:59:54
that's where Americans are suffering.
00:59:56
And that's where it would be great if
00:59:58
founders actually put their minds to it.
00:59:59
And the government's got to get rid of
01:00:01
all that regulation and let us cook in
01:00:03
those specific verticals. All right. The
01:00:05
market is in hyperdrive. Hyperscaler
01:00:08
revenue has made the markets move up. We
01:00:12
hit on this briefly, but we didn't have
01:00:14
you here. Fifth best EBG. Cloud
01:00:17
computing on a tear. I referenced it
01:00:19
earlier, but AWS is now on $150 billion
01:00:22
run rate. Azure 108 billion. GCP, Google
01:00:26
Cloud, 80 billion. There's a little bit
01:00:27
of fun with numbers there because Azure
01:00:29
and Microsoft include some of their
01:00:30
software products in there and Google
01:00:31
Cloud includes things like Google Office
01:00:33
or Google Suite in there. But the growth
01:00:36
numbers are tremendous. AWS, which is
01:00:39
the more pure play of the three, 28%
01:00:42
growth on a very big number. Azure 39%,
01:00:45
Google Cloud stunning everybody with 63%
01:00:49
growth. It is incredible what the ARR
01:00:52
numbers are. Google Cloud Edit 10, AWS
01:00:55
10, Azure 9.5. So basically 30 billion
01:00:57
collectively. Jammond Ball who works for
01:00:59
you uh I think put out some data on the
01:01:03
Twitter. Brad markets at all times high
01:01:06
all-time highs. Mag 7 cooking
01:01:11
Uber blowing out growth. Disney blowing
01:01:14
out growth. The consumer seems absurdly
01:01:17
strong based on those two bell weathers.
01:01:20
Tech seems extremely strong based on the
01:01:22
cloud computing. What's your take on the
01:01:24
overall market and overall economy?
01:01:26
Obviously, inflation up a bit, people
01:01:28
hand ringing about the never-ending war
01:01:30
and the cost of oil.
01:01:32
>> Let's just telescope way up. You know,
01:01:35
the level of criticism directed to this
01:01:37
administration, right? Tariffs were
01:01:39
going to cause hyperinflation. We're
01:01:41
going to destroy GDP. Uh conflicts in
01:01:44
Venezuela and Iran were going to do the
01:01:46
same. You know, we've heard all of the
01:01:48
negative stories, but what's happening?
01:01:50
Accelerating GDP, a tenure that's
01:01:53
sitting at 43, inflation totally under
01:01:56
control. AI, AI, AI, comput, compute,
01:02:00
compute. We're leading the world. It's
01:02:02
contributing massively, right, to GDP
01:02:05
growth in the country. We see the S&P
01:02:07
only up 8% this year, right? So, we're
01:02:09
not into bubble territory here. Meta's
01:02:12
trading at 17 times, fully tax gap
01:02:14
earnings. Nvidia at 19 times, Microsoft
01:02:17
at 20 times, Google at 24 times, and
01:02:20
then the memory stocks that everybody's
01:02:22
excited about. We have 25% of our
01:02:24
portfolio in SKH Highix five times fully
01:02:28
taxed gap earnings. Samsung six times,
01:02:31
Micron 7 times. Right? This is not the
01:02:34
stuff that bubbles are made of. You
01:02:36
know, David referenced it earlier. We
01:02:38
started the year Open AI and Anthropic
01:02:40
were doing combined about 30 billion in
01:02:43
revenue. Now combined 4 months later 80
01:02:46
billion in revenue. The the policies of
01:02:49
this administration on the economy are
01:02:52
working. They're working in in spades.
01:02:54
Our gap on the rest of the world in AI
01:02:56
is growing. And so from from my
01:02:59
perspective, you know, uh you know,
01:03:01
we've been all in on the market. I
01:03:02
talked about it earlier in the year.
01:03:04
We're heavily.
01:03:05
>> When did you make that switch to go all
01:03:06
in on the market? Because you were
01:03:07
bearish.
01:03:09
>> I would say toward the toward the end of
01:03:10
last year. The market had run up a lot.
01:03:12
We had a lot of these questions. Listen,
01:03:14
entering this year, there was a huge
01:03:16
question hanging over the market. Would
01:03:19
the AI revenues show up? If the
01:03:22
anthropic revenues hadn't shown up and
01:03:24
we didn't see this reaceleration out of
01:03:26
the hyperscalers, the market would be
01:03:28
down 10 to 15%. Because people would say
01:03:30
there's no ROI on all of this
01:03:34
infrastructure. Exactly. When I saw the
01:03:36
numbers start showing up in December and
01:03:39
into January, we went from medium to
01:03:41
large in terms of our exposures and 80%
01:03:44
of our exposures or more have been in
01:03:47
compute, AI, memory, etc.
01:03:49
>> And this is why it's great to operate in
01:03:50
the private market and the public market
01:03:52
because you can see things in the
01:03:53
private markets that inform the public
01:03:55
markets. But the question remains, Brad,
01:03:57
how much better would the economy have
01:03:59
been doing? you know, as much credit as
01:04:02
you're giving to the administration if
01:04:04
they didn't start a hundred billion
01:04:05
dollar war that we did not need to go
01:04:07
into according to all reports. And if we
01:04:10
didn't do a bunch of tariffs that wound
01:04:13
up being unconstitutional by the Supreme
01:04:15
Court, which Trump himself put in, we
01:04:18
would have been further ahead. That's uh
01:04:20
my take on it. We would be ripping even
01:04:22
more if we didn't have those uh
01:04:24
>> Jason, it's hard to it's it's hard to
01:04:26
imagine. Okay, just to set up again
01:04:28
here. It's hard to imagine a more
01:04:31
goldilock situation for the United
01:04:33
States of America. We have reset the
01:04:35
table geopolitically. The discount rate
01:04:38
globally is actually coming down, not
01:04:40
going up. Evidenced by markets at
01:04:42
all-time high and the bond market in
01:04:44
control. And then look at the private
01:04:46
markets. We have multiple trillion
01:04:48
dollar companies that have been created
01:04:50
in the private markets that are now
01:04:52
coming public. SpaceX coming public is
01:04:54
going to be a multi-trillion dollar, you
01:04:56
know, open AI anthropic. Like at some
01:04:59
point, you just have to acknowledge USA
01:05:01
is winning. Of course, there are always
01:05:04
things that we could be doing better,
01:05:05
but there's not a country on the world
01:05:07
that wouldn't trade all of its fortune
01:05:10
for the United States fortune today.
01:05:12
>> 100% in agreement, American
01:05:13
exceptionalism as embodied by the great
01:05:16
companies in America, SpaceX, Google,
01:05:19
etc., all the ones we've been talking
01:05:20
here. That is the story. And I give
01:05:23
infinite credit to this administration
01:05:24
for being businessfriendly. I do think
01:05:26
they've made two critical mistakes. I
01:05:28
think the tariffs were poorly executed
01:05:29
and I think we shouldn't have gone to
01:05:30
this war. Um, and we should find a quick
01:05:33
resolution to it, which the
01:05:34
administration seems to be desperately
01:05:36
doing. Sax, your take on the economy?
01:05:39
>> Well, look, we have an AI boom going on
01:05:40
right now, and I think that's thanks to
01:05:42
President Trump's policies. Remember the
01:05:43
first week he was in office, he
01:05:45
rescended the Biden policies on chips
01:05:47
and models. And what were those
01:05:49
policies? It was the approval regime
01:05:51
that we're talking about. Models would
01:05:54
have to go to Washington to get approved
01:05:55
if they were trained with some number of
01:05:57
flops. And then every sale of a GPU
01:06:00
worldwide would have to be licensed from
01:06:02
Washington unless it fit into some
01:06:04
narrow exemptions. So the whole approach
01:06:07
of the Biden administration that
01:06:09
President Trump inherited was everything
01:06:11
approved in Washington. He rescended
01:06:14
that. He declared that we had to win the
01:06:15
AI race and he unleashed uh our
01:06:17
companies to do that. Now one other
01:06:19
really important thing is energy.
01:06:21
Remember it was this president going
01:06:23
back a decade who said drill baby drill.
01:06:25
He said we have to unleash American
01:06:27
energy. That's the basis for the
01:06:28
American economy. It's also the basis
01:06:30
for AI. He also has said that he wanted
01:06:34
to allow our AI companies to become
01:06:36
energy companies so they could bring
01:06:38
their own power to these data centers.
01:06:40
So they're not drawing off the grid.
01:06:41
They're not competing with consumers for
01:06:43
electricity. They're generating their
01:06:45
own power. And it's thanks to this
01:06:47
president that we have seen this
01:06:50
bluecollar construction boom right now
01:06:53
powering all of this infrastructure.
01:06:54
What would the alternative have been? We
01:06:57
know I mean Bernie Sanders has said it
01:06:59
would have been ban on data centers. So
01:07:01
>> yeah raise tax data centers. It's that
01:07:04
would be a much a much worse choice. Uh
01:07:07
Chamal, I'll give you the last word here
01:07:08
as we wrap on the economy generally. I
01:07:12
think the markets are going to keep
01:07:13
going up for a while
01:07:15
and then at some point they're going to
01:07:16
go down.
01:07:19
>> Okay, I wrote it down. Uh, Chimath, you
01:07:21
said markets are up and then eventually
01:07:24
they're going to come down. Those are
01:07:26
the two things up and we put a U here
01:07:28
and then a D here.
01:07:29
>> He's got the I think you're doing an
01:07:31
impression of the taking notes emoji. Is
01:07:34
that what you're doing?
01:07:34
>> Yes. Okay. Ups and down. Wow. Thank you
01:07:38
for tuning in to AllIn where you can get
01:07:40
your great calls and market action
01:07:42
advice. It's going to go up and then
01:07:43
down folks act accordingly. Go Chamat in
01:07:46
all seriousness but what what what makes
01:07:47
you bullish you know let's say in the
01:07:50
next uh 6 months 12 months catalyst and
01:07:52
then what do you think the headwinds are
01:07:53
as well? Let's take the you know the
01:07:54
short to midterm 6 months to two years.
01:07:58
I think that in the short term the
01:08:01
people that makes the new thing
01:08:06
needs to get valued and needs to
01:08:10
demonstrate value. So who are the people
01:08:12
making the new thing? It's the Nvidias,
01:08:15
it's the memory makers, it's the
01:08:16
anthropics, it's the SpaceX's and uh
01:08:19
it's the open AIS. But eventually it all
01:08:21
comes home to roost. And
01:08:25
you can't just make things for a market
01:08:28
who then doesn't have a measurable
01:08:30
benefit themselves.
01:08:32
To be very clear and blunt, there is
01:08:34
literally not a cintilla of evidence
01:08:37
that AI has helped lift the operating
01:08:39
margins of the S&P 500. There's all
01:08:42
kinds of bluster. There's going to be an
01:08:44
important fork in the road. It's
01:08:46
probably two or three years from now.
01:08:48
One path will be opex shrinks, hence
01:08:52
margins increase and the other path is
01:08:54
revenues grow and margins expand and
01:08:57
opex stays flat but or maybe it even
01:08:59
goes up. Those two things are very
01:09:01
important differences because in the
01:09:03
former you're talking about shrinking
01:09:06
workforce and shrinking opex as a
01:09:07
percentage of operating margin and
01:09:10
revenue. In the latter you're actually
01:09:11
growing through it. The answer to that
01:09:13
question I think is critical about how
01:09:15
the markets will respond and how society
01:09:18
will respond. So I think we have kind of
01:09:20
call it 500 days where you just got to
01:09:24
be net long. But I think it's literally
01:09:28
in the hundreds of days from now 500
01:09:33
you're going to have to have an
01:09:34
important reckoning moment. The people
01:09:37
that are paying for all these tokens
01:09:38
need to see an actual benefit.
01:09:41
>> Yeah. That that's reasonable. Yeah,
01:09:42
>> that's a really interesting point. Let
01:09:43
me connect a couple of dots here between
01:09:45
something Jamal said and what Brad said,
01:09:46
which is Brad said at the beginning of
01:09:48
the year, we went into this year with
01:09:50
this massive capex, this massive
01:09:52
investment infrastructure, but people
01:09:54
weren't sure that the ROI was going to
01:09:56
be there in terms of model revenue. And
01:09:59
that was true and then the model revenue
01:10:00
has proven out. And now what Schmoth is
01:10:03
saying is that we're going to be at
01:10:04
another fork in the road soon in terms
01:10:05
of whether there's going to be ROI on
01:10:08
all those tokens that are being sold and
01:10:11
generating the revenue for those model
01:10:13
companies. And I agree with you that
01:10:15
that is not proven out yet. But I'm
01:10:17
optimistic that it is going to be proven
01:10:20
out and otherwise these you wouldn't be
01:10:22
seeing
01:10:23
>> look I
01:10:23
>> enterprises continue to to buy hold on
01:10:26
let me just point you wouldn't be seeing
01:10:28
enterprise continue their month
01:10:29
overmonth spend on coding tokens if they
01:10:33
didn't feel like the ROI was going to be
01:10:34
there but you make a good point which is
01:10:37
what is the impact on the economy going
01:10:38
to be when all of this new software this
01:10:42
bespoke software that's being created
01:10:44
through again all these coding tokens
01:10:45
are being bought It's going to power a
01:10:47
wave of productivity like I think we've
01:10:49
never seen before. So I think what
01:10:51
you're seeing is the ROI is getting is
01:10:54
sort of trickling down from
01:10:55
infrastructure to model to application
01:10:58
to end user and I think it's going to
01:11:00
create an economic boom.
01:11:02
>> I I got to I'm with you Saxs. This is
01:11:04
deja vu all over again. We watch this
01:11:06
happen with the PC revolution, the
01:11:08
internet revolution, cloud revolution,
01:11:10
mobile revolution. We had all this hand
01:11:11
ringing. Will this ever pay off? Should
01:11:12
I build an app? Should I build a
01:11:14
website? Should I not? Should I move to
01:11:15
the cloud? Should I keep it on prem? All
01:11:17
of these questions over and over and
01:11:18
over again. And then they went from
01:11:20
question marks to exclamation points. I
01:11:21
can tell you inside of my firm, we have
01:11:23
started, we were using agents, then we
01:11:25
started building code. And I've got
01:11:26
three people on the team who are making
01:11:28
all the interfaces and products that a
01:11:30
22 person investment firm should not be
01:11:33
making internally. They should be using
01:11:35
SAS software. And they are shipping
01:11:38
product day in and day out. The ROI is
01:11:40
fate complete. Brad, it is fat or
01:11:43
complete. I think this has been decided.
01:11:45
I think it's been decided.
01:11:46
>> It has not been decided at all. It has
01:11:48
not been decided at all.
01:11:49
>> You have 8090. You're working with the
01:11:51
big enterprises. I invest in 100
01:11:52
startups a year. I work with the small
01:11:54
ones. It is fate a complete with
01:11:56
startups. They are building software.
01:11:57
They're shipping. They are getting
01:11:58
massive value from these tokens. And
01:12:00
they're getting so much value that they
01:12:02
don't have to add, you know, but half
01:12:04
the number of employees that they would
01:12:06
with the same amount of capital. They're
01:12:08
getting further with less money. It is
01:12:10
working in startup land. I don't know
01:12:12
what's happening at 8090. you would have
01:12:14
a better picture obviously of the
01:12:15
enterprise. Tell us what you see there.
01:12:16
>> I mean, our business is doing well, but
01:12:20
what I'm trying to get across to you
01:12:22
guys is that you can't will profits to
01:12:26
go up.
01:12:28
Okay? So, ultimately what happens is I'm
01:12:31
just going to take a company randomly.
01:12:33
Annheiser Bush, they have to eventually
01:12:37
sell more beer.
01:12:39
Okay? Take Nike. they ultimately have to
01:12:43
sell more shoes. Take a medical devices
01:12:46
company. They have to sell more
01:12:47
artificial hips and knees. So the point
01:12:50
I'm trying to get across is right now
01:12:52
there's an enormous amount of very
01:12:54
constructive and creative
01:12:56
experimentation.
01:12:58
But I think it's what is also true is a
01:13:02
lot of that has not yet proven value. I
01:13:05
don't think that means it's going to
01:13:06
stop. All I'm just trying to say is
01:13:09
until a company can trace very directly
01:13:12
I spent X and I made Y where Y is now
01:13:14
greater than X and it's lifted my
01:13:16
margins.
01:13:18
That is the thing that causes the
01:13:20
flywheel to spin faster.
01:13:22
And right now we've started the first
01:13:24
part of that equation. We've spent the X
01:13:28
and we have not seen the Y.
01:13:31
You would see it in global GDP. You
01:13:33
haven't. You would see it in global
01:13:35
productivity. You haven't. You would see
01:13:37
it in the global profit margins of the
01:13:38
S&P 500. We haven't. It doesn't mean
01:13:41
it's not coming.
01:13:41
>> Brad, you want you want to pick up on
01:13:43
this because yeah, I'm definitely taking
01:13:44
the other side of it because I'm seeing
01:13:46
with a lot of these companies,
01:13:48
>> massive lowering of costs. Their ads are
01:13:50
getting more effective. At the same
01:13:52
time, they've stopped hiring. They're
01:13:53
not adding positions in a lot of cases.
01:13:55
And things like, just picked the Nike
01:13:58
example, a lot of the photo shoots they
01:14:01
used to do for their app, excuse me. A
01:14:02
lot of the imagery they used to make,
01:14:04
now they're able to make more of it
01:14:06
without having to hire photographers and
01:14:08
do that stuff. I know this example
01:14:09
because we have a startup that does this
01:14:10
specifically for brands like Nike.
01:14:12
They've seen a massive drop. We have one
01:14:14
that helped Door Dash with their food
01:14:16
pictures. All those pictures used to
01:14:17
have to be taken by photographers. Now
01:14:19
it's all done by AI. Massive reduction
01:14:20
in cost. And they're using ads and ad
01:14:23
creative now. that is, you know, you
01:14:26
know, double digit percentage more
01:14:28
effective while costing half as much.
01:14:30
So, I definitely think we're seeing it
01:14:32
on the earnings, but is is that true?
01:14:34
Are you seeing it in the earnings of
01:14:35
these companies yet?
01:14:36
>> Yeah. So, so two data points. Number
01:14:38
one, we just saw Azure grow 39% in the
01:14:42
quarter. We saw Google Cloud grow 63% in
01:14:45
the quarter. Headcount growth for those
01:14:47
companies the last 3 years, Mag 5
01:14:48
combined is about 3%. So, their
01:14:51
operating margins are all expanding. If
01:14:54
you look at the S&P 500 at large, in Q1
01:14:57
of 24, operating margins were about
01:14:59
11.8%.
01:15:00
That was up from 11% in 23. This year,
01:15:04
they're 13%. So, we've had a 200 basis
01:15:06
point improvement in the operating
01:15:09
margins of the S&P 500, which is
01:15:12
massive.
01:15:12
>> Do you think that's AI across those
01:15:14
businesses? And Chimath, I think that
01:15:16
that's the question where where it could
01:15:20
donuts, it's not AI.
01:15:22
>> Yeah. So, so any amount of money. I bet
01:15:24
you are the same financial engineering
01:15:26
that got these earnings to rise in the
01:15:28
last decade.
01:15:29
>> Yeah. So I I I think that's the
01:15:30
question. Are is this margin expansion
01:15:33
durable? The forecast the consensus
01:15:35
forecast and estimate is that margins
01:15:38
are going to continue to expand over the
01:15:40
course of the next 2 years. You and I
01:15:42
both know back in 22 23 we went from the
01:15:45
age of excess to the age of fitness.
01:15:47
Right? A lot of these companies were
01:15:48
able to shed people, you know, with the
01:15:50
excuse of AI just because they had
01:15:53
become uh, you know, uh, uh, too
01:15:55
excessive during the period of co. So I
01:15:58
I think it's a legitimate question
01:16:00
whether or not that's all from AI. But I
01:16:01
will tell you anecdotally, it maps for
01:16:04
me. I'm hearing like Jason and David, a
01:16:06
lot of these companies that are, you
01:16:08
know, uh, really growing their top lines
01:16:10
at an accelerating rate without
01:16:12
expanding headcount nearly at the same
01:16:14
pace. Okay, Sax, I'll give you the final
01:16:16
word while we're wrap.
01:16:17
>> Brad was talking about how we got all
01:16:19
these operating efficiency improvements.
01:16:21
The unemployment rate stayed at historic
01:16:24
lows during that time. I mean, the
01:16:26
economists consider full employment to
01:16:27
be 4 to 5%. And we've stayed at, you
01:16:30
know, the low low 4% 4.2% roughly during
01:16:34
this time. So, you're able to get these
01:16:36
efficiency improvements while
01:16:38
unemployment is still extremely low.
01:16:40
Moreover, there was just a big article
01:16:43
saying that the unemployment rate for
01:16:45
young college graduates has dropped. So,
01:16:48
you know, there was this whole narrative
01:16:50
recently that recent college graduates
01:16:52
were going to have the hardest time
01:16:53
finding jobs because, you know, there's
01:16:55
going to be no no work left for entry-
01:16:56
level jobs because of AI. And in fact,
01:16:59
it has gotten easier for recent college
01:17:00
graduates to find work recently. Maybe
01:17:03
that's because they're AI natives. Maybe
01:17:04
that's because they know how to use AI
01:17:07
better. So in any event, I mean, we're
01:17:09
just not seeing any evidence yet of
01:17:11
these theoretical downsides of AI around
01:17:14
job loss and unemployment, and we are
01:17:16
starting to see big productivity gains.
01:17:17
>> Yeah, I this is going to be a circular
01:17:19
discussion, but yeah, there's a lot of
01:17:21
conflicting evidence. The last piece of
01:17:23
conflicting evidence obviously is the
01:17:25
labor participation rate because if you
01:17:27
are not even opting in to participate,
01:17:30
then you, you know, don't get counted as
01:17:33
unemployed. And that's been, I think, a
01:17:35
big challenge. 61.9% in March. Labor
01:17:38
participation rate back in uh before we
01:17:41
were at 63.3. Yeah. And college
01:17:43
graduates are hearing different stories.
01:17:45
Certain degrees getting jobs, other ones
01:17:47
not getting jobs.
01:17:49
>> It's too early to tell. Uh I think is
01:17:51
probably what we all agree in. And it's
01:17:53
a it's a mixed bag.
01:17:54
>> Agree with that.
01:17:55
>> No. Look, whenever whenever I have data
01:17:58
to refute one of your narratives, you
01:18:00
always say it's too soon to tell.
01:18:01
>> No, no, no. What I do
01:18:02
>> online Wall Street Journal Nick put it
01:18:04
on the screen. College graduates are
01:18:05
finally catching a break in this job
01:18:07
market.
01:18:08
>> Yeah,
01:18:08
>> Jake, you should be happy about this.
01:18:10
>> I know. I'm Listen, I'm happy anybody
01:18:12
gets a job. But what you do is then you
01:18:13
say we don't trust the numbers and we
01:18:15
should get rid of the Fed and we should
01:18:16
get rid of the numbers. So, we all know
01:18:18
>> we should get rid of the Fed. That was
01:18:19
But you you get rid of the Fed because
01:18:21
we don't like the numbers. Listen, it's
01:18:22
all great. Welcome to the debate club.
01:18:25
>> What did I say?
01:18:25
>> You said abolish the Fed. Abolish the
01:18:28
Fed.
01:18:30
What is the Fed here for? All right,
01:18:31
listen. Enough. We're getting into Trump
01:18:33
derangement syndrome or Trump bend the
01:18:35
knee syndrome. It's the end of the show.
01:18:36
We had a great show, everybody. We had
01:18:38
some laughs. We all learned. We
01:18:40
workshopped some stuff. Let's leave it
01:18:41
where it is. Great job, President Trump.
01:18:44
>> I want to I want to congratulate
01:18:46
>> Oh, here we go.
01:18:47
>> I want to congratulate all of our
01:18:48
innovators. And I want to congratulate
01:18:50
Elon and um Daario D. Rockefeller on
01:18:53
their recent deal.
01:18:55
>> Oh, shots fired. He's getting straight.
01:18:57
Come on anytime, Daario. Hey, you know
01:19:00
Dario well, Brad. Well, get him on the
01:19:02
program next week. I want him on the
01:19:03
program. Have him come.
01:19:04
>> Will you ask him for me?
01:19:06
>> I um
01:19:07
>> Will you ask him for me?
01:19:09
>> Well, sure. I'll ask him. And yeah, you
01:19:10
know, the the fact of the matter is
01:19:13
>> I think our lucky stars that we have
01:19:16
Elon, that we have Anthropic, that we
01:19:18
have OpenAI, that we have Google, that
01:19:20
we have Amazon, all innovating in this
01:19:23
country. And you know, I know we like
01:19:25
to, you know, kind of poke poke fun on
01:19:27
the edges at these things, but the fact
01:19:29
of the matter is, you know, I see them
01:19:32
all showing up, sharing their models,
01:19:35
driving as hard as they can to innovate.
01:19:37
We have the best competitive framework
01:19:38
in the country. David's right. It's been
01:19:41
transformed over the course of the last
01:19:43
14 months. We need to stay the course.
01:19:46
We're on the winning horse. We just had
01:19:47
the derby last week. We're on the
01:19:50
winning horse. Stay on the horse.
01:19:51
America for the win. America for the
01:19:54
win.
01:19:54
>> There it is. Senator Brad Gersonner. I
01:19:56
think if you're gonna run, you got to
01:19:57
get rid of the red glasses. We gota get
01:19:59
maybe to shells in there. But I think
01:20:00
you got a you got a serious shot.
01:20:02
Senator Senator
01:20:04
>> I like I like secretary better.
01:20:06
>> Secretary.
01:20:07
>> Secretary. Who's Yeah. Secretary of the
01:20:10
Treasury. Brad Gersonner.
01:20:12
>> Secretary.
01:20:13
>> Secretary of State David Saxs. Secretary
01:20:16
of Cashmere and Wine. Chamath Paul. what
01:20:19
are you doing with your uh
01:20:20
>> there's so much fake news out there
01:20:21
because I mean look I totally agree with
01:20:22
everything Brad said look I poke fun at
01:20:24
at some of these companies for some of
01:20:26
the things they do but I am happy that
01:20:27
they are American companies and that
01:20:29
they're innovating here congratulations
01:20:32
on your dad
01:20:33
>> yes and look there's so much fake news
01:20:36
out there I mean we just covered on this
01:20:38
podcast how beneficial some of these
01:20:41
economic trends are you never get it
01:20:42
from the media
01:20:43
>> no
01:20:44
>> and they are trying to derail us from
01:20:46
from you know the the policies that have
01:20:48
been so successful Yes, but but they did
01:20:50
some great inspiring coverage of micro
01:20:52
looting. So get your microl looting on.
01:20:55
Congratulations, New York Times. We'll
01:20:57
see you next time, everybody. Bye-bye.
01:20:59
Love you, boys. Love you.
01:21:03
>> We'll let your winners ride.
01:21:06
>> Rainman David
01:21:10
and we open sourced it to the fans and
01:21:12
they've just gone crazy with it.
01:21:14
>> Love you, queen of
01:21:17
all. Winners,
01:21:23
>> besties are gone.
01:21:26
>> That is my dog taking your driveway.
01:21:31
>> Oh man, my appetiter will meet you.
01:21:33
>> We should all just get a room and just
01:21:35
have one big huge orgy cuz they're all
01:21:37
just useless. It's like this like sexual
01:21:38
tension that we just need to release
01:21:40
somehow.
01:21:43
>> Wet your feet.
01:21:46
>> That's going to be good. waiting to get
01:21:47
merch. I'm going
01:21:57
all in.