00:00:00
s have tequila I tasted 10 samples the
00:00:04
other day one was heads and shoulders
00:00:06
Above the Rest which was a good sign oh
00:00:09
so now we're taking that and making it
00:00:11
the new Baseline and I'm going to try 10
00:00:13
more samples and we're going to go from
00:00:15
there I think we're very close to having
00:00:16
something excellent great and then I did
00:00:18
a a side by-side taste test of the
00:00:21
sample I like the best compared to CLA
00:00:23
zul Ultra and it was already better than
00:00:26
CLA Ultra which retails for like 3,000 a
00:00:30
bottle really yeah holy so it's going to
00:00:32
be better v1's better than their final
00:00:35
product is what you're saying yes and
00:00:37
that was not just your taste test but
00:00:39
like you and several people it was me
00:00:40
and two other guys who I trust so we we
00:00:42
did it together now when you say trust
00:00:44
do you mean underlings or do you mean
00:00:46
actually like no he means alcoholics
00:00:48
yeah where in your pay hierarchy are
00:00:51
they he recruited them at Alcoholics
00:00:53
Anonymous in in West Hollywood are they
00:00:55
level two employees or level four
00:00:56
employees no one one works for me and
00:00:58
one's a consultant
00:01:00
they both both on the
00:01:03
pay let your winners
00:01:06
ride Rainman
00:01:10
David and instead we open sources to the
00:01:13
fans and they've just gone crazy with
00:01:15
[Music]
00:01:18
it okay are we over or underestimating
00:01:22
the impact of the AI boom Nvidia has
00:01:24
been on an alltime heater if you don't
00:01:25
know its market cap is now 1.8 trillion
00:01:28
freedberg this week it surpassed Google
00:01:32
and Amazon as the fourth most valuable
00:01:34
company in the world there it is folks
00:01:38
right behind Saudi aramco 16 months ago
00:01:42
stock was trading about $112 a share
00:01:46
since then share price has gone up 6.5x
00:01:48
on a big number shares went higher this
00:01:50
week after it was reported that Nvidia
00:01:51
plans to build custom AI chips with
00:01:54
Amazon meta Google and open AI in other
00:01:58
news finan Cal times reported openai hit
00:02:01
a $2 billion run rate for their
00:02:04
reoccurring Revenue it's about 165
00:02:06
million a month and they think they can
00:02:08
double that number in
00:02:10
2025 tons of competition coming out if
00:02:12
you didn't see it Google rebranded their
00:02:14
Genera of AI Suite to Gemini and they're
00:02:16
charging 20 bucks a month for it and
00:02:20
then there was this bizarre Wall Street
00:02:22
Journal headline that Sam was going to
00:02:23
raise $7 trillion that makes no sense
00:02:25
maybe that was the Tam we'll get into
00:02:28
that freeer thoughts on this Ascension
00:02:32
are we underestimating or overhyping AI
00:02:35
I mean it seems like the compute
00:02:37
buildout is
00:02:40
underway so this is going to continue
00:02:43
for some time I think the real question
00:02:45
that keeps getting asked is you know
00:02:47
when do you really see the tides recede
00:02:48
and what's what's there the compute
00:02:50
buildout is underway but are we seeing
00:02:52
application and productivity gains
00:02:54
associated with the capability of that
00:02:57
compute some would say yes some would
00:02:59
say no but you've got to build out the
00:03:02
infrastructure if you're going to assume
00:03:03
that there's going to be these
00:03:04
productivity gains in the application
00:03:06
layer and everyone's assuming they'll be
00:03:10
there so we need to meet the demand
00:03:11
which is almost like building out an
00:03:12
entirely new internet and so the core of
00:03:17
AI is this matrix multiplication problem
00:03:21
and you need these chips that do that
00:03:23
really efficiently they're not
00:03:24
traditional CPUs which are general
00:03:26
purpose chips but they're chips that are
00:03:27
very special purpose for a particular
00:03:29
set of
00:03:30
applications that render the output of
00:03:33
whatever model is being run it's sort of
00:03:35
like the internet like you could pretty
00:03:36
much create a bogey of any number on how
00:03:39
big this can get at the same time you're
00:03:42
seeing some of the other players with
00:03:45
different architectures and potentially
00:03:47
that are going to be beneficiaries of
00:03:48
this buildout because they're
00:03:49
participants in other parts of the data
00:03:51
center or alternative Solutions like arm
00:03:54
and I don't know if you saw this but in
00:03:55
just the last couple of weeks or last
00:03:57
week or two arm's valuation
00:04:00
doubled it's come down a little bit
00:04:02
since that big doubling today arm is
00:04:04
sitting at 140 billion
00:04:07
valuation last week it got as high as
00:04:09
156 and remember it was only last August
00:04:14
so if you guys remember arm was bought
00:04:15
by SoftBank and the vision fund and then
00:04:19
I think 25% of arm was held by the
00:04:22
vision fund and 75% was owned by
00:04:24
SoftBank last August SoftBank bought
00:04:28
from the vision fund the 25% that the
00:04:31
vision fund owned at a valuation of 64
00:04:33
billion so now soft Bank owns 90% of arm
00:04:37
the last 10% trades publicly so at $140
00:04:41
billion valuation softbank's position in
00:04:43
arm is 125 billion and I think it really
00:04:46
begs the question did Masa have a
00:04:48
masterful move here yet again because
00:04:51
everyone said for years with all the
00:04:53
bets this guy was making in the vision
00:04:55
fund and some of the misguided bets he
00:04:57
was making with SoftBank and the he was
00:05:00
applying on himself and SoftBank to
00:05:02
deploy all this Capital during the Zer
00:05:04
era everyone was saying you know this
00:05:06
guy had a one hit it was Alibaba and
00:05:08
that was it and clearly he doesn't know
00:05:10
what he's doing but now all of a sudden
00:05:13
his bet on AI was fairly preing and it's
00:05:16
paying off pretty powerfully he bought
00:05:19
the whole business for 32 billion and
00:05:21
now he has a a stake that's worth 125
00:05:24
billion seven years later yeah I mean
00:05:27
chip in a chair right I mean it's really
00:05:29
begging the you only need one the power
00:05:31
law right when you're making a longterm
00:05:32
investment you can very easily look like
00:05:35
a over the short term and to your
00:05:37
point the power law all it takes is one
00:05:40
of them to work out and it can wipe out
00:05:41
the losses on the rest of the portfolio
00:05:44
and we may be seeing mosac with his
00:05:45
whole AI Vision story that he's been
00:05:47
telling for seven years we may see mosa
00:05:49
clawing his way out of the hole you only
00:05:51
look like a to
00:05:53
losers meaning when people were throwing
00:05:57
shade at Masa it wasn't anybody that's
00:06:00
any good I didn't see headlines that
00:06:02
said John Malone thinks Masa is a fraud
00:06:05
Warren Buffett thinks Masa is a fraud
00:06:07
nobody was saying that because those
00:06:09
guys know what the game is and the game
00:06:11
is exactly what you said jcal you must
00:06:14
always
00:06:15
survive and you need to keep your wits
00:06:18
about you and over time if you're
00:06:20
prepared and do enough good work you'll
00:06:22
get waves of good luck that amplify your
00:06:24
hard work and Masa is an example of an
00:06:27
incredibly talented person that has
00:06:29
found a way to survive more than
00:06:31
anything else and so waves of success
00:06:33
will hit
00:06:34
him the people that complain are the
00:06:38
ones that are sort of a little bit stuck
00:06:40
on the sidelines and are more frustrated
00:06:42
with their own impotence and inability
00:06:44
to do anything now on arms specifically
00:06:48
the thing that I would keep in mind is
00:06:50
that those guys floated I think about 15
00:06:53
million shares so there's about 145 150
00:06:58
million shares outstanding 90% owned by
00:07:00
them so there's about 10 or 15 million
00:07:02
shares outstanding and there was about
00:07:04
125 million shares short going into the
00:07:06
earnings report and so the thing to
00:07:09
remember is that there was almost an
00:07:11
entire turn of the float that was short
00:07:14
so I don't know how much of this
00:07:17
reaction was just an enormous short
00:07:20
squeeze of people that were largely long
00:07:23
other chip companies and so they thought
00:07:25
they could short arm essentially as a
00:07:28
spread trait to minimize volatility but
00:07:30
this is an example of where if you don't
00:07:32
do this analysis properly GameStop was
00:07:35
another where you misjudge or you sub
00:07:40
analytically look at the actual share
00:07:41
count and what's actually free floating
00:07:44
you can get caught on the wrong side and
00:07:45
these things can be very violent and to
00:07:47
explain the float to folks you can make
00:07:50
a certain number of shares available to
00:07:52
the public to trade when you have a
00:07:54
small number of those correct shth
00:07:57
available to the public weird things can
00:07:59
happen
00:08:00
because there's not enough available to
00:08:02
short that's exactly right and so that
00:08:04
could have been the situation here with
00:08:05
this
00:08:06
runup but I think I think the overall
00:08:08
point of placing bets in being a capital
00:08:11
allocator is the important one to
00:08:13
discuss here saaks what are your
00:08:14
thoughts on all the crazy bets that Masa
00:08:18
made we watched a lot of our companies
00:08:20
or companies adjacent to our companies
00:08:22
get massively overfunded some of them
00:08:24
literally got drowned in capital but
00:08:26
then you only need to hit one and and
00:08:28
maybe win it all back what are your
00:08:29
thoughts on this high alpha you know
00:08:34
betting style of his look I mean he's a
00:08:36
gambler and he is playing a power loss
00:08:40
so he only needs one really big return
00:08:42
to basically make up for all the
00:08:43
mistakes and it seems like he's hit that
00:08:46
once again he did it with Alibaba too
00:08:48
right I mean he's a major Gambler most
00:08:50
people would be a little bit more risk
00:08:52
averse I don't think it was a great
00:08:54
strategy to be pumping $500 million
00:08:56
checks into what were effectively seed
00:08:58
companies like they were doing with the
00:08:59
vision fund however obviously he made
00:09:03
enough good bets that they're going to
00:09:04
pay off that fund well it's a good point
00:09:07
sacks I mean none of the big I don't I I
00:09:10
I probably need to be careful because
00:09:12
I'm not super familiar with the
00:09:14
portfolio but it doesn't seem like any
00:09:16
of the big winners were those seed bets
00:09:19
like he doesn't have any returns coming
00:09:20
out of that kind of investment strategy
00:09:23
door Dash door Dash yeah door Dash was
00:09:26
not like what point did they invest
00:09:28
billion or2 billion do was yeah pretty
00:09:30
important moment we've heard at dinner
00:09:32
Jason and I at dinner we've heard the
00:09:35
the waterfall from the soft Bank side
00:09:37
and it sounds like it was an enormous
00:09:39
number yeah that was definitely their
00:09:42
but it was a drisk business it was a
00:09:44
scaling business seems a little
00:09:46
different than some of these companies
00:09:48
that were fairly early on that got
00:09:51
monster valuations and large chunks of
00:09:53
capital and then struggled to find ways
00:09:55
to deploy that
00:09:56
Capital which led to unhealthy Behavior
00:09:59
I think the thing you have to give him a
00:10:00
lot of credit for is it's so easy to
00:10:03
give up when the peanut gallery is
00:10:06
chirping and so the thing that you have
00:10:10
to give him a lot of credit for is he
00:10:13
has the ability to tune those people out
00:10:14
now he probably isn't even living in a
00:10:16
world where any of those people that
00:10:19
chirping even gets to him and I think
00:10:20
that that's a very important
00:10:22
thing but kudos to him I mean yeah to be
00:10:26
able to tag something for 100 billion
00:10:28
years 70 billion there I mean keep
00:10:31
swinging and by the way at the end of it
00:10:33
all what is it going to change for him
00:10:35
absolutely nothing except he'll have
00:10:37
some incredible stories to tell about
00:10:39
the VA that he took and the roller
00:10:40
coaster R he was on and I think that's
00:10:42
going to be the most valuable thing he's
00:10:44
going to be able to tell his grandkids
00:10:46
these stories and his their eyes will
00:10:48
just be like what and his friends you
00:10:51
know I mean I would love to have a
00:10:52
dinner with him I don't know him let do
00:10:54
it the Pod come on the Pod Masa let's
00:10:56
talk about I've only met him once by I
00:10:58
have a great story to tell about Masa
00:10:59
when I was on Facebook I met him once
00:11:01
but anyways goad tell it I mean can you
00:11:03
tell it all I'm saying is I would love
00:11:04
to have dinner with him because I just
00:11:06
want to hear these stories of all like
00:11:08
up and down and up and down and am I had
00:11:11
two lunches with him we had lunch one
00:11:13
day we had such a good time he said can
00:11:15
we have lunch again tomorrow how long
00:11:16
are you in Tokyo I said yeah and I came
00:11:18
back the next day and I had lunch with a
00:11:20
voice that's amazing and it was great
00:11:21
like talking two or three hour lunches
00:11:23
where we're just talking about every Web
00:11:25
2.0 company this is in you know the 2009
00:11:28
Uber kind of Robin Hood period and he
00:11:31
was just fascinated by what was going on
00:11:33
and and all the disruption happening
00:11:35
door Dash series D $14 billion do Post
00:11:38
2018 he did well the question I was
00:11:40
asking is like door Das was a lot like
00:11:43
an arm there's a point of view that the
00:11:45
business is proven that there's a market
00:11:47
that it can grow into that there's a
00:11:48
management team that knows how to deploy
00:11:50
Capital versus a lot of these early
00:11:53
stage bets where when they got too much
00:11:56
money they ambled about trying well they
00:11:59
didn't amble about
00:12:00
they kind of incinerated it incinerated
00:12:04
because what do I do with this money I
00:12:05
don't yet have a machine or an engine
00:12:07
that can grow with this money and until
00:12:09
the engine can handle the capital Pizza
00:12:11
thing like the pizza thing Zoom Pizza
00:12:13
you're talking about yeah I mean if you
00:12:15
well there's dozens of them in that
00:12:16
portfolio right yes if you don't have a
00:12:18
business that can handle the gas you're
00:12:21
putting in the tank the engine's going
00:12:23
to blow up the engine has to grow it has
00:12:26
to build the volume has to build so that
00:12:28
it can actually handle the fuel and I
00:12:30
think that's been part of the challenge
00:12:31
and I will just say I've said this in
00:12:33
the past but there was an ERA in Silicon
00:12:35
Valley that started around
00:12:37
20 10 2009 kind of right after right as
00:12:41
right as Web 2.0 was starting to venture
00:12:43
into more traditional markets meaning
00:12:46
tech companies were becoming other
00:12:48
companies kind of the Mobile
00:12:49
Local era was was kicking off and there
00:12:52
was a strategic concept that got popular
00:12:55
in Silicon Valley on take the market
00:12:57
which is when you feel like you could be
00:12:59
the winner give the person all the chips
00:13:02
and say now go over bet and take the
00:13:05
whole Market because these are going to
00:13:06
be winner take all or these could be win
00:13:07
or take most markets so get first mover
00:13:10
Advantage here's more Capital than you
00:13:11
need and I remember being pitched this
00:13:13
when I was raising my series B that I
00:13:16
should raise more than I was trying to
00:13:17
raise because I should use the extra
00:13:19
Capital to take the market and then that
00:13:21
Mantra just became the deao standard
00:13:24
which is oh you have something that has
00:13:25
something it has some Edge it has some
00:13:27
Advantage it started to move in the
00:13:29
market here's a ton of capital you don't
00:13:31
really need now go take the market and
00:13:34
the reality is most of those businesses
00:13:35
had absolutely no way of using that
00:13:38
Capital to quote take any Market they
00:13:40
didn't even know what their Market was
00:13:41
yet yeah they when you in the product
00:13:43
Market fa product Market fit phase and
00:13:46
you don't have the engine as you called
00:13:47
it how do you make the engine go faster
00:13:49
you're literally trying to take a bunch
00:13:51
of car parts and then run a Formula 1
00:13:54
race the car's not even trackworthy yet
00:13:57
and Cera you not not to pick on any
00:14:00
companies but I remember catara which
00:14:02
was doing construction and they just
00:14:03
thought they're going to take over the
00:14:04
whole construction injur you know
00:14:06
industry with this company it raised
00:14:08
like $900 million and I believe it's
00:14:10
like a zero the whole way you cannot you
00:14:13
can't put the money to use Capital as a
00:14:15
weapon maybe works if you're uber lift
00:14:17
and door Dash and you have customers and
00:14:19
you can reduce the price but well you
00:14:20
know what's interesting is when you now
00:14:22
look at all bunch of these AI companies
00:14:24
every day what you see is every round is
00:14:26
a minimum of a hundred million
00:14:30
but the the Dirty Little Secret is not
00:14:33
that these companies have found product
00:14:34
Market fit or are scaling Revenue
00:14:36
massively I'm not talking about open AI
00:14:37
I'm talking about these other longtail
00:14:41
businesses the100 million is largely
00:14:43
going to compute but as we know and I
00:14:46
think we'll talk about this next week
00:14:48
when all these announcements are done
00:14:50
but you're about to see a one10 thing of
00:14:53
the compute cost and so what's
00:14:55
interesting for those companies is
00:14:56
they're going to be left with a lot more
00:14:58
money but back to free Brook's point
00:15:00
they may be ambling about with too much
00:15:02
money at that point so today you raise
00:15:04
$100 million because you need to spend
00:15:06
70 or 80 on compute tomorrow when you
00:15:08
only need to spend seven or eight for
00:15:10
the same amount of
00:15:11
throughput on one hand you could say
00:15:13
well this is great I have so much more
00:15:15
money I can spend but again we've seen
00:15:18
example after example where if you raise
00:15:20
too much money when you don't need it
00:15:21
you will just run yourself into a brick
00:15:23
wall you'll over hire you'll Mish hire
00:15:26
you'll misallocate and it just goes back
00:15:28
to the same thing that we've said before
00:15:30
which is there's just not enough people
00:15:31
that actually know how to run companies
00:15:33
yeah and it just becomes a huge
00:15:34
distraction you see sacks today Sora did
00:15:37
you see Sora get dropped by open this is
00:15:39
breaking news right now yeah this is
00:15:41
unbelievable if you haven't seen it Sora
00:15:44
is the text video model from open AI
00:15:46
they've been talking about this I guess
00:15:48
this uh would compete
00:15:51
with staple diffusion do video as well
00:15:54
anyway you give a prompt I'll just read
00:15:56
this one out loud beautiful snowy Tokyo
00:15:58
City is bustling the camera moves
00:15:59
through the bustling City street
00:16:01
following several people enjoying the
00:16:02
beautiful snowy weather and shopping at
00:16:04
nearby St stalls gorgeous sakur pedals
00:16:08
are flying through the wind along with
00:16:10
snowflakes and
00:16:12
this if you weren't studying it would
00:16:16
look like a major feature of film or TV
00:16:20
show yeah SX yeah I mean it's
00:16:23
unbelievable I mean it looks to me like
00:16:25
4K video I mean I'm not sure it actually
00:16:27
is 4K but it looks pretty darn close
00:16:30
yeah the rendering resolution can be
00:16:33
whatever you want it to be right the
00:16:34
fact that you can describe the the way
00:16:36
that you want the camera to move so I'm
00:16:39
going to speculate on this you guys know
00:16:40
I talked about this quite a while ago on
00:16:43
how I think we're going to be at prompt
00:16:45
to video meaning people are going to
00:16:47
have personalized video games and
00:16:48
personalized movie experiences and music
00:16:51
eventually where rather than everyone
00:16:53
watch the same product made centrally by
00:16:56
a producer we'll all make our own
00:16:58
products locally I spent some time
00:17:00
digging into this over the last year and
00:17:02
the Big Challenge was that the
00:17:03
traditional approach for rendering video
00:17:05
is you create three-dimensional objects
00:17:07
and then you have a rendering engine
00:17:09
that renders those objects and then you
00:17:10
have a system that defines where the
00:17:12
camera goes and that's how you get the
00:17:14
visual that you use to generate a 2d
00:17:16
movie like
00:17:18
this and the best tool out there is
00:17:20
Unreal Engine 5 which is made by epic
00:17:22
games I don't know if you guys have ever
00:17:23
seen the demos of it but it's is that
00:17:26
Star Wars uses with their like virtual
00:17:28
sets now when they did the Mandalorian I
00:17:30
think they're using the Unreal Engine or
00:17:31
unit they probably are and then there's
00:17:33
other tools if you're in movie
00:17:35
production you'll use different stuff
00:17:36
but Unreal Engine is just so
00:17:37
incredible but it requires the placement
00:17:40
and the rendering of 3D objects so where
00:17:42
do those objects go and then the physics
00:17:44
of the light moving off those 3D objects
00:17:46
to generate a 2d video is pretty
00:17:49
straightforward that's been done for 20
00:17:50
30
00:17:52
years and so the the density the the
00:17:55
rate of improvement the color contrast
00:17:57
Etc has all improved made it look
00:17:58
incredible now it's super photorealistic
00:18:00
you create photo realistic video but it
00:18:02
still requires this concept of where are
00:18:04
the 3D objects in 3D space this doesn't
00:18:07
do that this was a train model so we
00:18:11
don't know what's going on in the model
00:18:12
so how would you train a model to do
00:18:15
this without having a 3D space because
00:18:18
clearly there's thousands of objects in
00:18:20
that rendering if you look at that video
00:18:21
thousands of little objects the compute
00:18:23
necessary to Define each of those
00:18:25
objects place them in 3D space is
00:18:27
practically
00:18:28
impossible today so this is being done
00:18:32
through some model that starts to render
00:18:34
all of these things together so it is
00:18:37
very likely my guess is that open AI
00:18:39
used a tool like Unreal Engine 5 and
00:18:43
generated tons and tons of uh video
00:18:46
content tagged it labeled it and were
00:18:50
then able to use that to train this this
00:18:53
model that can for whatever reason that
00:18:56
we don't understand do this make the
00:18:57
output but what's interesting is that
00:19:00
this
00:19:01
output you're referring to synthetic
00:19:03
training
00:19:04
data exactly exactly and and and but
00:19:07
what's amazing about this model is that
00:19:10
underneath this this model on its own
00:19:12
has rendered physics so all of the rules
00:19:15
of physics all of the rules of motion
00:19:17
all of the rules of perspective all of
00:19:19
the rules of how objects sit upon one
00:19:21
another next to each other how fluids
00:19:22
move if you look at some of the other
00:19:24
demos on this tool you can actually see
00:19:26
fluid moving it's
00:19:29
the M of fluid or rering the M ofir
00:19:31
those are like physics models that would
00:19:33
typically deterministically written
00:19:35
meaning that a programmer an engineer
00:19:37
writes the physics and then the the
00:19:39
Unreal Engine renders the physics in
00:19:41
this case the model figured out the
00:19:43
physics and the model probably did it in
00:19:45
such a reduced way that you can use a
00:19:48
reduced amount of compute to generate
00:19:50
something that would otherwise take a
00:19:51
ton of compute time I mean look at this
00:19:53
it's truly amazing Nick find the one of
00:19:55
Golden Retrievers playing in the snow
00:19:57
this is what way better than this but
00:19:59
what's amazing is that the and by the
00:20:00
way if you're listening to this I really
00:20:01
recommend going on Twitter and typing in
00:20:03
open AI Sora s o r a and looking at some
00:20:06
of these demo videos it's Bonkers and I
00:20:09
think it really speaks to what's so
00:20:10
incredible right now which is that
00:20:12
software is writing itself and it's
00:20:14
replacing a lot of human determinism the
00:20:17
way that humans think about constructing
00:20:20
software is being completely redone with
00:20:22
these uh uh learn models this model
00:20:24
learned physics on its own and it
00:20:26
renders this thing that looks like only
00:20:28
way we know how to do this today is to
00:20:30
create 3D object models in 3D space and
00:20:32
use tons of compute to render the to
00:20:34
translate that from Geek to you know
00:20:37
people
00:20:37
listening this is incredible because
00:20:40
Pixar spent years maybe a decade trying
00:20:43
to figure out hair I remember talking to
00:20:45
Ed catull and he told me hair was just
00:20:46
so hard to get right with wind and
00:20:49
lighting they literally built the
00:20:51
physics engine for years like maybe even
00:20:53
a decade to make hair cross The Uncanny
00:20:56
Valley what you're saying here free ber
00:20:59
is somehow this language model this
00:21:01
image model just outputs it and it
00:21:04
doesn't actually have a physics model it
00:21:06
just outputs it correct yeah it it well
00:21:08
we don't know but the
00:21:11
model there may be some determinism
00:21:13
built into it so so I don't want to
00:21:14
assume anything but the general approach
00:21:18
determinism means you write the physics
00:21:20
and you write the formula of the physics
00:21:22
yep so that the software can execute the
00:21:24
physics and render the thing but in this
00:21:27
case it may very well be the case like a
00:21:30
lot of the other models that are being
00:21:32
trained today when you create a model
00:21:34
they're trained that it's being trained
00:21:36
and on its own it resolves to figure out
00:21:39
how to do the physics is there a limit
00:21:42
on the length of these Sora videos that
00:21:44
you can generate right it's one minute
00:21:46
right now 60 seconds today obviously
00:21:48
that'll come off at some point but by
00:21:50
the way if you can do 60 seconds you can
00:21:51
do two hours right the problem that you
00:21:53
have is I mean if we're talking about
00:21:55
creating a movie is making edits so I
00:21:59
obviously I haven't used sory yet
00:22:00
because it just came out today but I've
00:22:01
used Dolly which is just their prompt uh
00:22:04
image service and I was designing the
00:22:07
label for or the bottle for the all-in
00:22:10
tequila for you guys yeah and what I
00:22:12
noticed is that it would spit out a
00:22:14
bunch of ideas and then I would take one
00:22:18
and then want to just make little
00:22:20
changes and I would just say okay keep
00:22:22
this but just change this one detail it
00:22:24
would give me a completely new image so
00:22:28
the AI doesn't realize the layers
00:22:31
inherent in what it's producing somehow
00:22:34
it takes this prompt and generates a
00:22:36
response and if you give it that same
00:22:39
prompt a 100 times it'll give you 100
00:22:41
different answers and if you say just
00:22:43
modify this this or this it can't do it
00:22:45
can't reason it can't take criticism
00:22:47
well it seems like on some level it's
00:22:49
unaware of the output that it's created
00:22:51
again yeah a human creating this in
00:22:54
Photoshop would have a lot of layers and
00:22:56
every object in that image would have
00:22:59
its own layer and you could have a layer
00:23:01
for text and just quickly modify the
00:23:03
text when I want to do something as
00:23:05
simple as edit the text or change the
00:23:08
font again it would give me a completely
00:23:09
different image and I just wonder about
00:23:12
these videos if you want to create a
00:23:14
movie I think you're probably going to
00:23:16
be stuck with like the what they give
00:23:19
you and then maybe you can import that
00:23:21
at some point into an Unreal Engine and
00:23:23
then do work from there the editing part
00:23:25
is it doesn't exist yet when you do text
00:23:28
it's easy because you just ex you cut
00:23:29
and paste it into your grammarly or your
00:23:31
notion or whatever you use and you start
00:23:33
editing here there's no exporting like
00:23:36
you're saying it's to Final Cut Pro or
00:23:38
something I bet that when they turn this
00:23:40
into an API that when you're working in
00:23:42
Unreal Engine you know if you're like a
00:23:45
VR designer or something like that or
00:23:46
you're like a um a filmmaker you could
00:23:49
probably just get started with a prompt
00:23:51
and that could
00:23:53
be hugely time-saving if they can make
00:23:56
that integration work
00:23:58
I think that's pretty interesting I
00:24:00
looked at this a lot over the last year
00:24:01
this was one of the biggest challenges
00:24:03
is in order to have an interactive
00:24:05
approach to video production or video
00:24:09
game production you have to be able to
00:24:11
say we take that out put that in but
00:24:13
like like you said the model doesn't
00:24:15
distinguish the objects it doesn't
00:24:17
distinguish each of those little flowers
00:24:19
or each of those little cherry blossoms
00:24:21
or it doesn't even distinguish that
00:24:22
there's two people walking down the road
00:24:24
the model doesn't have this concept of
00:24:26
objects when it renders
00:24:28
and so um the traditional model where
00:24:31
you have objects in a 3D model that then
00:24:33
get rendered into a video allows you to
00:24:36
do those sorts of things so what may end
00:24:39
up happening is a fusion and this is
00:24:41
something we talked about with Steph
00:24:42
Wolfram and others that there are these
00:24:44
models that are called compute models
00:24:46
that do deterministic compute tasks then
00:24:51
there are these AI train models where we
00:24:54
don't really have a sense of how they
00:24:55
work but the fusion of the two will
00:24:57
unlock a lot of opportunities and that's
00:24:59
very likely the next generation of
00:25:01
models is not just purely um non-human
00:25:04
intervene train models but the
00:25:06
integration of those models with these
00:25:08
deterministic models with these compute
00:25:10
models and then you can start to do
00:25:11
things like instead of saying hey render
00:25:14
this whole scene you're saying render
00:25:16
all the objects that would go in the
00:25:17
scene they get placed in an Unreal
00:25:19
Engine and then the Unreal Engine
00:25:20
outputs the video and so that
00:25:22
integration work I think is the next big
00:25:25
phase in AI model development and
00:25:27
that'll unlock like insane yeah cuz
00:25:30
right now you can't use this to create
00:25:33
dialogue like actors in a movie scene
00:25:35
talking to each other and acting
00:25:37
especially like close-ups of those
00:25:40
actors with their facial expressions I
00:25:41
don't at least I don't see any examples
00:25:43
like that the way humans want to have a
00:25:45
creative process it doesn't work that's
00:25:47
right yeah yeah it'll get there though
00:25:48
and if you put something like this out
00:25:51
there to the public and said it's 20
00:25:52
bucks a month to do this or 50 bucks a
00:25:55
month you can have a lot of consumers
00:25:57
who want to buy this product so when I
00:25:59
see this I think people are going to pay
00:26:00
for this and then going back to Nvidia
00:26:04
or arm okay more infrastructure to your
00:26:07
point freeberg is like what
00:26:08
infrastructure do we need for everybody
00:26:10
to have Pixar on their laptop or on
00:26:13
their you know phone and people people
00:26:16
are willing to pay for these I don't
00:26:17
know if you guys saw this but there was
00:26:18
another big breakthrough announced today
00:26:20
with the um the new Gemini pro15 can you
00:26:23
just pull up sundar's announcement on it
00:26:24
yeah Sundar now is found the password to
00:26:27
his Twitter handle he's he's tweeting
00:26:29
yeah the reason I want to talk about
00:26:31
this if you just scroll down so when you
00:26:32
prompt an AI
00:26:34
model you put in a certain number of
00:26:36
words into like an llm model today and
00:26:39
it spits words back out the more words
00:26:42
you put into the model the more compute
00:26:45
intensive or challenging it is to do the
00:26:46
compute work needed to render the output
00:26:49
so open AI their gp4 turbo model has
00:26:55
128,000 tokens which means about 28,000
00:26:58
words can be fed into the model to get
00:27:00
an answer back out so you could put like
00:27:02
a whole essay or 100 pages of text and
00:27:05
it will read a document for you you get
00:27:07
a whole book and it can give you a
00:27:09
summary in a couple of seconds what
00:27:11
Google just announced with Gemini 1.5
00:27:12
Pro is a 1 million token context window
00:27:16
up to 10 million token context window in
00:27:19
total I don't know what's going on here
00:27:22
because the problem is the way these
00:27:23
models are kind of traditionally built
00:27:27
is that that creates a nonlinear scaling
00:27:29
on compute needs to be able to increase
00:27:31
that that context window like that so
00:27:33
there is some architectural work that's
00:27:35
gone on to allow these models to scale
00:27:38
up in a way that we did not think
00:27:39
possible when these models first came
00:27:41
out which is crazy to think was only
00:27:42
like 24 months ago that they started to
00:27:44
get kind of looked at like this 18 mon
00:27:46
May yeah and so there's two kind of
00:27:49
things that are happening one is that
00:27:50
these models are getting chopped up in a
00:27:52
way that you can run small models in a
00:27:54
highly trained way locally and you don't
00:27:56
need a massive general purpose model and
00:27:58
so that's going to make compute needs
00:28:00
very small for very specific tasks and
00:28:02
the other one is that there's other
00:28:03
architecture that's happening in
00:28:04
deploying the models that's allowing
00:28:06
massive context Windows like this so it
00:28:08
totally changes how scalable they are
00:28:10
with the same number of chips or the
00:28:12
same amount of computer the same amount
00:28:13
of power so there are all these amazing
00:28:16
mors law effects happening right now in
00:28:19
AI technology development on the chip
00:28:21
level on the architectural level and the
00:28:22
how they're being built and how they're
00:28:23
being deployed that completely changes
00:28:26
the concept that I think most folks had
00:28:27
when these first came out two years ago
00:28:29
and the potential so I think sky is
00:28:31
still the limit we don't really know how
00:28:32
big this is going to get but all of
00:28:34
these efficiency gains mean that we can
00:28:36
do so much more with so much
00:28:38
less and by the way when you can do more
00:28:40
with less it doesn't mean you need less
00:28:41
chips it means you need a lot more chips
00:28:44
because the fact that you can do a lot
00:28:46
more with much fewer chips means
00:28:48
everyone's going to want to do it now
00:28:49
more people can afford to do it so it
00:28:51
unlocks the opportunity for uh for
00:28:53
compute and so it really starts to
00:28:55
rationalize the Nvidia thesis that it
00:28:58
could be worth1 trillion do potentially
00:29:01
and for people in the audience to
00:29:02
understand what this context window
00:29:04
means I'm sorry if I was too Technic no
00:29:06
no no no I'm just I always like to just
00:29:07
give an example or two after you explain
00:29:09
it the context window for example you
00:29:13
could upload the transcript of this
00:29:14
podcast right and there were only
00:29:16
certain services like Claude maybe where
00:29:18
you could put a PDF attachment boom you
00:29:20
put it in there and say hey summarize
00:29:21
this using your language model and using
00:29:23
this content now you could be able to
00:29:25
say hey take me every Mr Beast video
00:29:28
take me you know take the top 10,000
00:29:31
videos on YouTube take the transcripts
00:29:33
summarize them for me give some analysis
00:29:35
on it and that was only available to
00:29:37
developers prior to these large context
00:29:40
Windows being open and that's going to
00:29:43
just open up a massive amount of
00:29:46
application right shim the results are
00:29:48
actually not that much better there's
00:29:49
these third party service benchmarks
00:29:51
lombata is probably the most often
00:29:52
referred to but what you see if you just
00:29:54
look at for example all the gpts even as
00:29:57
they increase the number of parameters
00:29:59
in a model the quality of that model
00:30:02
doesn't necessarily increase with the
00:30:03
rate of parameter increase so you have
00:30:06
this thing of a little bit of
00:30:08
diminishing return which is more a
00:30:09
problem of what we said before which is
00:30:12
eventually when you're scouring the open
00:30:13
internet doesn't matter how fast or how
00:30:16
how big you scour the open internet the
00:30:17
open internet is a bounded space and so
00:30:19
you get the same answer so all of these
00:30:23
things are they're interesting marketing
00:30:26
wear but I would just encourage us all
00:30:29
to keep in mind that the shift is not
00:30:33
necessarily these kinds of
00:30:35
things it's kind of like saying you know
00:30:37
in a chip like oh wow we built this chip
00:30:39
on 3 nanometer okay what does it mean no
00:30:42
I think the piece respectfully you might
00:30:44
be missing here is like your proprietary
00:30:47
data might not be in that language model
00:30:50
so by having this context window and you
00:30:52
can add something that you have on your
00:30:54
hard drive that's not how it works so my
00:30:55
point is the context window is not
00:30:58
correlated to the quality of the model
00:30:59
outputs correct but consumers and
00:31:03
business users haven't been able to put
00:31:05
their data into the models because that
00:31:06
has nothing to do with the context
00:31:08
window that existed when the context
00:31:09
window was smaller that that's what you
00:31:11
think you do things like rag with right
00:31:13
but most people don't have access to
00:31:14
that this is like consumers can now just
00:31:16
take something off their hard drive rag
00:31:19
is different than what this is okay like
00:31:22
this is this what what you guys just
00:31:24
spent all this time talking about is
00:31:26
equivalent to saying I built this chip
00:31:28
with on a 7 nanometer process or a 3
00:31:31
nanometer process and what I'm saying is
00:31:34
it is technically more burdensome as
00:31:36
freedberg said it's technically much
00:31:38
more sophisticated and
00:31:40
complicated but the the translation to
00:31:44
Quality in the consumer is not what that
00:31:48
means so a 3 nanometer chip is not two
00:31:51
times better than a 7 nanometer chip for
00:31:53
the consumer
00:31:55
similarly a context window that's five
00:31:57
times bigger than the one before it
00:31:59
doesn't make a model five times better
00:32:01
to the consumer these are orthogonal
00:32:03
ideas it's not to say that it isn't an
00:32:05
important thing but I I I would say that
00:32:07
it's a reminder for me that we're in the
00:32:09
early
00:32:10
phases where there's a lot of hay that
00:32:14
people can
00:32:15
make but it's important to look at these
00:32:18
third-party benchmarks that try to sort
00:32:20
through and sift through what matters
00:32:22
and what doesn't matter and one of the
00:32:24
interesting things that thirdparty
00:32:25
Benchmark service have figured out is
00:32:29
that irrespective of the exponential
00:32:31
growth of these things like parameter
00:32:33
size quality is only growing linearly
00:32:36
and now
00:32:38
sublinearly so it's just an interesting
00:32:40
point I understand your point now my
00:32:42
point is man when we when you are a
00:32:45
consumer of this and I've been talking
00:32:47
to people trying to get their data into
00:32:50
you know the context window to be able
00:32:51
to ask a question they're always like
00:32:52
how do I put more in there how do I put
00:32:54
more in there and then these language
00:32:56
models
00:32:57
because of the copyright issues they're
00:33:00
not able to upload Blade Runner and then
00:33:03
say hey make me something but I was able
00:33:05
to and just to give a very real example
00:33:07
I did um a podcast around the ray Croc
00:33:11
movie the founder that we were
00:33:12
referencing earlier I uploaded the
00:33:14
screenplay I found a PDF of the book and
00:33:17
I found two other like essays about Ray
00:33:19
Croc like HBS studies and I said give me
00:33:21
like a timeline of it and I use
00:33:23
something called Google notebook which
00:33:24
has a big context window and you can
00:33:25
upload all this stuff and that's where
00:33:27
I'm seeing
00:33:28
consumers and what these things are
00:33:30
capable of Visa the context window it it
00:33:34
just unleashes in people's mind so many
00:33:36
applications that you could do there are
00:33:38
two examples today that caught my eye
00:33:41
that I think are worth bringing up the
00:33:42
first one is from Meta Meta published
00:33:44
this paper today which was like pretty
00:33:47
crazy and essentially what they said is
00:33:49
that for the realm of the software
00:33:51
worker the software
00:33:53
engineer what they basically built is
00:33:56
the this llm tool that runs over llama
00:33:59
called testgen and it essentially does
00:34:01
automated unit testing to such a high
00:34:04
degree of accuracy so you know in the
00:34:06
software development if you want to spit
00:34:07
out a product you have to go through
00:34:09
this thing where you have to
00:34:10
functionally run unit tests and make
00:34:12
sure the thing works as advertised and
00:34:15
this paper essentially now allows you to
00:34:17
Traverse a code base understand it
00:34:19
generate these unit tests with high
00:34:21
accuracy so where is this valuable think
00:34:24
of remember the software bugs in like
00:34:26
the bow triple 737 Max issue jcal where
00:34:30
it turned out that it was just bad
00:34:31
software tools like this now will able
00:34:34
will be able to run unit tests where you
00:34:37
will not be able to get out get that
00:34:38
plane out the door unless all of these
00:34:41
functional unit tests not only have been
00:34:43
written but run and have been passed
00:34:45
successfully and it'll all be done by
00:34:48
Bots which means that you can't gain
00:34:50
that system so that's that's something
00:34:52
today that it's the last thing
00:34:54
developers get to is writing the unit
00:34:55
test right it's always like is why
00:34:57
there's all kinds of bugs and security
00:34:58
holes in code now I thought that that
00:35:01
was a pretty large breakthrough and I
00:35:03
think we're going to go through a bunch
00:35:04
of fine tunes on this that make it
00:35:06
interesting and then I saw this other
00:35:08
thing which caught my eye even more
00:35:11
which was this thing that Nat fredman
00:35:13
tweeted out which is this thing called
00:35:14
Magic dodev and what essentially it is
00:35:17
is like a coworker and what they are
00:35:19
claiming and I don't know if it's true
00:35:22
is that now you can
00:35:24
basically have this thing under
00:35:27
understand your style of code and
00:35:29
actually be able to write code that is
00:35:31
just as good as the way you write code
00:35:33
so now it's not a co-pilot but
00:35:35
effectively a
00:35:36
coworker and this is the first real
00:35:38
example of what I've seen of a claim
00:35:40
that basically says hey you know we can
00:35:43
now be an add-on to your
00:35:46
Workforce and this to me is just so
00:35:50
gargantuan because like if you look at
00:35:53
the amount of Opex inside of any company
00:35:56
that sits inside of R&D if you look at
00:35:58
the amount of stock based
00:36:00
compensation right if you replace humans
00:36:02
with Bots which is what this is you're
00:36:05
going to crush those things SBC at the
00:36:07
limit goes to zero because you don't
00:36:09
have to give stock to a bot right you
00:36:11
have to give stock to
00:36:13
people and Opex gets shrunk because now
00:36:16
you just pay a license to magic dodev or
00:36:19
I don't know how they're going to make
00:36:19
money on this but really really really
00:36:22
fascinating stuff it is a scary there
00:36:25
was another open source project I think
00:36:28
it's called Maggie mg I don't know if
00:36:30
you saw that freberg but Apple has been
00:36:33
working on an open source image project
00:36:36
and so you're kind of going to see I
00:36:38
think in iOS 18 or the next iPhone 16
00:36:42
they're going to have an image model on
00:36:43
onto that phone where is where is Apple
00:36:46
in all of this it's unbelievable no they
00:36:48
they're where are they they are building
00:36:51
an image model and you're going to be
00:36:53
able to talk to your phone and they
00:36:54
announced anything I think you're going
00:36:56
to see a
00:36:57
on the next set of mobile
00:37:00
devices that will run models locally yep
00:37:03
and so this goes back to the point about
00:37:05
how the hell is the government going to
00:37:06
regulate AI models which is the
00:37:08
stupidest thing I've ever heard um well
00:37:10
not the stupidest it's probably you know
00:37:12
probably top 20 but the fact that
00:37:15
they're going to like be able to be
00:37:17
fully vertically designed that they can
00:37:19
have the hardware and the software and
00:37:23
integrate all the way through to the
00:37:24
application layer they're going to have
00:37:25
I would imagine incredible intuitive AI
00:37:28
features built into all of their
00:37:30
products that can run locally rather
00:37:32
than all the other stuff that's going on
00:37:34
where you've got to make a request on
00:37:35
the internet and get the results back
00:37:36
over the Internet it'll be pretty
00:37:38
powerful but obviously no data right we
00:37:41
don't know what they're doing what just
00:37:42
speculating here well no I mean you have
00:37:44
breadcrumbs and the breaks are they have
00:37:46
an image model and I just dropped in the
00:37:48
hugging face of it you can basically
00:37:49
upload an image and just give it a text
00:37:51
prompt of how to change the image so you
00:37:54
know if Nick were to use this he could
00:37:56
upload a a picture of Youth and say you
00:37:58
know give him five o' Shadow give him a
00:38:00
mustache whatever and Nick Nick will you
00:38:02
do that real quick while we're talking
00:38:04
while you're doing that can you bring up
00:38:06
the video of the robots life in a cyber
00:38:09
Punk setting it's beautiful is this a
00:38:11
SORA rendering yeah this is Sora so
00:38:14
beautiful so it's really cool so but
00:38:16
here's the the limitation and this goes
00:38:18
back to what me and fre were talking
00:38:19
about earlier okay so you've got these
00:38:22
people in the background let's say Fe
00:38:25
the feeder backwards by the way you can
00:38:26
see human feet aren't correct sort of
00:38:28
like a vol problem anyway what's cool is
00:38:30
that okay so those cars aren't moving
00:38:32
they're static you got some movement of
00:38:34
these background people right at least
00:38:38
so they're kind of they're the extras in
00:38:41
a movie but if you wanted to tell Sora
00:38:45
this is just my guess that okay I want
00:38:47
the people in the background to be doing
00:38:48
this it would render a whole different
00:38:50
image like your robot might change too
00:38:53
right well SX this is where your context
00:38:55
window matters so if you repeat keep the
00:38:57
statement If instead of writing and then
00:39:00
iterating if you just reprompt it and
00:39:03
your prompt got longer and longer and
00:39:04
longer with the degrees of specificity
00:39:06
you needed that's the way you use these
00:39:08
models today and to your point what
00:39:10
humans are looking for from a ux
00:39:12
perspective is an iterative approach to
00:39:15
rendering is an iterative approach which
00:39:17
isn't feasible in a world where the this
00:39:21
this is all a matrix it's a
00:39:23
three-dimensional Matrix there's two
00:39:24
dimensions of pixels and then the Third
00:39:26
mention is the next image and the next
00:39:27
image and the next image and so that
00:39:29
Matrix is rendered from scratch it
00:39:32
doesn't have a concept of what are the
00:39:35
objects that I'm rendering it has a
00:39:36
concept of a matrix of pixels and so it
00:39:39
will render a matrix of pixels based on
00:39:40
a redefined prompt but what you're
00:39:43
describing would require a very
00:39:44
different model architecture than the
00:39:45
way all of these train models operate
00:39:47
today this is going to
00:39:49
revolutionize pornography I think it's
00:39:51
the first no no no I'm not saying it as
00:39:53
a joke I think that's where you're going
00:39:54
to see this first because then all these
00:39:56
issues potential issues of underage
00:39:58
people or exploited folks it goes
00:40:01
completely away your gonna cut
00:40:04
this no and and you know it'll kill only
00:40:07
fans yeah say gambling is where every
00:40:10
Revolution starts that's where exactly
00:40:12
technological that's where payments
00:40:14
that's where the web started that's
00:40:15
where payments got refined but I I think
00:40:17
that in
00:40:18
pornography you're going to see this
00:40:21
basically destroy pornography and only
00:40:23
fans well I I don't want to bring up
00:40:24
something uh that's super offensive but
00:40:27
there was a certain pop music star who
00:40:29
had a bunch of images just a couple of
00:40:31
weeks ago that were AI generated yeah
00:40:35
fakes deep fakes right yeah there were a
00:40:37
bunch of deep fakes of a pop star I
00:40:38
don't to like actually highlight the
00:40:40
person's name because then it just I
00:40:41
think it sense more of it but X did a
00:40:44
really good job of blocking all of them
00:40:46
watermarking them and saying you can't
00:40:48
do this and so this is like a whole new
00:40:50
world you can take any person on the
00:40:52
planet and make porn out of them that is
00:40:54
realistic well I don't I don't mean that
00:40:55
I just mean that you'll be able to make
00:40:58
pornography in a way that is totally uh
00:41:01
nonhuman involved which is generally
00:41:04
better I would say for All Humans
00:41:06
involved what's great about the Deep
00:41:08
fakes is that if you have a sex tape
00:41:10
come out you can just say it's a deep
00:41:11
fake totally I was about to ask Cham
00:41:13
what is the prompt you're GNA use to
00:41:14
make love to yourself
00:41:16
tonight with the new Sor model chamat
00:41:20
Embraces chath all I hope is I I may
00:41:23
just need to publish some data so that
00:41:24
whatever deep fake gets some part it's
00:41:26
accurate this is Apple Oh short Apple
00:41:28
stock this is terrible it's not very
00:41:31
good that's not very good what is this
00:41:32
model this is an apple model it's called
00:41:34
Maggie mg e this is Apple's open source
00:41:38
image model it's way behind just looks
00:41:40
like GPT 0.1 they got work to do but
00:41:43
it's open source it's going to go faster
00:41:44
it's very hard to make fine tune changes
00:41:46
in these models or to get the result
00:41:48
that you want but that's fine this is
00:41:49
just a this is v0.9 so if you guys want
00:41:52
to see a great fine tune the Juggernaut
00:41:55
XL fine tune uh oh I would just say try
00:41:57
it it is a fine tune that does not
00:42:02
allow bad-looking people the whole the
00:42:05
whole concept is only beautiful people
00:42:07
so it doesn't matter what your prompt is
00:42:09
but these people come out the most
00:42:12
incredibly good-looking people I've ever
00:42:14
seen as if the standard wasn't already
00:42:16
high enough with filters and everything
00:42:18
else but this is going to be
00:42:19
revolutionary for independent film I
00:42:21
mean this is obviously just tip of the
00:42:22
iceberg but you know the problem with
00:42:23
independent film right now is that the
00:42:25
economic model literally doesn't work
00:42:27
it's too expensive to make an
00:42:29
independent movie relative to the return
00:42:33
and so the independent movie business is
00:42:35
almost dried up almost completely this
00:42:38
is going to revolutionize it because
00:42:39
it'll bring the cost down close to zero
00:42:42
now I I don't I don't agree with
00:42:43
freeberg that everyone's going to design
00:42:44
their own entertainment because most
00:42:47
people are lazy no I'm not saying design
00:42:49
the way you think about creative
00:42:51
production processes is what's going to
00:42:53
change it's all going to reduce down to
00:42:55
sitting down and it's like when you ask
00:42:56
your spouse or partner or whatever what
00:42:58
do you want to watch tonight and they
00:43:00
answer something that something will
00:43:02
then be delivered to you instantly
00:43:04
that's what's going to be so powerful
00:43:05
well you mean the AI is going to create
00:43:07
it on the spot for you it certainly will
00:43:09
but I I think that there's a certain
00:43:11
element of culture that gets integrated
00:43:13
into
00:43:15
personal I mean I don't think we have
00:43:17
time for this but I have like a whole
00:43:18
philosophy people still want curation I
00:43:20
mean if you look at something like
00:43:22
YouTube um or really any user generated
00:43:24
content site there are rough 100 times
00:43:27
more people who are passive than people
00:43:29
who are active and actually create stuff
00:43:31
whatever the ratio is there's always a
00:43:33
lot more passive viewers than there are
00:43:36
creators so there's I'd say most of the
00:43:39
content will be created by people who
00:43:40
actually care enough to do it again
00:43:42
let's call it one or two% of the
00:43:44
population and then of those 1% might
00:43:47
actually be really good at it yeah but
00:43:49
here here's where freeberg I think has a
00:43:51
really good point because I have been
00:43:52
looking into this imagine if you took
00:43:56
the soprano something you and I love ZX
00:43:58
and you just said hey for each season
00:44:00
give me 10 more minutes per episode and
00:44:02
give me like two more characters or
00:44:04
whatever now you when you watch it the
00:44:06
next time you could say just fill it in
00:44:08
I want every episode to be 90 minutes
00:44:09
instead of 50 minutes or I've been doing
00:44:12
this thing uh with my daughters we're
00:44:14
into Star Wars and the Clone Wars Etc
00:44:17
people have been
00:44:18
cutting shows like Andor or Rogue one
00:44:23
Clone Wars the prequels Etc and the bad
00:44:26
batch and they're putting in together
00:44:27
like timelines and they're making their
00:44:29
new versions of it by super cutting the
00:44:33
different series that are filling things
00:44:34
in so you're going to be able to say hey
00:44:36
tell me the whole Star Wars saga with a
00:44:39
little more perspective from Obi-Wan or
00:44:41
give me a little more perspective from
00:44:43
this person and that that's going to
00:44:45
just make these things like these
00:44:46
franchises so much more watchable and
00:44:48
deep and Rich yeah so this is my point
00:44:50
sex is I don't think it's about groundup
00:44:54
personal media creation I think that
00:44:56
it's about the curation model changes
00:45:00
the Creator or the central creative
00:45:02
Force comes up with a genre with a bunch
00:45:05
of characters like a Star Wars and
00:45:07
general like arcs of stories and then
00:45:09
you can go in and you can view it from
00:45:11
any character's point of view imagine
00:45:13
watching the Star Wars movie from Darth
00:45:15
Vader's point of view imagine watching
00:45:17
the Star Wars movie where you just
00:45:20
stayed on Endor the whole time and you
00:45:23
watch the news on the TV while you're
00:45:25
hanging at fantasy you with the you with
00:45:27
the
00:45:28
ewo so you're right that there's a great
00:45:31
opportunity for some of these cinematic
00:45:32
universes to kind of open up the
00:45:34
creative assets and let the fan base do
00:45:36
whatever they want with them or create
00:45:38
you know these Alternatives I still
00:45:40
think that it's that 1% of the
00:45:43
viewership that actually cares enough to
00:45:46
do that and the reason is because that I
00:45:49
think what you see with all these uh
00:45:51
demos is that it's really easy to get
00:45:53
the first 90% it's really really hard to
00:45:56
get the next 9% and then the final 1%
00:45:59
you know to get to the level of of
00:46:01
course yeah to get it dialed in to the
00:46:03
point where again the the first thing it
00:46:05
gives you but the the benefits could
00:46:07
outweigh those costs and this is true in
00:46:09
3D video game playing I don't know if
00:46:10
you play fortnite but there's all sorts
00:46:12
of stuff that you feel could be better
00:46:14
but the overall value of that experience
00:46:17
far outweighs the missing elements of it
00:46:19
not being truly photo realistic and I
00:46:22
think that's where this stuff goes it's
00:46:23
ski week next week
00:46:26
and I've done this for all both of my
00:46:29
older boys and I'm doing it now for my
00:46:32
youngest boy who is four and a half
00:46:34
almost five we are going to watch this
00:46:37
entire week all of the Star Wars movies
00:46:40
fantastic and then I got a downloaded
00:46:42
history movies all the movies yeah from
00:46:45
and I got an order the order was not
00:46:47
what I thought it's not as released
00:46:49
obviously no you want to do timelines so
00:46:51
you want to have the prequels I did that
00:46:54
once and I will tell you if
00:46:57
Ro one uh the Star Wars the original
00:46:59
sequel well you're saying this order is
00:47:01
not good here's what I was told this is
00:47:03
what was the internet was the most
00:47:04
popular you start with The Phantom
00:47:07
Menace yeah then attack of the Clones
00:47:10
then revenge of theth yeah then solo is
00:47:13
optional option yeah Rogue one optional
00:47:17
y Rogue one not optional then four five
00:47:19
and six obviously a new hope that's when
00:47:21
your kids will lose their mind they'll
00:47:22
be like what is this it's like Empire
00:47:24
Strikes Back return to the Jedi six are
00:47:26
too slow for kids yeah like the better
00:47:28
seven the force awakens and eight The
00:47:30
Last Jedi then nine the rise of
00:47:32
Skywalker skip the last three yeah the
00:47:34
last three are garbage they fall asleep
00:47:36
during 456 yeah God 456 are the best
00:47:39
prequels are probably the most watchable
00:47:41
they're very they in in retrospect
00:47:43
they're great but there's also the Clone
00:47:44
Wars which is absolutely fantastic I
00:47:47
would sprinkle in some Clone Wars
00:47:48
Episodes Natt and my daughters I'm
00:47:50
forcing them to sit with us for this
00:47:52
entire Matt's going to watch all this no
00:47:54
she's ready to shoot her your in the
00:47:56
face yeah this is her worst nightmare
00:47:57
she and my daughters are like I put the
00:47:59
over under on Nat sticking through maybe
00:48:00
a 100 minutes of this and she's done
00:48:03
she's done Amore why do you poison the
00:48:06
kids with this nerd stuff they're never
00:48:08
going to get the
00:48:10
girlfriend what this is not a felini
00:48:13
this is a how about they watch something
00:48:15
where the people make a love instead of
00:48:17
the lights they able to cut people hand
00:48:18
off you watching this exactly it's said
00:48:21
no there's no love and a joy uh there's
00:48:23
uh people with light it's no fun but to
00:48:28
bridge the gap between your two Vision
00:48:29
Sachs and
00:48:31
freeberg just ma voice is so
00:48:35
great chamath and I will talk when we
00:48:38
talk to each other for literally 20
00:48:40
minutes as Nat we'll just have a convers
00:48:43
but it's a conversation we're having I
00:48:45
don't know why Chata this deal has uh
00:48:48
such a high interest rate on the
00:48:50
convertible note why would you do that
00:48:52
to me I'm a your friend who's going to
00:48:53
be on the board that this a person looks
00:48:55
like a dum dum what's going on I don't
00:48:58
understand well it's an independency jth
00:49:01
you don't mean to have a put the super
00:49:03
smart a different perspec is the
00:49:06
opposite of a white truffle the white
00:49:07
truffle is delicious this a product that
00:49:10
smells like a
00:49:11
poo but anyway
00:49:15
between we're literally just
00:49:18
talk coming tonight ask are you making
00:49:20
it for dinner or not we need I have to
00:49:21
I'm not I'm going to try to make it by
00:49:23
dinner I just have a I have to give a
00:49:24
keynote on a podcast conference oh my
00:49:26
God are you wait sorry are you kidding
00:49:28
me no I'm not Kidd you guys banned me
00:49:31
from having advertising here I have to
00:49:33
make cheddar I got bills to pay just
00:49:35
because there's an eight in front of the
00:49:37
Uber stock price doesn't mean I can
00:49:38
retire will you be there by 7 p.m. or
00:49:42
not I'm going to make it by 7 yes I keep
00:49:44
me in for we are going to start dinner
00:49:48
when you get there perfect I will not do
00:49:50
David saaks time okay thank you but
00:49:53
between freeberg and Dave you know SA
00:49:55
actually said it's like one or two% of
00:49:56
the audience for you know show with a
00:50:00
million viewers or 10 million viewers
00:50:01
that could be tens of thousands of
00:50:03
lunatics and I don't know if you knew
00:50:05
this but you remember when Luke
00:50:06
Skywalker showed up spoiler alert in the
00:50:08
Mandalorian everybody was like oh his
00:50:10
young face does not skip The Uncanny
00:50:12
Valley this is why you got to watch the
00:50:14
Mandalorian so then some kid did his own
00:50:17
AI rendering of Luke Skywalker you know
00:50:21
after Return of the Jedi but you know
00:50:23
after he defeats Emperor and they then
00:50:27
hired that kid to work at Disney and
00:50:29
Lucas Arts so what's happening is all
00:50:32
these people who are making content for
00:50:33
the Star Wars Universe they are going to
00:50:36
be the next generation of creators so
00:50:37
between a Mr Beast way over here and
00:50:41
Lucas art there's going to be a new fill
00:50:43
in and it's going to be all these great
00:50:45
1centers making interesting with IP
00:50:48
that's owned by other organizations I
00:50:50
think it's a whole new paradigm for
00:50:52
media and content it's going to be so
00:50:54
different the automated stuff that's
00:50:55
customized for you it's almost like what
00:50:57
YouTube and Instagram did for user
00:51:00
generated content there's going to be
00:51:02
this whole new model of AI generated
00:51:05
content the the real winners are going
00:51:07
to be the technology platforms that
00:51:09
bring these tools in a simple intuitive
00:51:11
way that to Sax's point doesn't
00:51:13
challenge the user to do work but makes
00:51:15
their life feel easier and better and
00:51:17
brings them a more Rich experience and
00:51:19
they're getting today from any other
00:51:20
media source and that's going to be a
00:51:21
game changer all right Lots going on in
00:51:23
AI I thought I'd bring up really
00:51:26
interesting story that happened just
00:51:29
around stock options and employees and
00:51:33
advice you're going to get generally
00:51:34
speaking just tons of chaos at the
00:51:36
startup bolt if you don't remember
00:51:38
that's a one-click checkout startup it
00:51:40
was founded by a guy named Ryan breslo
00:51:42
he's been on my other podcast this week
00:51:43
in startups he went viral in 2022 after
00:51:47
we call YC and stripe like mob bosses
00:51:49
and had this whole conspiracy theory of
00:51:51
how they were trying to kill this
00:51:52
company don't know if it's true or not
00:51:55
but at its peak bolt was valued at $1
00:51:57
billion chth and they're buying back
00:52:00
shares at a $300 million valuation down
00:52:02
97% yada yada what makes this
00:52:05
interesting was that in 2022 breslo
00:52:08
announced a new stock auction program
00:52:10
for his employees on Twitter and he did
00:52:12
this in a very bombastic lecturing kind
00:52:14
of way to the entire industry saying
00:52:16
that bolt would offer loans to any
00:52:18
employee to exercise their options early
00:52:20
we all know about these type of loans
00:52:22
from their balance sheet that is the
00:52:24
idea here yes and he called it the most
00:52:27
employee friendly plan possible and said
00:52:30
that half of both 600 plus employees
00:52:33
took the
00:52:35
deal he since deleted the entire thread
00:52:37
but at the time a lot of VCS who had
00:52:40
been through this before you know
00:52:41
explain to Ryan this a terrible idea
00:52:43
here's ggv's Jeff Richards for what it's
00:52:46
worth almost every private company s
00:52:47
finally did this in the 90s was an
00:52:49
absolute disaster employees spent years
00:52:50
paying back loans for worthless stock
00:52:52
tax bills for merely exercising Etc yada
00:52:55
y and his now deleted tweet was over
00:52:59
half our employees chose this plus I
00:53:00
would strongly encourage my family and
00:53:02
friends to choose this but VC say it
00:53:04
didn't work in the 90s so it's a
00:53:05
disaster VC Twitter pumps the tweet this
00:53:08
is why VC run companies are never able
00:53:10
to make strides for employees obviously
00:53:11
all those options at 11 billion are
00:53:14
worthless now but sorry they did this
00:53:16
when they were worth 300 billion 11
00:53:18
billion and it was reported that at the
00:53:20
same time in a new industry publication
00:53:23
that reslo
00:53:26
was selling his shares in secondary
00:53:28
preso confirmed with that newsletter
00:53:31
that he had sold 10 million and uh
00:53:35
sources uh Anonymous sources y yada take
00:53:38
that forward it's where it said he
00:53:39
didn't tell the team he was doing
00:53:41
that two investors also accused the
00:53:44
company of misleading shareholders and
00:53:45
raising capital on inflated metrics SEC
00:53:48
investigated but did not recommend any
00:53:50
action sex jth I don't know if either
00:53:53
want to take this this is really
00:53:54
mechanical but it does speak to what
00:53:56
happens when things implode should
00:53:58
employees be taking loans to buy these
00:54:00
kind of stocks yeah at a minimum
00:54:01
hopefully these loans are non
00:54:06
recourse and at the maximum if they were
00:54:09
recourse but it looked like these guys
00:54:10
were doing these kinds of shady things
00:54:12
they have to kind of forgive the loans
00:54:13
and eat it there's a reason why almost
00:54:15
no one does these types of exercise
00:54:16
loans anymore I mean whoever the VC was
00:54:19
who pushed back was right I mean these
00:54:20
things were common 20 something years
00:54:22
ago people stopped doing it and the
00:54:24
reason why why is that well there are a
00:54:27
couple reasons so the
00:54:29
mechanics of the loan are that the
00:54:31
company loans you the money to pay down
00:54:33
the exercise price the strike price for
00:54:36
the options and then you're supposed to
00:54:38
pay the company back at some future date
00:54:40
basically when you can sell the shares
00:54:42
and in effect from the point of view of
00:54:45
the balance sheet of the company this
00:54:46
was cashless because it would loan the
00:54:49
exercise price to the employee who would
00:54:51
then exercise and pay that money back to
00:54:52
the company so everyone's like well
00:54:54
what's wrong with this this just seems
00:54:56
like simple mechanics where no one loses
00:54:59
the employee gets to start the clock on
00:55:02
their shares for long-term capital gains
00:55:04
that's why they would do it the problem
00:55:06
is that what people discovered is that
00:55:10
if the company didn't work out and those
00:55:11
shares became worthless then the
00:55:14
employee is still on the hook for the
00:55:16
loan and if they don't pay the loan back
00:55:20
then there's a loan forgiveness issue in
00:55:21
other words the forgiven amount of the
00:55:23
loan or the def faed amount of the loan
00:55:26
becomes ordinary income and so what
00:55:28
happens is you end up having to
00:55:29
recognize as income the loan that you
00:55:32
never pay back so you can lose
00:55:34
furthermore if there's spread at the
00:55:38
time that you exercise between the
00:55:39
strike price and the fair market value
00:55:41
of the shares then although if if it's
00:55:44
the right kind of option if it's an ISO
00:55:46
you can defer paying tax on that spread
00:55:48
however people got whammied by the AMT
00:55:51
that that spread still counted for
00:55:53
Alternative Minimum Tax so there were a
00:55:55
lot of people during the first Doom
00:55:57
bubble who exercised thinking they can't
00:56:00
lose because the market was always up
00:56:01
and to the right and then the company
00:56:03
would go bust and they would get hit
00:56:06
with a big AMT and then they would get
00:56:07
hit with a loan forgiveness issue so
00:56:09
that's why everyone stopped doing these
00:56:10
things yeah yeah I mean really
00:56:12
importantly getting taxed on that spread
00:56:15
you owe that money no matter what no
00:56:16
matter what no matter what the IRS
00:56:19
demands that money and the California
00:56:21
Franchise Tax Board demands that money
00:56:23
so I just want to be very specific so
00:56:25
people understand this if the value of
00:56:28
your company's gone up to $2 a share as
00:56:31
a private company and your options are
00:56:34
10 cents a share you pay 10 cents a
00:56:36
share to get those shares you owe tax on
00:56:39
the $190 difference and ultimately if
00:56:42
the company's now worth less than a
00:56:44
dollar when you're finally able to sell
00:56:46
or it shuts down you actually have money
00:56:49
you have to pay to the IRS the franchise
00:56:50
tax board and you did not make enough
00:56:52
money to cover the tax bill the way a
00:56:53
lot of these companies do this is they
00:56:54
do do a cashless early exercise or
00:56:56
cashless exercise loan meaning you don't
00:56:58
even have to put money up to to pay the
00:57:00
10 cents a share the company is just
00:57:02
covering it for you to sax this point
00:57:05
you end up getting taxed for that later
00:57:07
if you don't pay it back so the whole
00:57:09
thing creates a lot of tax obligations
00:57:12
and it's a very complicated structure
00:57:14
and system and the reason options exist
00:57:17
is so that people could buy their
00:57:18
options and not have to pay taxes at the
00:57:20
time that they receive them because if
00:57:22
you just get shares in a company rather
00:57:24
than options the value of those shares
00:57:27
is your tax bill even if you can't sell
00:57:29
the shares well it's income so it's
00:57:31
income and so options were created as a
00:57:33
way that you would have a strike and the
00:57:35
strike was the fair market value so
00:57:37
you're actually not getting any value
00:57:39
today and so you can avoid the tax bill
00:57:41
but now it leads to all these other tax
00:57:43
Loops way too complicated for R and file
00:57:45
employees is what it comes down to like
00:57:47
I I saw this with like senior Executives
00:57:49
trying to get like the top five
00:57:50
positions you know a CFO or
00:57:52
something is to reduce your tax B bill
00:57:55
yeah from 50% to 20% and at the end of
00:57:58
the day you open up all of these other
00:57:59
problems if you try and do that
00:58:01
completely one of the stupidest things
00:58:03
about the way this was used is that they
00:58:06
would give the exercise loan to
00:58:08
employees who were leaving the company
00:58:10
because they said that you know normally
00:58:12
when you're a a departing employee from
00:58:14
a company you have three months to
00:58:17
exercise your options or you or you lose
00:58:18
them so they were saying that this would
00:58:20
be a solution to that right we'll allow
00:58:22
you the money to exercise all they were
00:58:24
doing is is creating a huge tax problem
00:58:26
for that former employee and you're
00:58:27
right the average employee is not
00:58:29
sophisticated enough to navigate all of
00:58:32
these
00:58:32
issues it's not clear also why you'd
00:58:35
want to create this complicated benefit
00:58:37
for departing employees it seems like if
00:58:40
you're going to do this you would want
00:58:41
to do it for a key exec who's joining
00:58:44
the company that's the only context in
00:58:45
which I've seen it by the way yeah same
00:58:47
yeah so they were trying to solve a
00:58:49
problem using this complicated device
00:58:53
that that really backfired
00:58:55
but it's not even a problem that you
00:58:56
should try and solve in that way I mean
00:58:57
there's a much simpler solution which is
00:58:59
extend the period of time that employees
00:59:01
have to exercise if you don't want them
00:59:03
to lose their option if they don't
00:59:04
exercise within three months give them
00:59:06
10 years and there are a bunch of
00:59:07
companies that have done that where they
00:59:10
give you a lot longer to exercise the
00:59:11
option the trend I'm seeing with
00:59:13
startups like early stage startups seed
00:59:15
stage startups which are really
00:59:16
indicative of what startups will do
00:59:18
later on hiring people Cash basis paying
00:59:22
them hourly outside the United States
00:59:24
not having to worry about stock options
00:59:26
not having to worry about healthare just
00:59:30
finding a global Workforce and using
00:59:32
these thirdparty tools there's a lot of
00:59:33
third party tools to manage employees in
00:59:36
Portugal or Manila or Canada or whatever
00:59:38
South America and they're just saying
00:59:41
you know what stock options are going to
00:59:42
be for like an elite group the top 10%
00:59:44
of the company and then we're going to
00:59:46
automate the rest of the jobs with AI
00:59:48
and we're going to Outsource and have
00:59:50
hourly workers or full-time contractors
00:59:53
outside the United States it's the unden
00:59:55
that's terrible why that's throwing the
00:59:58
baby out with the bath water one of the
00:59:59
most magical things about stars in
01:00:01
Silicon Valley is that when it worked
01:00:03
everyone got rich I mean you have the
01:00:04
story of the Google Chef who made $100
01:00:07
million or the Secretary of Microsoft
01:00:09
who got to retire it you know and those
01:00:11
are extreme examp graffiti artist at
01:00:13
Facebook the graffiti artist he really
01:00:15
what was his name David David show the
01:00:18
fact of the matter is that Silicon
01:00:19
Valley had a share of the wealth
01:00:20
mentality where the rank and fall
01:00:22
employees did participate in the owner
01:00:24
ship at the company and they got equity
01:00:26
and it was highly motivating or it is
01:00:28
highly motivating sure for not just like
01:00:31
lower level employees but mid-level
01:00:33
employees and Senior employees to
01:00:34
basically get equity in the company
01:00:36
that's one of the things that makes
01:00:38
Silicon Valley company special or
01:00:40
companies created it outside Silicon
01:00:42
Valley but using this Valley template
01:00:45
yeah the know getting rid of options
01:00:47
alog together is a terrible over what
01:00:49
they're doing is more like for the core
01:00:51
team that's in Silicon Valley options
01:00:54
but then like for work that is not
01:00:56
considered as essential you know just
01:00:58
hiring people offshore who don't well
01:01:01
try to get those people then to work
01:01:02
nights and weekends the way that you
01:01:04
might be if you're the owner of the
01:01:06
company they're not g to have an
01:01:06
ownership mentality yeah I agree I mean
01:01:10
get nine to five workers that way and I
01:01:12
think what people are starting to
01:01:13
realize is like maybe there's with these
01:01:15
remote workers a lot of people are doing
01:01:17
nine-to-five work and maybe that like
01:01:20
hardworking culture just doesn't exist
01:01:22
in Solon Valley like it used to you may
01:01:24
be a to find hardworking entrepreneurial
01:01:27
like
01:01:28
folks that don't necessarily speak
01:01:31
English that now with these AI tools
01:01:34
basically become some of your best folks
01:01:36
but there are many countries that don't
01:01:38
value Equity don't have an understanding
01:01:40
of it and so to Jason's point it's
01:01:42
almost more of a complication to try to
01:01:44
give it so that may be the balancing
01:01:47
thing sacks that happens which is that
01:01:49
all these tools actually allow somebody
01:01:51
in some country that speaks some
01:01:54
language that's not English but who can
01:01:57
be completely committed who becomes this
01:01:59
Lynch pin of a business and gets paid an
01:02:02
extremely great salary in American terms
01:02:05
that makes them rich in their country
01:02:06
get to the same place maybe maybe I
01:02:08
don't know their infrastructure needs to
01:02:10
be built for stock options in other
01:02:11
countries because I don't know that that
01:02:13
exists or just pay people more cash
01:02:14
because a lot of folks have different
01:02:16
again stock options make sense when you
01:02:18
have a differential in taxation between
01:02:21
stock equity and inome yes but in other
01:02:25
places where you have a flat tax or no
01:02:29
income tax whatsoever what's the
01:02:31
difference yeah if you're in Dubai
01:02:32
paying 1% like you could just ship it
01:02:35
right yeah ship it I me you can always
01:02:39
go back to a model where people vest
01:02:41
shares rather than vest options and
01:02:44
those shares vest on a trigger of time
01:02:47
and a liquidity event and that way they
01:02:49
can't exercise or real or they won't get
01:02:51
tax prior to the liquidity event
01:02:53
happening but then they will will have
01:02:55
to pay ordinary income tax rates on
01:02:57
their shares hold on we're making it
01:02:58
sound like exercising options is a bad
01:03:00
thing it can be a very good thing I mean
01:03:02
if you're an early employee of a startup
01:03:06
and the options are worth pennies and
01:03:10
you can exercise for I don't know a few
01:03:12
hundred dollars then that's a great tax
01:03:14
benefit because you start the clock
01:03:16
taking on long-term capital gains
01:03:17
treatment and when you sell those shares
01:03:19
eventually if there's a liquidity event
01:03:21
you get turn tax at a capital gains rate
01:03:24
not an ordinary income rate that's the
01:03:26
tax benefit that's why people want to be
01:03:28
able to exercise their options in the
01:03:30
first place the problem is that as you
01:03:33
join the company later and later the
01:03:35
strike price of the options goes up and
01:03:37
so therefore if you want to exercise
01:03:40
you're going to have to write a pretty
01:03:42
big check and so you're going to take a
01:03:43
meaningful risk some people do that and
01:03:45
that's fine the trouble comes in when
01:03:48
you try to get cute and go around taking
01:03:50
that risk by saying okay well the
01:03:52
employee doesn't have to go out of
01:03:53
pocket for their exercise price we're
01:03:55
going to loan them the money well they
01:03:57
thought that they had figured out a way
01:03:58
to avoid taking that risk but as it
01:04:00
turns out all you end up doing is
01:04:03
potentially incurring AMT and a loan
01:04:05
forgiveness issue so there's no way
01:04:07
around the fact that if you want to
01:04:09
exercise your options to get the
01:04:11
potential tax benefit of a lower tax
01:04:14
rate you have to also take the risk of
01:04:17
losing the strike price that of that
01:04:20
check that you write into the company
01:04:22
that's a fair trade right if you want
01:04:24
the potential benefit you take the
01:04:25
potential risk but that doesn't that
01:04:27
doesn't mean that you should never
01:04:28
exercise there's a lot of cases again I
01:04:31
think if you're early at a company you
01:04:32
should low thousands of dollars low
01:04:34
hundreds of dollars go for it there's
01:04:37
all kinds of ways where when the basis
01:04:39
is extremely
01:04:40
low it's nominally not a lot of money so
01:04:44
you can take the risk but when you start
01:04:46
talking about tens of thousands or
01:04:47
hundreds of thousands of dollars or
01:04:49
millions of dollars and then using loans
01:04:51
and leverage and you got the risk of
01:04:53
Ruin coming out
01:04:54
News Bears all the way around no bu all
01:04:57
right SX here your read me did you see
01:04:59
the poll about Biden's age and his the
01:05:04
unfavorable percentage of people who
01:05:06
think he is too old how can I not see
01:05:08
that I think the most important thing
01:05:10
that we should talk about is that
01:05:13
President Biden has decided to not go
01:05:15
through a cognitive test as part of his
01:05:17
annual physical it's the first time it's
01:05:20
ever happened Joe Biden now 81 years old
01:05:22
82 at the start of his second if he wins
01:05:24
he'll be 86 at the end making him by far
01:05:27
the oldest president ever Trump is 77 he
01:05:30
would be 78 and a half when he gets his
01:05:33
second term if he won
01:05:34
82.5 at the end of his term here's a
01:05:37
histogram of age of presidents these are
01:05:40
the oldest just to give you a everybody
01:05:42
obviously refers to Reagan as the oldest
01:05:44
president prior to this he was a spry 69
01:05:47
at the start of his presidency and by
01:05:50
the end he was 77 which is younger
01:05:53
obviously than Biden and Trump
01:05:57
today so saak thoughts or shth thoughts
01:06:02
let's just say that we've had a bunch of
01:06:03
these issues over the last few
01:06:06
years president Trump didn't disclose
01:06:09
his tax
01:06:10
returns is that a huge issue I think
01:06:14
people made it to be a huge issue but
01:06:16
it's more of an issue of judgment but I
01:06:18
don't think it fundamentally compromises
01:06:20
his ability to leave the
01:06:23
country you had a different issue not
01:06:25
with the president but with the current
01:06:27
SE defense secretary who had prostate
01:06:31
cancer the disclosures were let's just
01:06:34
say somewhat
01:06:37
suboptimal he had a UTI he
01:06:39
got emergency admitted to the hospital
01:06:43
he just had another acute issue and
01:06:45
those disclosures weren't
01:06:48
good but you're talking about an acute
01:06:51
kind of a thing where if you're
01:06:52
incapacitated there's a chain of of
01:06:54
command that immediately kicks
01:06:56
in those are a class of issue that I
01:07:00
think people can find things that are
01:07:03
wrong with probably correlated to one's
01:07:07
political affiliation the other side
01:07:09
probably cares more than the side that
01:07:11
the that the leader on but I think this
01:07:13
issue is a really big one and I posted
01:07:16
this clip of kjp basically saying
01:07:19
nothing to see here move on and the
01:07:21
first comment was Elon which I think
01:07:24
nailed it he said a basic cognitive test
01:07:27
should not be something that someone who
01:07:29
controls nuclear strikes is allowed to
01:07:31
skip and I think that that is part of
01:07:34
the fundamental issue that I have with
01:07:37
what they decided to do here which is
01:07:39
that it just massively increases the
01:07:41
risk that you could
01:07:43
have somebody that was fundamentally
01:07:46
unelected running and making decisions
01:07:48
and I think that is a huge issue and
01:07:51
it's an enormous reason why for me this
01:07:54
was sort of the straw that broke the
01:07:57
camel's back and why I just absolutely
01:07:59
cannot vote for Biden anymore I was
01:08:01
trying to find okay I was trying to find
01:08:05
a lot of good reasons to support him
01:08:09
stability normaly all of this
01:08:11
stuff but this is not what stability
01:08:14
looks
01:08:15
like and I go again back to this very
01:08:18
simple thing a cognitive test for
01:08:20
somebody that can control the ability
01:08:22
for us to go to war or for us to
01:08:25
completely eviscerate entire populations
01:08:27
through Weaponry that he and he alone
01:08:31
controls should be something that is
01:08:33
just mandatory you can't skip it this is
01:08:35
not a thing where it's a nice to have I
01:08:38
think it's a must have and in the
01:08:40
absence of it I think it compromises his
01:08:42
ability to be a legitimate choice and we
01:08:44
do have a minimum age Sachs and what
01:08:47
shamat saying is what the majority of
01:08:49
Americans are saying ABC News ran a poll
01:08:53
8 6% of Americans across both parties
01:08:56
think Biden is too old to be president
01:08:58
59% both said both Biden and Trump were
01:09:01
too old you said on a previous episode
01:09:03
Sachs that you thought the public should
01:09:05
decide not and that that doesn't need to
01:09:06
be a cognitive test are you sticking
01:09:08
with that or do you think you're going
01:09:09
to maybe move to chat's position and
01:09:11
mine cognitive test should be mandatory
01:09:13
to be president well I think it's a tell
01:09:17
that they won't take the cognitive test
01:09:19
because they're worried that he might
01:09:20
fail just like it was a tell that they
01:09:22
refused to do the Super Bowl interview
01:09:25
which the president always does that's
01:09:26
the most softball interview that Biden
01:09:29
could possibly have and they refused to
01:09:32
do it obviously because his AIDS and
01:09:33
advisers are nervous about how that
01:09:35
might go we had the her report come out
01:09:39
where the special counsel investigated
01:09:41
that classified documents case for Biden
01:09:43
said that we can't prosecute this guy
01:09:45
he'll be too sympathetic a defendant
01:09:47
because he's manifestly scile and he'll
01:09:50
come across as an old man who can't
01:09:51
remember what happened so that's why
01:09:53
they decided not to prosecute him the
01:09:54
Babylon be had a hilarious headline or
01:09:57
take on this which was man deemed too
01:10:00
cile to stand for trial still fit to run
01:10:03
country that's the situation we're in he
01:10:05
can't stand for trial but he's still
01:10:06
running the country yeah and the thing
01:10:10
that was most damning about the her
01:10:12
report was not just what the special
01:10:15
counsil wrote but it was Biden's
01:10:16
reaction to it apparently he demanded
01:10:19
that he wanted to go out and do a press
01:10:21
conference that very day to refute the a
01:10:24
that his memory was bad or that he was
01:10:27
scile and he then did this press
01:10:30
conference and it completely backfired
01:10:32
he came across as kind of an old man
01:10:35
shaking his fist yelling get off my lawn
01:10:37
I mean that was basically the the vibe
01:10:39
that he gave to the press and then the
01:10:42
thing that he got wrong was he confused
01:10:44
Egypt from Mexico in regards to the
01:10:47
whole crisis in Gaza so it would have
01:10:50
been one thing if it was just her saying
01:10:54
that Biden was cile but then Biden
01:10:56
himself came out and seemed to confirm
01:10:59
it by making these mistakes and gaffs so
01:11:02
they know they have a problem so to my
01:11:04
original question would you be in favor
01:11:07
and will you change your position here
01:11:08
that we should have cognitive test for
01:11:10
the presidency of the United States yes
01:11:12
or no I think that the real issue is
01:11:14
that Biden would fail such a test I mean
01:11:15
that's the problem is that he is
01:11:17
cognitively impaired and the thing to
01:11:19
realize about these sorts of conditions
01:11:21
is they're not static there's a
01:11:23
trajectory to them so in the same way
01:11:25
that Biden is manifestly worse than he
01:11:30
was a year ago or three years ago yeah
01:11:33
when he took off he was obviously
01:11:34
sharper yeah whatever this condition is
01:11:36
it's going to be worse a year I wouldn't
01:11:38
ask you if there time to answer the
01:11:40
question but you were choosing not to
01:11:41
answer the question so fre when you say
01:11:43
when you say that should the president
01:11:46
take a cognitive test sure he should
01:11:47
take a cognitive test should it be the
01:11:49
law the land just like we have a 35 year
01:11:52
minimum age I'm just asking you to be
01:11:54
specific you could make him take a
01:11:55
cognitive test but the point is that
01:11:57
let's say the doctor says yeah he's
01:11:59
cognitively impaired is that gonna bar
01:12:02
someone from running for office
01:12:03
constitutionally no that's not going to
01:12:05
be part of the the Constitutional
01:12:07
requirements you have to pass a
01:12:09
constitutional amendment yes would it be
01:12:11
a good idea for someone to take a
01:12:12
cognitive test part of their physical to
01:12:14
be president sure but is that going to
01:12:16
be a constitutional requirement no you
01:12:18
can't have some doctor and by the way
01:12:20
that that's my whole point like you you
01:12:22
could take a a phys Bill Clinton did it
01:12:24
and people used to joke about his
01:12:26
cholesterol levels and he would go for a
01:12:27
jog and stop at a McDonald's so I'm not
01:12:30
saying by any means that the results
01:12:32
should bar you from the ability to run
01:12:34
it should still be up to the voters but
01:12:36
just like we would make sure that acute
01:12:39
or chronic health conditions were well
01:12:41
understood right the idea that this
01:12:44
entire body of judgment which is around
01:12:46
your brain around the Judgment of the
01:12:48
president of the United States to
01:12:51
basically say no we're good here we can
01:12:53
skip skip it I think that that's really
01:12:54
dangerous and it's part and parcel
01:12:58
of a
01:12:59
systematic set of Behaving now by the
01:13:02
Democratic party which I think is really
01:13:04
not where they want to be so the only
01:13:06
person that's really stood up to all of
01:13:08
this is Dean Phillips yep but he is
01:13:10
being prevented and railroaded from even
01:13:13
being able to put that issue on the
01:13:14
table so that the voters can decide my
01:13:17
fundamental view is the voter should
01:13:18
decide but all of these tests should be
01:13:21
done and skipping them is a form of a
01:13:24
tell as David said and I think that this
01:13:26
is the most important tell that we've
01:13:29
been given about the amount of cognitive
01:13:31
decline that he suffered and John
01:13:33
Stewart by the way had the funniest
01:13:35
version of this John Stewart's
01:13:37
essentially monologue was like hold on
01:13:40
everybody is out here giving sound bites
01:13:42
that say he is a chess grandmas in
01:13:45
private but then he shows up in public
01:13:46
and it looks like he can't even play
01:13:48
checkers yeah freeberg let me let me
01:13:49
jump to you there you shared the clip
01:13:50
with us what were your thoughts what are
01:13:52
your thoughts on this whole ISS isue
01:13:53
cognitive test I don't I mean I don't
01:13:55
think a cognitive test should be the law
01:13:57
I because you're starting to get into
01:13:58
edge cases where you'll create lots of
01:14:00
different requirements like what's the
01:14:01
Strategic test and you know what's the
01:14:04
under pressure test and there's
01:14:06
obviously uh a lot of ways that folks
01:14:08
can game and Challenge and create new
01:14:10
systems
01:14:11
for um a new criteria the system of
01:14:14
democracy however is meant to solve
01:14:16
these problems the system of democracy
01:14:18
is meant to allow individual voters to
01:14:20
say we judge that someone is not
01:14:23
equipped or prove us wrong which is what
01:14:26
the voters are saying and doing right
01:14:28
now based on the polling data the
01:14:30
majority of Voters in the United States
01:14:32
are saying we do not think that this
01:14:34
individual is equipped to be
01:14:36
president rather than respond to that
01:14:38
with a cognitive test we are stuck with
01:14:41
a single candidate and I think it speaks
01:14:44
more to the problem of the democratic
01:14:45
caucus and how the electoral system
01:14:48
works in the United States today where
01:14:50
the Insiders get to decide who the
01:14:52
nominee is for the party and those
01:14:55
insiders are self-d they are getting
01:14:57
benefits by putting someone in the seat
01:15:00
for next term because that person is in
01:15:02
the seat this term and is approving
01:15:04
things that they want this term and so
01:15:07
it feels very corrupt and it feels very
01:15:10
unfair and very undemocratic I think it
01:15:13
leads to either a sense that the
01:15:15
individual voters rise up and say we
01:15:17
want a third party because the way that
01:15:19
this two- party system works and the way
01:15:21
that the caucus system works isn't
01:15:24
or we end up seeing a really unfortunate
01:15:27
outcome where you have a similar sort of
01:15:29
like January 6 type event where folks
01:15:31
are just fed up with the whole system
01:15:32
and don't trust it and if the citizens
01:15:34
lose faith in democracy we're we're in
01:15:37
Peril but as John Stewart said in his
01:15:39
monologue it is the responsibility of
01:15:41
the citizens of this country to step up
01:15:44
and not lose Faith but to take action
01:15:45
and to resolve to a solution we have the
01:15:47
power we have the authority as voters
01:15:50
and if enough of us get together and say
01:15:52
there should be a third party or we do
01:15:54
not want this individual to be the
01:15:55
candidate something will change rather
01:15:58
than just resolve to letting the
01:15:59
Insiders set the terms of the caucus
01:16:02
third party is seeming more and more
01:16:03
viable as we've talked about here people
01:16:06
forget the case of Ross perau getting
01:16:10
19% RFK is looking pretty favorable and
01:16:14
the Third Party candidate something I
01:16:17
think the two parties dismiss all the
01:16:18
time but it is completely possible so
01:16:22
maybe this is the that this is the
01:16:23
breakout season for it as saaks alluded
01:16:27
to the her report also came out it was
01:16:30
released special counsel Robert her
01:16:33
Republican and Trump appointed AG
01:16:35
declined to prosecute Biden because he
01:16:39
feared they couldn't reach the standard
01:16:40
of Reasonable Doubt as saxs pointed
01:16:43
out and I'll just quote from the report
01:16:45
we conclude that the evidence does not
01:16:46
establish Mr Biden's guilt Beyond a
01:16:48
reasonable doubt they also said he was
01:16:53
an elderly man with a poor memory or
01:16:55
that's how a jury would view Biden
01:16:57
Democrats felt this was partisan
01:16:59
obviously and then pointed out a bunch
01:17:02
of times that Trump has misstated
01:17:07
people's names or been confused is there
01:17:09
any correlation here between the two
01:17:12
candidates in your mind chamath is Trump
01:17:14
right behind Biden in terms of this and
01:17:16
is he too old to run as well and do you
01:17:18
think the gaffs that he's made are
01:17:21
comparable I think that
01:17:25
he looks much younger and sprightlier
01:17:27
than Biden does to be totally honest
01:17:30
yeah and look at the schedule he
01:17:31
maintains is nothing like Biden's I mean
01:17:34
look Democrats are trying to both sides
01:17:36
the issue because it's their only choice
01:17:38
they're stuck with a candidate who's an
01:17:40
obvious Decline and the problem for
01:17:42
Democrats wasn't just the her report
01:17:44
which like you said jcal they can
01:17:46
dismiss as partisan because they control
01:17:47
the media the problem is that Biden
01:17:50
himself demanded to go out in front of a
01:17:52
press conference to refute the her
01:17:54
report and he ended up confirming it and
01:17:56
the people of the United States can just
01:17:58
look at Joe Biden and every time he
01:18:00
talks no matter how much they try to
01:18:02
hide him when he does speak the American
01:18:04
people are left with the impression that
01:18:06
this is somebody who's invisible decline
01:18:08
Trump although he does misstate a word
01:18:10
or two here and there does not give that
01:18:13
impression Trump maintains a vigorous
01:18:15
schedule not only is he running a
01:18:16
presidential campaign he's battling he's
01:18:18
in court every other day fighting all
01:18:20
this lawfare he is willing to give
01:18:23
interviews in fact he barely turns down
01:18:24
a press interview and most importantly
01:18:27
he's willing to do a hostile interview
01:18:29
he's willing to step into the lion's den
01:18:30
he's will to go on CNN to those town
01:18:33
halls and basically battle I think your
01:18:36
with these reporters I think Biden won't
01:18:38
even sit down for the Super Bowl
01:18:40
interview which is as much of a softball
01:18:43
interview as you're ever going to get
01:18:44
yeah yeah no no I mean that's your
01:18:46
that's your best point I
01:18:48
think who does a two or three if you
01:18:50
can't do a two or three hour interview I
01:18:53
don't think you could be presid who do
01:18:54
you think is going to run the country in
01:18:55
the second term if Biden were to win you
01:18:57
know that that is the question you think
01:18:59
it's Biden no the staff runs it listen
01:19:02
this will be this will be a regent
01:19:04
president where the presidency becomes
01:19:06
an office instead of a person and it's
01:19:08
run by powerful ministers like in some
01:19:11
sort of European system that's what
01:19:14
we'll get or happened under Reagan and
01:19:16
the fact of the matter is or or sax as
01:19:18
happened under Reagan you can say only
01:19:20
in the last couple of years of Reagan
01:19:22
exactly what happed the simility start
01:19:24
to or not simility he had Alzheimer's
01:19:25
and it it really I think started to
01:19:28
become a factor in the last couple of
01:19:29
years it was the last two years yeah but
01:19:32
the first six years were very strong 25%
01:19:34
of his presidency who was running the
01:19:36
country is the question yeah it's it's a
01:19:38
valid question do you think there's a
01:19:39
difference J Cal between 25% and 50% no
01:19:42
no it I I think it's like we only have
01:19:45
I'm just looking for other corollaries
01:19:46
where this happened and what happened uh
01:19:48
who who is no I think the this isn't a
01:19:50
corollary exactly because when he ran
01:19:52
when bran ran for a second term he was
01:19:55
cognitively sharp it was true that he
01:19:58
was diagnosed with early onset
01:19:59
Alzheimer's in the second year of his
01:20:01
second term and then that was not
01:20:04
properly disclosed to Americans that was
01:20:06
just it was just the beginning in the
01:20:07
last two years if you look at if you
01:20:09
look at the 1984 campaign the morning in
01:20:11
America campaign Reagan still did
01:20:13
debates he was still sharp on his feet
01:20:16
he was debating Mondale remember yeah he
01:20:19
gave speeches all the time he maintained
01:20:21
a vigorous schedule there was no
01:20:23
evidence of there being a problem my
01:20:25
only Point here chof is like does the
01:20:26
public has a right to know when these
01:20:28
things happen and it seems to have
01:20:29
happened a couple of times in history
01:20:31
well so I would take freedberg side I'm
01:20:33
not sure that the public you could say
01:20:35
irrespective of whether the public has a
01:20:37
right to know the public has a right to
01:20:39
judge and I think in order to judge you
01:20:43
have to make yourself available so that
01:20:44
you can actually see it in front of your
01:20:46
own eyes or hear it or listen to it or
01:20:48
watch it or what have you and that's the
01:20:50
second problem that I think we're
01:20:52
starting to see which is it is so
01:20:56
controlled how we get to interact with
01:20:59
with our leader and I think that should
01:21:02
give people some pause so that they ask
01:21:04
themselves why would this
01:21:08
happen and I think that that's worth
01:21:10
internalizing here freeberg would you
01:21:13
vote for a presidential candidate who
01:21:16
does not debate uh or who does not
01:21:18
subject himself to you know serious
01:21:20
strenuous confrontational adversarial
01:21:23
interviews those two things are those
01:21:25
two things a requirement for you to vote
01:21:27
for
01:21:28
somebody I don't know how much that
01:21:30
matters anymore the problem with the
01:21:31
two- party system is you vote for your
01:21:33
party no I know but to
01:21:35
you if like a trump a Biden they just
01:21:37
refuse to do it because we we're getting
01:21:39
to the point where maybe they're just
01:21:40
going to refuse to actually Biden or
01:21:42
Trump trump has refused to do any
01:21:44
debates so maybe they're just both going
01:21:46
to refuse to debate would you vote for a
01:21:48
candidate if they refuse to debate is my
01:21:50
question to you
01:21:51
freeberg
01:21:54
I mean I'm just in a very different
01:21:55
mindset on this stuff I'm not excited to
01:21:58
vote for a candidate that's not making
01:22:00
their priority to fix the fiscal problem
01:22:03
of the
01:22:04
US that topic it's not that topic it's
01:22:07
like for me it's like there's a car
01:22:08
driving towards a wall and you've got
01:22:11
your foot on the gas pedal it's like
01:22:12
which guy do you want to keep the foot
01:22:13
on the gas pedal like I don't care like
01:22:16
we got to take the foot off the gas
01:22:18
pedal that's the point you perceive
01:22:19
there's a difference between the two in
01:22:20
spending you're talking okay so
01:22:23
you're you're voting for a third party
01:22:24
candidate that would lean towards
01:22:26
chamath would you vote for some Party
01:22:27
candidate can't win in the system well I
01:22:30
mean they can't until they can right
01:22:31
until they do so we'll see chamath would
01:22:35
you vote for a candidate who didn't
01:22:38
debate or subject themselves to a
01:22:41
rigorous adversarial interview not
01:22:43
unless I wanted to Shadow government
01:22:46
okay sack same question to you Jason
01:22:48
you're acting as if we need to have all
01:22:51
of these tests because we don't don't
01:22:52
know whether Biden has a cognitive
01:22:55
problem or not we know I was including
01:22:56
Trump in this I'm including Trump in
01:22:57
this like the debate issue I think we
01:22:59
also know about Trump trump maintains a
01:23:00
vigorous enough schedule he takes enough
01:23:03
difficult interviews he does enough
01:23:04
speeches what about the debate part
01:23:06
that's the one I'm curious if he refuses
01:23:08
to debate would you would you still be
01:23:10
willing to vote for him you know or any
01:23:12
candidate you're acting like we need to
01:23:14
judge this question based on proxies for
01:23:17
the real issue when the real issue is
01:23:19
that we know one of these candidates has
01:23:21
basically dementia he's invisible
01:23:24
decline I'm not asking you this about
01:23:26
Biden I'm asking you in general because
01:23:28
we're talking about Norms one of the
01:23:30
Norms of you know releasing your taxes
01:23:32
I'm I'm not asking about but I'm asking
01:23:34
subject from the core substantive issue
01:23:37
we have already all agreed that Biden is
01:23:39
in cognitive decline I'm asking you
01:23:41
another question I'm opening up the
01:23:43
aperture as the moderator here and I'm
01:23:44
asking you to address another question I
01:23:46
think candidates should debate period
01:23:48
candidates should debate that's what I'm
01:23:49
asking you I don't know why you're going
01:23:50
back and telling me what I think because
01:23:52
I think that the tactic of Democrats the
01:23:54
mainstream media right now is that they
01:23:56
can't there's no way for them to paper
01:23:59
over or deny the visible reality of
01:24:01
Biden's decline so they're trying to
01:24:03
muddy up the waters they're trying to
01:24:05
claim well Trump is old too or that uh
01:24:08
Trump refused to debate too they're
01:24:10
trying to both sides the issue I think
01:24:12
we all know that Biden has a problem
01:24:15
that Trump does not have now I think
01:24:18
that chamath is also right that there's
01:24:20
a lot of Democrats who are comfortable
01:24:22
voting for a shadow government they're
01:24:24
fine with the presidency being a
01:24:26
figurehead or a rubber stamp and I think
01:24:29
the reason why they're fine with it is
01:24:30
because Democrats basically control most
01:24:32
of our ruling institutions really all of
01:24:34
them I don't think you can name one
01:24:36
major institution run by Republicans and
01:24:38
they don't need a president to do
01:24:41
anything other than be a rubber stamp
01:24:43
Republicans are looking for something
01:24:44
different Republicans need a president
01:24:46
who's willing to push back on the
01:24:48
permanent bureaucracy in Washington push
01:24:50
back
01:24:51
on the Deep State or the state
01:24:53
department creating new Wars they want
01:24:55
someone to push back on Wall Street they
01:24:57
want someone to push back on all these
01:24:59
institutions Republicans are populists
01:25:02
are looking for a Tribune of the plebs
01:25:04
basically to push back on this
01:25:06
senatorial class that runs Washington
01:25:09
for their own benefit Democrats are
01:25:10
different Democrats are the ruling Elite
01:25:13
and they're fine with the president
01:25:17
basically just being a a figurehead who
01:25:19
will sign their legislation veto veto
01:25:21
bills they like does he drop out jth at
01:25:24
this point is is this the moment we're
01:25:25
going to look back on and say this is
01:25:27
the peak when that 86% poll comes out
01:25:29
and all this negative press do you think
01:25:31
Biden is there's an exit ramp here and
01:25:34
do you think there'll be a different
01:25:35
candidate fielded by the
01:25:38
Democrats no this is uh this was their
01:25:40
way of saying we're going for it okay
01:25:43
freeer you have any thoughts on that you
01:25:44
think they'll make a switch on Biden
01:25:47
yeah I think there's a chance I think
01:25:49
there's a chance last week I had a few
01:25:50
conversations that led me to think that
01:25:52
there not super influential people in
01:25:54
the party but people that are in the
01:25:55
party that you know have connections to
01:25:58
people of influence in the party I'd
01:26:00
love to hear Jam's
01:26:01
like conversations he may have had
01:26:04
around this stuff too but people are not
01:26:06
happy in the party yeah very like very
01:26:09
much unhappy and everyone would love to
01:26:12
see someone else now the assumption is
01:26:14
Gavin is going to be shoved in that seat
01:26:17
and that's also a losing proposition so
01:26:18
if you shove Gavin in the seat you're in
01:26:20
a worse place than you are with Biden in
01:26:22
the with respect your shot at winning
01:26:24
and it's a little late now but I still
01:26:27
think that there's a chance saak what do
01:26:28
you think Biden drops out first of all
01:26:30
the only reason these people are unhappy
01:26:32
is because they think that Biden's
01:26:33
problems have become so manifest that it
01:26:35
might cost him the election that's what
01:26:36
they're worried about they don't care
01:26:39
about the fact that the B Biden
01:26:40
presidency will be run by staffers as
01:26:43
long as those staffers will basically do
01:26:45
what they want sign their legislation or
01:26:47
veto drop the question we're asking you
01:26:49
will he drop out no he's already said he
01:26:51
won't look the problem that people in
01:26:53
the Democratic party have who have this
01:26:56
concern about B Biden losing the
01:26:58
election because of this cognitive issue
01:27:01
is that there is no mechanism to replace
01:27:03
a candidate who has won all the
01:27:05
necessary primary votes against his will
01:27:10
Biden has said he's going to be the
01:27:11
nominee he has no desire to to drop out
01:27:13
so unless he changes his mind or there's
01:27:15
some Act of God he's going to be the
01:27:17
nominee the time for Democrats to make
01:27:19
this change was a year ago it required
01:27:22
mounting a primary challenge to Biden
01:27:25
the way that Dean Phillips has done the
01:27:28
problem is that the Democratic party
01:27:29
elders and the powerful people in the
01:27:31
party said we're going with Biden and so
01:27:33
no one wanted to run against him and
01:27:35
they
01:27:36
basically set up the primary calendar so
01:27:39
that Biden would win and he is going to
01:27:40
win he'll have all the primary votes he
01:27:42
needs locked up by the end of March and
01:27:45
Dean Philips is not going to be able to
01:27:47
basically stop him so there is no
01:27:49
mechanism to remove Biden he's going to
01:27:51
be the candidate you know barring Act of
01:27:53
God unless he can be persuaded to step
01:27:56
down so it's not going to happen what do
01:27:58
you think about this new third party
01:28:00
effort in DC No Labels have you heard
01:28:03
about it have you been pitched it I've
01:28:04
heard of it I haven't been pitched it
01:28:06
jamat have you been pitched it I hosted
01:28:08
a dinner at my house with some folks
01:28:11
about it and yeah I mean I think like it
01:28:14
could be interesting I think that the is
01:28:16
there room for a third party in the
01:28:17
country you think yes and I and I think
01:28:19
that there is a very reasonable chance
01:28:22
and I would put reasonable somewhere
01:28:24
between 30
01:28:27
and
01:28:29
50% that this third party gets created
01:28:33
after the presidential
01:28:35
election and that is the one of the very
01:28:38
likely outcomes of rfk's
01:28:41
candidacy because it is an
01:28:44
infrastructure that will have gotten on
01:28:48
presidential ballots in 50 States and
01:28:50
it's an infrastructure that if people
01:28:52
are willing to fund specifically no
01:28:54
labels and other people I think the
01:28:56
artifact could be a third party and if
01:28:59
if RFK can
01:29:00
get 30 plus perc support which in some
01:29:04
states I think he will I think that
01:29:06
there's a decent chance that would be an
01:29:09
enormous outcome obviously the most
01:29:10
disruptive outcome for RFK would be that
01:29:12
he won the presidency but I'm just
01:29:14
saying
01:29:15
that a very powerful outcome that has
01:29:17
huge longitudinal benefits to America
01:29:20
would be that his candidate see
01:29:22
essentially is the is the founding story
01:29:25
of a third party
01:29:28
yeah and that was the shame of what
01:29:30
happened with perau I think that there
01:29:31
was a chance that perau could have done
01:29:32
that and he must have been under just an
01:29:34
enormous amount of pressure I can't even
01:29:36
imagine what he must have dealt with but
01:29:39
that could have been the artifact back
01:29:41
then but it definitely can be the
01:29:42
artifact today
01:29:44
so we'll see they haven't announced
01:29:47
either and there's no timeline for them
01:29:49
announcing they've asked to come on this
01:29:51
program many times to just talk about
01:29:53
the concept I think like they they want
01:29:55
to have like the executive director on
01:29:57
or yeah they've asked over love yeah I
01:30:01
was like well maybe when you pick a
01:30:02
candidate that would be a better time
01:30:04
than just talking about it in the
01:30:06
abstract I think it's a great idea I
01:30:08
think that's what Americans want so
01:30:10
maybe we the elections in nine months
01:30:12
we're running out of time I mean it's
01:30:14
entirely too late to be talking about
01:30:16
new candidates who aren't even in the
01:30:17
race not going to happen I mean this
01:30:19
whole Gava new some things a fantasy
01:30:21
even if Biden were to see the light and
01:30:24
step down be pressured to do that KLA
01:30:27
Harris just said I'm ready to serve she
01:30:29
wants to be president and even though
01:30:32
she's not scile she's actually even more
01:30:34
unpopular than Biden so you have that
01:30:37
huge problem in the Democratic party
01:30:39
that even if somehow they could Prevail
01:30:40
on Biden to step down now you have KLA
01:30:43
Harris ready to serve so how do you pass
01:30:45
over her he's not stepping down he
01:30:47
didn't go through all of this craziness
01:30:49
this past week to step down they didn't
01:30:52
evade the cognitive test for him to step
01:30:54
down they're not doing all of that
01:30:55
they're not keeping him away from the
01:30:57
media so that he can step down none of
01:30:59
that is happening they're not having
01:31:01
people come out with the sound bites
01:31:03
about how he's got Mastery of everything
01:31:05
in private so that he can step down he's
01:31:08
running Donald Trump is running and RFK
01:31:11
is running those will be the three
01:31:13
people on the ballot in November of 2024
01:31:17
Biden's staff are the absolute last
01:31:19
people who would ever pressure him to
01:31:20
step down why because they'd be out of a
01:31:23
job I mean from the staff's point of
01:31:26
view having a weak president who
01:31:28
basically they administer it's not just
01:31:31
a job it's like you you allocate the
01:31:33
future as the President of the United
01:31:34
States you decide it's the best it's the
01:31:36
best situation possible for a staffer is
01:31:38
to have a president who is borderline
01:31:41
incapacitated and you get to just do
01:31:42
whatever you want you are the
01:31:44
president yeah jcal give us a rundown of
01:31:47
the Tucker Putin thing what' you think
01:31:48
just as a programming note it came out
01:31:50
it dropped right after we taped we would
01:31:52
have covered it obviously we W not
01:31:53
covering it for a reason I give him a B
01:31:55
on it could have been better but it
01:31:57
wasn't bad all things considered I don't
01:32:00
think any journalist is ever going to be
01:32:03
able to trap Putin or get good
01:32:05
information out of him or any of those
01:32:07
things and if you have somebody like
01:32:09
Tucker who is obviously very sympathetic
01:32:11
to Russia and who believes that you know
01:32:14
the West forced Putin to invade Ukraine
01:32:17
you know you're not going to get like a
01:32:18
really great interview I did think he
01:32:20
had a great moment where he asked for
01:32:23
the Wall Street Journal reporter back at
01:32:25
that moment to take back with him he did
01:32:28
a lot of softball questions he asked 43
01:32:30
questions during the
01:32:31
interview about 15 of them were kind of
01:32:34
softballs and he had about six fast
01:32:36
balls but he lost control over the
01:32:38
interview very early obviously if you
01:32:40
watched it and that's just
01:32:44
kind you're like Chris Collinsworth here
01:32:46
I'm not asking for a technical analysis
01:32:48
of the play I'm asking you what did you
01:32:50
think of the interview what did you take
01:32:52
away from it what did you think going in
01:32:55
and was there a Delta and a change going
01:32:56
out just a normal no CU I I've watched
01:32:58
the other interviews I watch Megan Kelly
01:33:00
as I watched NBC's he's done I did a
01:33:03
research on it he's done about nine
01:33:04
interviews in the last 10 years so it's
01:33:07
not like this was anything new and when
01:33:09
you watch the other ones it's just Putin
01:33:11
spinning and pontificating and nothing
01:33:14
gets accomplished he doesn't answer hard
01:33:15
questions so I had six questions I I
01:33:18
said he should ask him he did ask two of
01:33:20
them but he didn't asked the other
01:33:22
really hard questions I had and so you
01:33:25
know he's
01:33:27
not I wouldn't say tucker is like the
01:33:30
useful idiot like Hillary Clinton said
01:33:32
but he is biased and he didn't ask
01:33:34
enough hard questions I would have asked
01:33:36
three or four harder questions but the
01:33:38
you're not going to get a lot out of
01:33:39
Putin this guy's like a KGB stazzy
01:33:41
trained manipulator of media it's his
01:33:44
speciality I had two really interesting
01:33:47
takeaways the first was if you had
01:33:51
wanted to see see some crazy madman you
01:33:54
actually left that interview thinking
01:33:57
okay this is a person that's leading a
01:33:58
nation that has a deep historical
01:34:00
context that goes back to basically the
01:34:03
dawn of
01:34:05
Christianity of which he has a pretty
01:34:07
deep grasp of and it seems that his
01:34:11
decisions are rooted in a multi-
01:34:13
thousand year history I didn't know that
01:34:15
and that was an interesting takeaway for
01:34:17
me so that was a broad strategic
01:34:19
takeaway so he was a little professorial
01:34:22
he was a little historical and all of
01:34:25
that to me made it harder for me to
01:34:28
leave that interview thinking that he
01:34:29
was a crazy person in fact it's more
01:34:31
like he's a bit of a
01:34:34
methodical thinker I guess the second is
01:34:37
much more tactical and I thought super
01:34:39
salacious and I wish Tucker would have
01:34:41
dug into it which is that in 1999 when
01:34:44
he was meeting Bill Clinton he asked
01:34:46
Bill Clinton can we join
01:34:49
NATO and Bill Clinton said to him yes
01:34:52
and let's go hash out the details over
01:34:55
dinner and then that night they have
01:34:57
dinner and Clinton says my team says
01:35:00
no and to me when I looked at the Arc of
01:35:04
that interview that was one of these
01:35:06
really unique moments where it explains
01:35:09
to me so much which is a country that's
01:35:13
at the that's birthing a
01:35:15
democracy where Putin transitioning from
01:35:19
yelton had to make a lot of very
01:35:21
important difficult decisions the idea
01:35:24
that he would have made an in treaty to
01:35:25
Clinton to say Can We join NATO and then
01:35:28
to be told yes and then rebuffed I
01:35:31
suspect has an underlying amount of
01:35:34
emotion that explains a lot of what's
01:35:36
happened since and I think that that
01:35:38
should have been explored much much more
01:35:40
I think they glossed over it and they
01:35:41
went on but it was a really important
01:35:43
moment those are my two big takeaways
01:35:44
from that from that from that interview
01:35:47
did you watching freeberg any thoughts I
01:35:49
don't understand I think when he he said
01:35:51
that point he was talking in the
01:35:54
theoretical not in the I'd like to join
01:35:56
NATO because the intention of NATO was
01:35:58
to create a unified defense against the
01:36:02
Russian Armament which is the most
01:36:04
significant besides the US and at that
01:36:06
point before China's Rise um ahead of
01:36:10
China so I think it doesn't make sense
01:36:13
to be like hey there's two sides of this
01:36:16
defense system let's unify the two sides
01:36:18
of the defense system it literally is a
01:36:21
sensical propos not true not true in
01:36:23
1999 that's totally not true what was
01:36:26
the other side 1999 was where we've
01:36:29
tored this whole thing down it's no
01:36:31
longer the Soviet Union so we're going
01:36:34
to democracy it's probably just an
01:36:36
organization that looks like any other
01:36:37
organization like aan what's aan well
01:36:41
aan is an organization could they have
01:36:43
military defenses supposedly it's but
01:36:46
it's a grab bag of countries what's the
01:36:48
bricks it's another grab bag of
01:36:49
countries this this is a defense
01:36:51
Alliance this isn't an economic Alliance
01:36:53
but in that moment the the the
01:36:55
justification for NATO was the least
01:36:57
powerful it had ever been so in some
01:36:59
ways joining NATO would have been akin
01:37:01
to dismantling NATO and saying we don't
01:37:03
need this anymore that's the point like
01:37:04
why would he say why would the US and
01:37:06
the Western Europe allies say let's
01:37:09
dismantle NATO that's effectively what
01:37:11
historical context of that moment is
01:37:12
important he did it at the moment where
01:37:14
they were transitioning to a democracy I
01:37:16
don't know it's like saying let's get
01:37:17
rid of the defense system when Russia
01:37:18
was transforming into a democracy well
01:37:20
when the US was breaking down and
01:37:22
transitioning to a democracy yeah how
01:37:24
did that democracy go if there was if
01:37:25
there was massive disarmament in Russia
01:37:27
the question the question is what would
01:37:28
have happened had they said yes and
01:37:30
Clinton took a different path all I'm
01:37:32
saying is guys without prejudging it
01:37:35
that that was I think a very important
01:37:37
context that is not understood in
01:37:40
history I wouldn't dismiss it no no I
01:37:42
certainly I think everybody raised their
01:37:44
eyes like really that happened what did
01:37:46
you think of that moment and also
01:37:48
Clinton who is so steeped in history and
01:37:51
policy is not somebody that would say
01:37:54
something like that in an uncalculated
01:37:56
way only to pull back that does not seem
01:37:59
and I know Bill Clinton that is not who
01:38:01
he is so just in that context he is a
01:38:04
very thoughtful person so he would have
01:38:06
said that thinking wow this makes a lot
01:38:08
of sense on many dimensions and again
01:38:11
the concept was it was a it was kind of
01:38:14
like let's sit down for dinner and
01:38:15
figure out the details and it's like my
01:38:17
team says no and of course the team
01:38:19
would say no because it's like well
01:38:22
also we haven't heard from Clinton you
01:38:24
know confirming this or how it was
01:38:26
structured this doesn't guys it sounds
01:38:28
like it was a theoretical like that was
01:38:30
my
01:38:30
interpretation he said it and kind of
01:38:32
shocked Clinton he's like let me look
01:38:34
into that that's crazy uh what did you
01:38:36
think of that moment Sachs and then
01:38:38
overall well Jason the fact that Putin
01:38:41
sought to join NATO or at least made
01:38:43
that offer is a welln well-documented
01:38:46
historical fact this is definitely not
01:38:47
the first time this has been raised more
01:38:50
generally I think what you heard in that
01:38:51
interview once you got past the the sort
01:38:53
of the ancient history lecture is that
01:38:57
Putin when he first came to power sought
01:38:59
cordial relations with the US and with
01:39:02
American presidents and he was rebuffed
01:39:05
his attempt to join NATO was rebuffed
01:39:08
more
01:39:09
generally the US reneged on its promise
01:39:12
to gorbachov not to expand Nato one inch
01:39:14
Eastward and instead we began multiple
01:39:17
waves of of NATO expansion and I think
01:39:19
it's very important to understand that
01:39:21
you don't need to believe Putin on this
01:39:24
or trust Putin American diplomats were
01:39:27
saying at the time what the
01:39:29
ramifications of this would be and I
01:39:31
just want to put two data points in
01:39:33
evidence one is George Kennan who was
01:39:36
the father of the containment Doctrine
01:39:38
during the Cold War he was our most
01:39:40
respected Cold War Diplomat said in the
01:39:43
late 90s when they were first debating
01:39:45
on NATO expansion he said that quote
01:39:48
expanding NATO would be the most fateful
01:39:50
ER of American policy in the entire
01:39:52
postc Cold War era so think about it
01:39:55
this is somebody who's been around since
01:39:56
the 1940s he was our Chief Diplomat and
01:40:00
defined our containment policy towards
01:40:02
the Soviet Union and he said that
01:40:04
expanding NATO would be the most fateful
01:40:06
era of American foreign policy and he
01:40:08
said that a decision like that would
01:40:10
inflame the Nationalist anti-western and
01:40:13
militaristic tendencies in Russian
01:40:15
opinion it would have an adverse effect
01:40:17
on the development of Russian democracy
01:40:19
and it would restore the atmosphere of
01:40:20
the Cold War so this is somebody who
01:40:23
really knew what he was talking about
01:40:24
and warned us not to go down this path
01:40:26
and then in 2008 second data point is
01:40:29
that Bill Burns who is our ambassador to
01:40:30
Moscow who is now Joe Biden's own CIA
01:40:34
director wrote a memo to conal arice who
01:40:38
was then Secretary of State called net
01:40:39
means net and what he said in that memo
01:40:43
is that NATO expansion to Ukraine
01:40:45
specifically is the brightest of all red
01:40:47
lines for the entire Russian Elite not
01:40:49
just Putin and in other words the idea
01:40:52
of bringing Ukraine into NATO was an
01:40:55
idea that was rejected not just by Putin
01:40:57
or by hardliners but even by liberal
01:40:59
reformers inside of Russia and
01:41:02
nevertheless we persisted in this policy
01:41:04
and so I
01:41:05
think what you hear in that interview
01:41:09
are a lot of points by Putin around how
01:41:12
he sought good relations with the west
01:41:15
but what was our reaction to him
01:41:17
pressure pressure pressure we supported
01:41:20
uh the maidon in 2014 we tried to expand
01:41:22
NATO into his backyard on his most
01:41:25
vulnerable border and then finally we
01:41:26
supported these hard right ultr
01:41:29
nationalist Neo-Nazi elements in Ukraine
01:41:33
now whether or not you believe
01:41:35
everything that Putin said my takeaway
01:41:37
is that this was a rational actor this
01:41:40
was not a Madman this is somebody who we
01:41:43
could have negotiated with and in fact
01:41:46
there was still an opportunity before
01:41:47
this war to prevent the war by taking
01:41:50
NATO expansion off the table we refused
01:41:52
to do that and even today there was a
01:41:55
new press story that in the wake of this
01:41:58
interview the bid Administration still
01:41:59
rejects entering negotiations with
01:42:02
Russia even though Putin held out that
01:42:04
Olive Branch in that interview
01:42:06
so you know again it's just uh very sad
01:42:10
how we have conducted American Russian
01:42:14
relations over the last 25 years we
01:42:16
turned what could have been an ally into
01:42:20
an enemy
01:42:21
and proven George Kenan correct I think
01:42:23
this has been the most faithful air of
01:42:24
American policy in the last 30 years all
01:42:28
right uh what an amazing episode has
01:42:30
this been I think we talked about Tucker
01:42:31
and Putin enough love you boys for the
01:42:33
sultant of science the dictator and the
01:42:36
Rainman I am the world's greatest
01:42:38
moderator we'll see you next time byebye
01:42:42
bye-bye let your winners
01:42:45
ride Rainman
01:42:49
David in said we open source it to the
01:42:52
fans and they've just gone crazy with it
01:42:54
love you queen of
01:42:56
[Music]
01:43:02
K
01:43:04
[Music]
01:43:05
besties my dog taking your
01:43:08
[Music]
01:43:10
driveway oh man myit will meet me at we
01:43:14
should all just get a room and just have
01:43:15
one big huge orgy cuz they're all this
01:43:17
useless it's like this like sexual
01:43:18
tension that they just need to release
01:43:20
somehow
01:43:23
your
01:43:26
your we need to get
01:43:28
mered
01:43:31
[Music]
01:43:35
all
01:43:38
[Music]
01:43:40
I'm
