TR38: H20, GPT-5, Ajinomoto Build-up Film, Intel (featuring Wei from SemiAnalysis)

the definitely monthly podcast that you

all have been waiting for. Um, on this

week's Wait, hold on, Doug. I thought we

scheduled it literally every week.

>> Yeah, but we show up like maybe one out

of every every four. So, you know, it's

uh

>> link flaps. There's all kinds of

failures. You know, the model just isn't

converging, bro. There's all kinds of

reasons. But um today,

>> dude, you know, I think I think at least

we have a higher success of training

good models than OpenAI does.

>> Okay. Okay. That's the spicy take and

and with us today to give spicy takes

like that is uh Weey Weey is a new

member of the semi analysis empire. Uh

now the sun doesn't set on semi analysis

and also we outnumber everyone else on

this podcast today. So we here um from

semi analysis so is Dylan Patel of semi

analysis of course Jordan Schneider

China talk a future semi analysis

property and John

>> I'm currently USA funded

>> yeah USA funded and John

>> anymore John dude that's why we're

questioning why you're here on the

podcast and John

>> that's that's how we know you're joining

something else soon you're out of you're

out of funds because US aid funding is

frying

And then you'll have no choice but to

come crawling to Doug and me.

>> Leverage going down by the

>> dude. Where does that leave you? Back to

us. Um anyways, and then John of

Asianometry also a future semi analysis

platform and uh yeah that that that's

it. So uh yeah, Transistor Radio Chip

chat and we actually have a we actually

have content today which is unlike any

other episode. So uh please enjoy your

slop. Um

>> all right. Yeah,

>> we're starting off with a clip. I

prepared for this

>> that China already has

and I deal with Jensen who's a great guy

and Nvidia. Uh the chip that we're

talking about, the H20. It's uh it's an

old chip. Uh China already has it in a

different form, different name, but they

have it. Uh or they have a combination

of two will make up for it. and even

then some. Now, Jensen also has Jensen's

a very brilliant guy. And Jensen also

has a new chip, the Blackwell. Do you

know what the Blackwell is? The

Blackwell is super duper advanced. I

wouldn't make a deal with that.

Although, it's possible I'd make a deal

uh a somewhat uh enhanced in a negative

way Blackwell. In other words, take 30%

to 50% off of it. But that's the latest

and the greatest in the world. Nobody

has it. They won't have it for 5 years.

But the H20 is obsolete. You know, it's

one of those things. But it still has a

market. So I said, "Listen, I want 20%

if I'm going to prove this for you, for

the country, for our country, for the

US. I don't want it myself." You know,

every time I say like uh like 747, I

want I want Yeah. for the Air Force. So

I just wanted So when I say I want 20, I

want for the country. I only care about

the country. I don't care about myself.

And he said, "Would you make it 15?" So

we negotiate a little deal. So he's

selling a essentially old chip that

Huawei has a similar chip, a chip that

does the same thing. And I said, "Good.

If I'm going to give it to you, because

they have a uh, you know, they have a

stopper, what we call a stopper, not

allowed to do it. The restrictive is

really known as a restrictive covenant."

And I said, "If I'm going to do that, I

want you to pay us as a country

something because I'm giving you a

release. I release them only from the

the H1." Now, on the Blackwell, I think

he's coming to me again about that, but

that will be a unenhanced version of the

big one. Like, I don't you know, we will

sometimes sell fighter jets to our

country and we'll give them 20% less

than we have. Do you know what I mean?

I'm I'm I'm shocked. I'm shocked that he

uh I I haven't watched that clip yet. I

just like saw it linked and I didn't

have time to watch it yet. Oh my gosh.

This is this is an incredible clip.

>> I think we're all speechless. I mean,

where do you even begin?

>> Super so the question. No, no, no, no,

no, no, no, no. The real question is H20

was already below Nvidia's per average

gross margin. It was less than H100 and

H200 margin. Um, and now with this 15%

haircut, is Nvidia going to finally sell

GPUs at AMD's margin?

>> I think might be taxed deductible.

>> Price, they'll do a price raise. That's

what they should do.

My question is, if we're trying to hook

China on America's bad AI, why aren't we

subsidizing the export of H20s? Why are

we making it more expensive?

>> Look here, we got Mr. USA talking about

government subsidies again. Doesn't want

America to win. He wants America to be

indeped forever.

>> All right. So the so the $1 billion

dollar that we'll make, what are what

are we putting it towards? Can we just

use that money to buy drones for Taiwan,

can that be like a straight shot in the

budget?

>> How does this even mechanically work?

Can you like

>> can you like this is the thing I don't

understand like can you explicitly tax a

product leaving the like well I guess it

never touches the US shores but like how

do you how do you do this? It's I mean

it's actually illegal.

>> The export tariffs are illegal. It says

in the constitution you can't do export

tariffs.

>> This isn't a tariff. It's not an export

tariff. The H20 never touches the US

shore.

>> It is manufactured in Taiwan with Korean

memory and it is assembled in Taiwan and

then is shipped to China.

>> But American

>> never touches the US shores.

>> I think the export control conversation

because everyone's like, "Oh, it's

illegal." That feels like a skill issue

if I'm honest with you. Um, Donald Trump

has acquired like a billion dollars of

proono legal hours. I'm sure those

proono legal, you know, brain theorist

will think of something. If I had to

structure it, I'll be like, "No, no, no.

It's not a export control against

another country. It's um, it's a rev

share."

>> Yeah.

>> It's a license. Yeah.

>> It's a donation. Donation.

>> Yeah.

>> Yeah. It's a fee.

>> Put it in the super pack. Hold

on, hold on, hold on, hold on. This it's

it's not 15% to the US government. It's

15% to Donald Trump's re-election

campaign for his fourth term, you know,

because he had he had two terms

previously. He's on his third. Um, the

second one is kind of contested, but you

know, and this is re-election for his

fourth term.

>> Well, my question is,

>> did Jensen bring this up? Because once

Trump had a number in his head, he was

like, "Oh, he said 10. I'll say 20 and

then we'll meet at 15 and then we have a

deal and like you stop thinking about

whether or not it's a good thing to sell

the H20s to China and then you're just

in this like oh I'm in like dealmaker

reptile brain

like hats off to the Nvidia GA team if

they if they came up with that real 10

points for them.

>> Um I mean I mean Jensen's gone to to

meet Trump how many times? Um I think I

think like there's people within the US

admins who wanted to sell H20. there's

people who want to ban it. Um,

and somewhere in the middle were all the

people that are like, "We want to make

deals with China." And Trump probably

saw his advisers all around him. He's

like, "Well, if this thing is so like

hotly contested, you know, maybe we

should get paid for it." I I I do you

really think Jensen would have suggested

this?

>> I think if he's smart, he would have.

>> Or do you think Trump came up with it

himself? I think I think there's

literally only two possibilities, right?

Jensen came up with it or Trump

suggested it. because it is so illegal

that you have to actually be like you

can't be an export control lawyer and

come up with this because it's just not

it's just

>> yeah but export control lawyers have

proven that they're very incompetent and

enable to actually do their job. So

that's besides the point.

>> So just for the record where do we stand

on selling black wells to China?

>> There's a price for everything. Jordan

Well, like like let's be honest. I mean,

if the if GPT5 is as good as China can

get, I think it's fine.

>> Just sell it.

>> Whoa.

>> John from the Raptors.

>> I think I think like Black is obviously

like very bad.

>> Like like why does this matter? Like

just keep letting him sell. Let him buy

them.

I'm tired. Okay. But but

>> does this increase or decrease invasion

risk?

>> This decreases invasion risk, right?

>> You tell me. I don't know how how much

how much better are your drones going to

get with those uh

H20 or you know [ __ ] Blackwell trained

models

to command your drone swarms.

>> I don't know. Wait,

>> like Dylan's going at like one frame

every two seconds.

>> Dylan's going one frame a second.

John John has a point.

>> I I want to hear I want to hear your I

want to hear your [ __ ] shilling H20

sales. Like, let's hear it. Let's hear

it.

>> Hey. Hey. All right. All right. We We We

may have We may or may not had a a bit

of a war in the semi analysis slack

between me and Dylan. Uh taking a little

bit different sides. I I've been I've

been of the mindset that we should flood

China with compute cuz as John said, if

this is as good as we're going to get

with GPT5, I mean, might as well make

some money off of them, right? before

this before this AI bubble bursts. So,

we should uh we we should just

>> believe you literally left your hedge

fund if thought the AI bubble was going

to burst.

That you thought you should leave your

hedge fund.

>> Dylan, I I'm not known to make

economically wise decisions. All right,

this is the pure vibes.

>> Yeah, he

>> but also

>> the chart's pretty good, dude. The the

the the upside down pyramid that was

pretty good. I really like

>> explain where it's it's a it's a it's a

chart to introduce myself Jordan you

know it's called from riches to rags and

it basically goes by aum starting with

the first fund I work for which was

pretty large and then the second fund

which we shall not name for now and then

you know the fund I launched and then

semi analysis at zero billion aum and uh

it kind of it kind of it kind of uh it's

it's consistent with my thought process

over time but no I mean I I think I

think I think I have somewhat of a

point, right? Like I'm I'm I'm I think

Jensen wants to sell as many of these

chips as he can to lock the, you know,

China into our ecosystem. The one thing

that's durable is our chip ecosystem,

right? Um and yeah, it probably reduces

the need for China to invade Taiwan if

we uh sell them what was Trump's words

like three-year-old hardware. I mean,

the hopper's been out for like two

years, so but if we if we tell Trump

this is three hardware, I'm sure he'll

believe it.

And also, if AMD can't do it, China I

guess China will probably do it faster.

Let's be real. The real rock and bull

case is that uh Chinese hardware

engineers are going to fix the AMD

software stack faster than uh they can.

And boom, parody, almost instantly.

>> Give them an economic incentive to fix

Rockom. I love it.

>> Dylan, what's the what's your what's

your side of this? I I'm I'm curious how

like how was how was AMD able to get the

yes we can also pay 15% for MI308X? Like

was it like Trump talked to Jensen and

then they agreed to 15% and then some

bozo and they a random person the admin

is like yo AMD by the way Jensen got

this deal for y'all and they're like oh

sick. Like is that what happened or

>> Jensen probably calculated the operating

profit break even for AMD and said you

know what 50%'s good enough there to

nique out them

>> those guys are making [ __ ]

>> this is like AMD um this is like Intel

like giving AMD the low-end like PC

laptop market in like Brazil.

and like saying, "Yeah, okay, you guys

can. We're not a monopoly. We're not a

monopoly. Look at all these units

they're shipping."

>> Yeah. Jensen does a giant uh you know,

huge brand and he's like, "Lisa makes

exactly $1 from this deal." She, you

know, she does she she did all the work.

She's like $14%.

She would have made a profit. 16, you

know, she would lose money. $1, 15%.

That's how the deal came about. Jensen

and his Galaxy brain use all those GPUs.

And uh Lisa's Lisa is probably happy

about it.

>> I still want

>> I actually don't understand why it's

illegal though. Like can you explain why

it's illegal?

>> I mean I can read you. It's like clause

four of the constitution. You can't

>> Yeah, there is no export tariff. There's

this is not being exported

in the traditional sense of it. Sure.

But since it's an American product made

by American company and everything that

is an American IP is, you know, in a

very broad interpretation and export,

then I guess I don't know, man. This is

for the lawy lawyers to figure out.

>> I mean, the United States administration

has seen the US Constitution as like a

flexible thing. I mean, whatever. Like

sometime this will get litigated like 3

years from now.

>> This is a skill issue and and and Trump

has more lawyers.

>> It's not it's not export. Like there's

literally nothing in here about an

export. Um now could they create a new

designation and be like oh we're going

to tariff we're going to tax this sale

of this designated good at a certain

rate then fine

>> like but that also requires

congressional approval

>> find something.

>> It's one big 15% admin admin fee l like

admin processing fee.

>> Yeah.

>> Anything else we want to talk about

before we move on? Well, yeah. I want to

hear the other side.

>> What if what if instead of drones, what

if instead of drones, we fund GIS export

restriction, uh like end use

verification on GPUs and um you know,

prevention of circumvention of export

controls on equipment and HBM with this

billion dollars that that the US

government makes. Ooh. Ooh. Now we're

talking. It's

>> why can't we just let them get cloud

access to the H20s they buy and put them

all in Malaysia and Singapore

>> then Singapore uh they were doing that

with

>> they were doing that in GB 200 first of

all and uh why they need to be in China

is because China wants that. What do you

mean? Like

>> but like why do we need to if they're

going to buy them any like like how much

purchasing is diverted from Nvidia or

how much demand is diverted from Nvidia

to Huawei because your cluster is in

Malaysia instead of Changhai?

>> I mean if you're saying that like I

don't know what you're trying to say

here but if you're saying that like

Huawei can't be exported to Malaysia

right? So then you

>> No, no, no. If I'm a Chinese, if I'm a

Chinese hyperscaler

>> and I'm going to be one of these people

who's buying these H20s, if I can have

access to the same H20, but it lives in

Malaysia.

Like, am I going to buy less of them

because they're in Malaysia versus being

able to have my own data center in

China?

Yeah, why not? I mean, China's got more

power. It's got all there's bandwidth

issues. And then like you is didn't

Xiinping say that every province wants a

data center in their kind of in in their

city? Look, everyone looks bored about

about this. Let's move on.

>> All right.

>> I can see everyone's faces and they're

just like,

>> "Yeah, dude. I don't care, bro. We're

dude. This is the GPT5 episode and

you're telling China GPT5 episode we're

talking about."

>> Yeah. This is so US A funded, dude.

>> Dude, US A is still impacting us today

even though we defunded them. I'm going

to have to do another show tomorrow

because you guys are not taking the bait

here. All right, GPT5.

>> You can you can send this to whatever

the sub.

>> Why did you get us so excited?

>> Wasn't nice of you.

>> Um,

>> yeah. Dylan, did you think OpenAI

actually thought GPT5 was going to be a

good model? Cuz the amount of hype was

disproportionate.

>> Like, don't even just

>> It's just a black box. Well, you Well,

you mean

>> they think GPD5 is amazing if you're if

you're finance free.

>> That's not true. People are freaking out

in finance right now, dude. Dylan,

you've been fielding calls all day

saying, "Is this over? Should we cut our

GPU estimates in half?"

>> Yeah, actually that's true. I mean, no,

I think uh there was Was there the

Twitter hype train that there usually

is? Do you remember the strawberry hype

train for 0 for 01 03?

There was just not like because because

you know like half of Twitter or X now

is funded by OpenAI like OpenAI

influencers are real. So they just

weren't that hype about it.

>> I don't know. GPT5 feels like a Oh,

goodbye Dylan. Um feels like a launch

for the pores aka all the free users.

That's what it's about in my opinion.

It's my

>> Is there any indication that GPT

whatever is underneath GPT5 is actually

better than whatever they're doing or

that they really did just add this

router layer on top?

>> Uh, we got to I don't know if we're

going to we're going to leak the hot

take on Transistor Radio, the most alpha

drop ever or not. Is Dylan But Dylan's

not here so

>> he's probably

>> Yeah. Changing on the streets of SF. Um,

wait. What What's the Okay, honestly,

let's just chat about Let's just talk

about GPT5. First and foremost, router

layer. Uh, it's effectively probably the

same. It's like similar to 4. My our

vibe is that it's not bigger by a lot.

Um, it's probably a similar maybe a

larger model. I don't know what the

pre-training is. I don't know. But like

I think the thing is everyone expected

the pre-training to be up a ton and it's

clearly just RL to hell, right? like

this is a an RL max um model and I think

>> it's it's like GPT 40.5 or 40.2 or

something like that you know like very

>> 4.5 actually I think that that's the

number I'm thinking of but I think they

actually used that already so so they

had to do

>> I think they had they were like whoa

whoa whoa we run out we can't do 4.6 six

and

>> their naming quality, the naming and

quality increased, but inversely the

model improvements decreased.

>> Got got worse.

>> You can have one or the other, right? An

incredible model with [ __ ] naming or a

really mid model with really good

naming.

>> Yeah, that's

Yeah, I think it's a I think it's a I

think it's a name issue for real. I

don't think they should have called it

five. Um but I don't know what they else

they're going to call it. So, and they

wanted to do a product launch. I mean, I

think they they collapsed the model

architectures, right? Like before they

had this Chinese menu of of of model

names and you had no idea which one was

the best.

>> Well, no, no, we we got Quorum here. We

got Quorum.

>> Chinese menus are really big. I think

that's like

>> Yeah, Chinese menus are really big,

right? And they're real comp.

Yeah, it's like hot oil or uh like

>> sour or it's steamed.

>> Some would say this dish was pre-trained

like all the other dishes. It wasn't a

new dish, you know. It was uh it was

using this but you know, same chef,

right? Same chef and same ingredients

but but new improved.

>> I mean

>> I I think I think they tried to like

Yeah. My pet theory is they're trying to

save GPUs for actually like research and

training and by collapsing the the the

menu count, right? You know, like like

like In and Out, they have like two

items and they're super profitable. So,

I think that's what they're trying to

do.

>> But then so why would they call this

five? Why not like 4.6 4.7 or something

like that?

>> No one's going to tune to 4.6.

>> Yeah, no one's going to tune to 4.6. You

can't you can't post a you can't tweet

an image of the Death Star and then call

it

>> dude. Also, I got to say they really

shouldn't have posted an image of a

Death Star. Like that was pretty

>> Come on. And then did you see the Did

you see the the Gemini? They tweeted the

There's a guy a guy who's like dev

relations and he tweeted the Yeah. the

Rebel. Yeah. The X-Wings. And you're

just like, "Bro, Gemini as the Rebel

Alliance." I mean,

>> can we talk about the energy on their

live stream for a second? There was like

one person who seemed happy to be there.

>> Maybe one and a half.

>> I kind of felt bad for them. It's like

if you're not into it, like just don't

do don't don't talk on it. I don't know.

>> Do you think do you think it's like a

transistor radio where we put it on the

calendar and then we have to come and

talk about things even when we're not

about them?

And so

>> getting launched on Transistor Radio.

You heard it here first.

>> So they're just sitting there and

they're like, "Oh [ __ ] I got to be

really excited about this." And they're

like, "Yeah, today it's it's really good

at I don't know. It was a pretty low

energy launch." Um, and then also maybe

we should talk about uh like we actually

just tweeted about this uh about the the

Sweet Bench uh hack, right? because

everyone made fun of them for the chart

and that was pretty I'm not gonna lie

that was pretty funny. Um because you

know clearly it was like a bad chart.

Maybe obviously maybe maybe the whole

thing was done

>> I thought they planted that bad chart.

>> Maybe maybe they did the whole thing

>> they showed

>> they did the whole thing with with GPT5

so it just didn't do a good job

>> that or they swapped out the model last

minute. They had a much better model

that fit the lines. Ooh,

>> that's that's a good Now now we're

cooking way. I get why we're cooking.

>> If you're if you're raising 500 billion,

if you're raising on a $500 billion

valuation, you want to be able to show

the prospective investors something

secret you have coming out later,

>> right? I mean,

>> that's part of the feature making.

>> They they could literally think like,

yo, mission accomplished, right? We

soped the most gullible topish Japanese

investor in the history of Earth to give

us another10 billion dollars and we have

this stuff coming. So like we're good,

right? Like we can just we can just

launch slop for the rest of until the

next race

that that's one

>> I dude I

>> are we worried about like talent density

at OpenAI yet? How are we feeling on

that front?

Mas a way question, not a Doug question.

>> That No, that's a that's a Dylan

question. I mean, I don't think

>> I I think there was there weren't that

many core research that took the offer,

but yeah, the big question is what are

they cooking over there because what

we're being served is uh AI slop

and we know they have more GPUs.

>> We know they have a ton more GPUs.

>> Wait, I want to I want to freaking take,

dude. I don't want to make a take right

now,

>> but you should put out the take, man. Do

it.

>> Okay.

Sam tweeted about it. Sam tweeted about

it right?

>> Did he Did he tweet about it?

>> Sam tweeted about how many people on the

free tier, which is like the vast

majority of people using this were using

thinking mode.

>> Okay. Okay. Okay. That's true. That's

true. Thinking is increasing. Okay. So,

let's let's let's actually talk about

what GBT5 is. Big brain galaxy finance

time, brothers. Okay, TLDDR, there are

650 million free users and those guys

cost a bajillion dollars and there's a

prito curve and they hit their rate

limits and all that crap. Um, but that's

where most of the the users are always

going to be and I think GPT5 is for

them. Five is probably a cheapish as

hell model to do and then with the

routing could be really cheap because

instead of asking these really dumb

questions, it'll go to the the smallest

model possible. But even bigger brain,

what five is is the route routing is the

launch effectively. And what we think

that this could possibly be setting them

up for is a new way to monetize for

users specifically for advertising andor

some kind of agentic purchasing. So just

think about it this way. One of the like

one of the core things of the router is

you ask you know how hard is this

question to answer? what if you have one

more capability that says how economic

is this query to serve effectively? And

so today they're putting in the

infrastructure to launch what will

eventually be some kind of new some kind

of new advertising way. Um I have crazy

speculations about it just cuz I think

uh you know I think OpenAI is allergic

to the concept of like display ads and I

don't think anyone wants display ads.

But let's just kind of also talk about

the timeline here. Uh, I can't pronounce

her first name. Simo got Fiji. Sim Fiji.

Yeah, I think I think but it's it's

spelled it's spelled weird. So, I'm just

going to Simo. Uh, Simo got hired in

May. Okay. And then all of a sudden,

last month on a podcast, Sam Alman is

like, we're uh so last year's like

that's the last possible thing we'll

ever do. And then last month on a

podcast, he said, uh, we're seriously

considering putting advertisements in

free pod uh in into the free product.

And then today you get a router. So what

I think this is is uh you know the

silent launch of um yeah what could

potentially be a new way to monetize the

vast majority of their users. And like

let's be real dude do they really care

about pro and plus users? Like those

guys are losing them money handover fist

especially the pro tier right? Those

guys are and like you know the pro high

taste people on Twitter. They're all

complaining because like now they can't

blast that 03 deep research. Dude, those

guys are the guys like burning a hole in

their pocket. Um, you know, the 1% of

users that are using, you know, 20x, you

know, they're uneconomic. And so what

this is really focused on is the vast

majority of chat GPT's consumer base,

which is a free user base and puts into

place what the infrastructure that will

eventually support a new monetization.

So yeah, that's my flaming.

>> Explain how the monetization works. So,

like when I search shirt search for like

give me the best home insurance, it like

cranks up an incredibly good answer with

some special links.

>> So, special links is one way, but if

you're going to be like, and this is

just I feel like now we've we've

Congratulations, Transistor Radio. You

got the whole alpha alpha drop from me.

Okay. Um, this we we posted about this

today and I think in uh in semi analysis

like you know right before this podcast.

Um, so pretty much I think the and

please can you delay this a little bit?

Um, uh, just so like our, you know,

freaking fund guys can read it for a

little bit or you know, the industry

guys. Um, so, uh, the thing that I think

could happen that really sounds

compelling to me is like agent

purchasing. So something where it's like

okay I can just imagine Sam Alman be

like you know we're against ads but

clearly what track GPT is becoming is an

adviser in your everyday life a

companion a person that you trust as a

relation you know to help your life and

that's why today we're advis we're

launching chat GPT advisor who will

agentically do things on your behalf and

then the whole monetization is

effectively a referral scheme right so

it's like okay chat GPT book me a flight

in uh you know, the last week of August

and see what the best time for my

calendar is. And then on the back end,

it's linked to whatever if booking and

and Expedia still exist, whatever. And

they they do the whole thing. They

purchase the whole thing. It goes to a

big modal that says, "Are you cool with

me purchasing this thing?" And you say,

"Yes, your payments linked up or some,

you know, crazy bullshit." And then, uh,

everyone on the back end gets paid in

referral fees. And then what you've done

is you completely collapsed the entire

search, you know, the entire customer

journey. Instead of them paying for XYZ

search tickets or, you know, search uh

keywords, all that stuff, you can, you

know, and then I think honestly if that

if they really did that, I think that

that would be a really really really

good product that I would probably use

very quickly. Um, especially as it got

better.

>> That's not advertising though, right?

That that's more like

>> OpenAI financial advisor, Open AI

doctor. I mean, there's so much like

>> Yeah. And and it and it actually fits

into their ethos, right? Because they

want to be a consumer super app. This is

a consumer super app, right? And so it

would be like, okay, my favorite example

is always the DUI lawyer cuz like the

lawyer query keywords, I don't know if

you know, they're like number one, dude.

The C the click-through rate on a

messthemoma lawyer is like three bucks,

which is like unheard of. So if you just

click the link, it costs someone $3 if

you just Google meloma lawyer right now.

so valuable because the the intention is

so high. But imagine like uh you know a

lawyer agent. Okay. Uh I just got a DUI.

I probably need a lawyer right now. It's

like okay uh where are you? What state

are you? Blah blah blah. Let's make a

plan. This is what you should shouldn't

say. I'm going to reach out to a lawyer

on behalf. I looked at a few on I looked

at a few. Made an email already. Here's

the phone number when you when you call.

Would you want to start this plan? And

you say yes. and someone on the other

side is like, "Dude, I'll pay 10 bucks,

20 bucks, 100 bucks for that lead

because the DUI lawyer is going to cut,

you know, 10 to 20 grand from from that

whole engagement." So, the cost of

acquisition is so so so so low. And then

just imagine the conversion if you have

this really welllaid out thoughtout plan

that tells you it handholds you. It just

says click this button and we will

start, right? That level of conversion

you can throw a lot of tokens at too.

You could you could spend a hundred you

know you can spend a hundred bucks of

tokens or something if you know the

conversion rate is 100%. Right? So

that's kind of like against the point

though like if if the lawyer if they

know that the lawyer is paying money to

be featured by Chad GBT isn't this isn't

that kind of like a problem.

>> It's your it's your sophantic advisor.

Okay. Slurp slurp sucks suck and tells

you what you want. Um that that's that's

probably true. But I think there's also

a little bit of um I think there's also

>> in the beginning there's going to be

it's going to be like a you sign up for

the auction and then the complete the

auction god router thing tells you what

is the right or wrong answer and then

purchases or doesn't purchase completely

based on whatever it wants to do. So

like in like let let me let's use the

flights example because I think that's a

little bit more fair. Uh can you book me

a flight? Sorry, we only have Delta

flights and we don't have American

flights. Well, then boom, everyone's on

there. Then all of a sudden you have 80%

market share and then it's just like,

you know, airlines offering flights. It

quickly becomes the status quo. Um, you

can also argue that the keywords on top

of a Google search are the exact same

thing, but people people pay a lot of

money because it's extremely economic

and it seems to work out. I think it

will just be like uh I'm sure some

clever aggregator in the lawyer case

will like add it all up and be like this

is the lawyer who's highest rated near

you that works with your type of client

and at this cost rate you know and it

will maybe it does the final choose and

between the last two at you know not ads

inventory that is being offered for the

take rate there is a choice but if you

were a lawyer and you knew

that you could put your inventory into

this, you know, DUI agent. I love DUI. I

don't know why I keep using DUI agent

into the DUI agent bot god. Um, I think

I would always say yes until the

business stopped working cuz how it

would feel on the other side is you wake

up with uh inbound that is so hot and

ready to go. It is effectively a whole

thing and it's and maybe maybe it's so

agentic that it you know you start you

talk with the agent first and the agent

gets this all kicked off instead of you

having to go back and forth. this whole

thing. There's a lot of different

verticals that you can start. If I had

to guess, it's going to be flights

first. Um, also you should something

that kind of leaks or not leaks, but

like kind of points to they have

customer service benchmarks that I felt

like were weirdly out of place. It was

telecom, healthcare, and uh and

airlines. Uh, and that kind of I think

um like a good example is uh T-Mobile. I

just moved to a new apartment. Can you

set up can you set up my internet for

me? And then it just says, "Okay, here's

the two service providers in your area.

Do you want to use X or Y?" And it's

like, "Uh, which one is faster and

cheaper?" And it's like, "Well, this is

cheaper and this one's faster." And

you're like, "I want the faster one."

And then on the back end, you know, that

customer is going to be there for 2 and

a half years, whatever. There's great

LTV cows. They know they understand

their customer. And they're like, "We

would be more than happy to pay 3% of

the contract value today if we knew that

this was a conversion." And so that's

that's a lot of money. And that's more

than like they're more than willing to

to to bake that. That to me is something

I can see as like an an agentic

purchasing process that people would

really appreciate and like the customer

experience, but also they would really

um I think the the the end seller into

the market would really enjoy. It's a

it's a two-sided marketplace. So

>> John, I think this is literally like the

affiliate business model, but you know,

reinvent it, right? And the affiliate is

the AI. I mean, they're they're already

holding a lot of traffic effectively

hostage, right? If you look at u like

websites and how much traffic they're

getting from AI, it's like growing a

lot. Uh and so my guess is like that's

the end state what Doug's describing,

but the intermediate state is just like,

hey, we have a bunch of traffic to send

you pay us or else, you know, or

something like that. Um yeah, like

people

are is it are you okay?

>> What's up?

>> Your your voice got really weird. I was

like, "Bro, you spoke a cigarette just

now."

>> No. Also, he just sounds all formal and

[ __ ] Doug, is this how he sounds on

client calls?

>> Wait, dude. I don't

>> John, I'm new. I'm new. I'm new to being

famous.

>> Weey doesn't have Weey doesn't have the

temperature control. Okay. I don't think

he he's either serious or he's extremely

unserious, of which there are tales in

the Slack. Or not unserious, but just

Weey. I don't know. You speak for

yourself. You're on the pulpit. John, is

it now that we admit to everyone who

actually invented the Hitachi magic wand

meme?

>> It was way

>> you heard it here first. It was heard

here first.

>> We way explored bravely by himself. He

plumbed the depths and to figure out

what John's greatest potential video

could be. He's like, "This is something

I am a user. I'm a net promoter and I

and I need to promote it to my most my

most influential friend to make a

YouTube video and that's how it all came

to be.

>> John, what was your character breakdown

on that one?

>> Have you made this video yet?

>> No.

>> I have the research for it.

>> This is you know you know. Okay. Okay.

If if if John joins the semiinal's

empire, he might not announce it, but

you'll know. There will be signs and the

and the sign the sign will be the first

day pinned to the top of YouTube.

Hitachi, the story of the magic wand,

east meets west

or eastern pleasure. Yeah, there you go.

Much better. New

>> pleasure.

>> All right. Can we talk Can we talk

intel?

Uh,

>> I kind of want to talk more on magic

wand but

>> yeah,

by all means. By all means, John, what

have you found so far? Give us a

preview.

>> I'm not going to give you a preview.

This is just ruins the video.

>> You said you wanted to talk more about

it. What What else do you want to talk

about?

>> How about some of your primary due

diligence so far?

>> I'm trying to track down some of the

people who used to work in that

business, but um I've gotten some emails

But yeah, it's it's been

it's interesting. Yeah.

>> Okay. There's I was talking about user

experience, but I'm

>> the the the the forgotten salary men

that invented one of the you know, God's

greatest gifts to to to

>> Have you Have you seen the AI Monoto

buildup film? A salary man hype video.

>> Yeah, dude. Doug, that is one of my

favorite films of all time. The one

>> 100% so good.

This is literally commissioned by a Jim

modems himself and it walks you through

how

>> Wow, John.

>> Oh my god, John.

>> Okay, John, you're Dude, come on. This

is what you're supposed to be doing. Way

you can explain. It's so good. I'm

happy. All right. All right. I don't

know what his name was. Let's not call

him Mr. Ojinamoto. But anyways, this

guy, he he discovered this buildup film,

this resin that had no product market

fit, right? And so like he was I don't

know he was experimenting whatever

figured this out and they didn't know

what to do with it. They couldn't sell

it. It was too expensive. And so you

know he his team was getting disbanded

but he like placed every one of his team

members into different parts of ginamoto

which is a giant like uh condiments and

food and

>> they make MSG.

>> They make MSG and and build up resin.

And anyways, and then years later when I

think Intel was the first customer,

right, with with with E-MIB or whatever,

basically like the, you know, multi like

advanced packaging. Uh, he like called

the the the Avengers back and he got his

team back and, you know, he saved their

jobs. It's the greatest

>> really great story. He talks about he he

talks about how like everyone gets fired

and it's just him and he's super built

on trying to figure this out and then he

h there's like a business so it's like

you know the tech guy there's even a

business guy who believes in him and

kind of keeps him funded and then

eventually it's like you know it's like

20% of their sales and like you know 30

or something percent of their profit.

It's like the most profitable part of of

ABF and it's like or aim it's like now

what I think of them for at least and

I'm just and they have this amazing

corporate video. It's like 30 minutes

long.

>> It's It's like a documentary.

>> Is this a video? Is it a video?

>> Yeah.

>> It's like It's like embedded in the

bottom of the film of the page of AB

about ABF.

>> It's actually good.

John, this is the content you should be

doing. The hell, dude.

>> I mean I mean I'm trying to not do too

many Japanese ones. So, it's like one

Japanese, one Korean, and one Chinese,

and one like I'll get around to it

again.

>> What about Malaysia? Huh? What about

What about all the other ones?

>> Yeah. Hey, what about Thailand, dude?

>> Indonesia. So much disrespect for

Indonesia.

>> So much so much disrespect, dude.

>> Yeah, but the CPM rates are just

terrible in Indonesia.

>> But the market is so big.

>> Instant.

>> John, you sound like an AI model. Stop

anchoring everything on CPM, dude.

>> All right,

DUI lawyer a chance.

>> Yeah, give that up.

He has low MPS,

>> but he says I'll do it for a lower rate.

>> I'm just trying to get my feet wet.

Okay.

>> I got some laws you have to pay down.

>> Yeah. I can't wait for these AI models

to select your DUI lawyer based upon

their background. Like, you know, how

disadvantaged they were when they grew

up.

>> I you know,

>> dude, I

get affirmative action like through

affirmative action.

>> True. broke AI

>> CPM adjusted for affirmative action. I

love this.

>> You guys, you guys have this totally

wrong. You know, now you have the second

biggest market of AI customers is not

the free users. It's the paid lawyers

who are like create the most sickopantic

slop so I get [ __ ] selected for this

DUI lawyer contract and they're paying

$500 a month to beg. And they're like,

"Dude, the lawyer next door, he pays

$600 a month." And they're just, you

know, they're just racing, racing,

racing, racing so they can get those

leads, dude. Oh my god. I can see I

could see it all. It's It's going to be

beautiful.

>> I'm excited for this new world, this new

ecom world where

>> we have to write stick authentic product

descriptions or pitch it on a video to

the chat chatbt AI.

>> Wow, that's such an amazing insight.

>> You're such a good AI bot. And I want to

tell you today why I

>> You have great taste in liquor.

>> Yeah,

>> I'm really proud of you for having that

extra smearing off ice.

>> But next time, buddy, don't get behind

the wheel.

>> Or don't. It doesn't matter to me.

>> As a DUI lawyers with plenty of DUIs

myself,

>> you can really relate with the customer.

Okay, that's pretty

like you know there'll be like a little

AVI. It's an anime PFP future. It's like

a slobinly drunken lawyer. Oh, I can see

it. I'm so excited, guys.

>> Over. So am I.

>> Oh, buddy. Your rent's late. So is mine,

dude. But I'm going to need that check

soon. Um, so anyways, um, Jesus, dude.

DUI lawyers, man. There there's a bull

market somewhere and that's there's a

huge opportunity for everyone. Actually,

they're not even drinking, so this might

not even We need to think of a better

like a a Gen Zcoded DUI lawyer

equivalent. No, there's going to be some

like meme. Like the financial advisor

thing is really terrifying because

there's like a meme stock purveyor

like buy my ETF.

Yeah. I don't know, dude. I'm like post

terror. I'm just like sure, dude. Let

let Mo Moolok happen.

>> All right.

>> Yeah.

>> Can we talk intel a little bit?

>> What was intel about? I don't know, man.

Trump tried to get him fired.

>> He changed his mind. Apparently they're

buds now.

>> You're like baiting us. I think you're

jating us,

>> dude. LBT I think everyone. So, okay,

let's let's actually walk through the

whole thing. Trump is surprisingly well

informed actually. He's like the

Blackwell who which is actually a lot

better. It's it's even better than the

crap total dog Huawei chips. I'm like

this guy is reading semi analysis, man.

What the heck is going on? He actually

he's been consistently pretty informed

and the intel thing. I mean, it's like

>> right at the you can you can go back to

the first Transistor radio we were

talking about.

>> Yeah, he's he's actually really he's

Dude, the guy's clocked in on

semicopters. Like I'm like Okay, so have

you heard the like how he's been

described? He's like very instinctual.

He's like, "Oh, this guy isn't lawyer

loyal. He's fired." like very but like

he's very for whatever reason super

duper duper clocked in on Intel and on

semiconductors. So since day one, you

know, we have pharma tariffs, but that's

like who really cares about that? It's

all it's been semiconductors. This man

has been a semiconductor man since day

one. So yeah, man,

>> he knows.

>> Do you want him fired?

>> Liboutan. No, dude. Come on, man. The

car just got the wheels again. You want

to do you want to do another car crash?

Bro, there's only so many car crashes

this thing can handle. Just keeps you

know the like the car crash. They put

the wheels back on. It's dented

everywhere. It's leaking employees. It's

It's all We can't fire another guy. And

like who's going to do it, guys? It's

literally Dylan Patel's next. Like

that's that that's like like we are

running out we're running out of

chances, guys. Like I'm not joking.

Who's who's next? Anyone got idea? I

don't.

>> I do think it's okay to scare him a

little bit.

>> Um because it is

>> I think that's how that's the art of

that's the art of the deal, bro.

>> He said this guy should get fired and

he's like actually I like his story.

>> You better work hard.

>> The quote is his rise is an amazing

story.

>> Mr. Tan and my cabinet members are going

to spend some time together and bring

suggestions to me during the next week.

>> Thank you for attention to this matter.

We're paying attention here on

Transistor Radio. But shouldn't

shouldn't Trump be

>> tariffs and then

>> and then minimum guaranteed price for

CPUs? I don't know.

>> Yeah. Shouldn't he be threatening

Intel's prospective customers that

refuse to commit to 14A? I I mean, he

seems to be threatening the wrong person

here right?

>> Get all the tech giants to invest $5

billion into Intel and then

>> dude, I know where the billion dollars

is going.

DUI lawyers.

>> After the DUI lawyers is going to Intel,

he's going to he's he's going to he's

going to piece this thing together one

rev share at a time. How much are you

paying for semianalysis

leads

when people ask uh you know when someone

in the White House ask chat GPT what

should I do about you know

>> um that's a great question

you have such an insightful brain there

Jordan Schneider uh we will have um

soped and sloppied the training data so

much that they will come to us anyways.

That's what real that's real. It's SEO

and is making 100,000 10k word bangers

from semi analysis. The training set the

entire training data set is just us.

They try to reference another firm. It's

still semi analysis. That's how we win.

So wait wait wait coming back to your

thing maybe this is the way it works out

is like in Intel has to like sell itself

to

>> I don't know

>> who

>> who Google Amazon

>> oh no Google

>> just like or just like sell like little

bites of it

>> like the capital's got to come from

somewhere right

>> it'll be a coalition coalition of the

unwilling you know like these companies

that don't want to waste their time and

resour resources will be forced to waste

their time and resources on this.

>> So they're going to wait their Okay, so

in my opinion, they're going to waste

their resources. It's definitely going

to happen. I think that's the reason why

you still need Leboutan. You still have

to have a guy who cares. Like guys, I've

learned a lot of things while working

here at semi analysis, but number one

thing I've learned is that if everyone

is like, "This is a great idea." And no

one is going to do it, it's not going to

get it's not going to get done. Okay?

You can have every single person at semi

analysis work an idea together and it

will not finish. The article will not be

done. One dude has to be like I am

sitting down on this puppy and we are

taking this article all the way. Okay,

that is the Bhutan. Now your 35 helpful

contributors that is the hyperscalers

the semiconductor companies and the

semicap companies. That's just how it is

man. Like someone has to have the will

to see it through. And Lip Bhutan

weirdly enough really wants the job. So

freaking give it to him. It's that or

Pat Gellzinger, Jesus boy, who like I'm

honestly I the relig religiousness is

really important. I'm like I'm hoping

LBT is really religious. I really you

need some God in in this. Um and you

have to like see it all the way through.

And then Yeah. And then also LBT is like

really realistic. I thought Okay, so

people really hated the Intel earnings

call. I actually kind of liked it. Um

because just just put some context on

it. Uh earnings was good. They guided

pretty decent, but in the guide they

were like, "Yeah, that's all tariff.

Pull forward, baby." Um I don't know

what Q4 looks like. Uh but that really

doesn't matter. So, uh let's continue

going forward. So, stock dumped on it,

blah blah blah blah blah. But I kind of

like that there's no more like I don't

think there's any expectations

management anymore. They're like, "Yeah,

totally throw that one out of the book."

Who gives a [ __ ] dude? this thing is

like a you know an 8090 billion call

option work you know running out of time

each and every day who cares about

what's next quarter's earnings like it

really doesn't matter um and I thought

the guide and being very like they're

pretty transparent on the on the and

then you go to AMD's and they're like oh

CPU's up big time and yeah it might be

non-seasonal but we you know we really

see a great refresh cycle just beginning

Windows 10 or Windows 10 end of life and

I'm like dude that's tariff pull forward

you guys delusional. You're like

messaging and managing and Liputan's

like it's going to be bad.

>> What do you think about him saying uh

you know we might not make it to 14A

>> that

>> which is uh which dude that was fire

tweet by your boy. I fastest finger in

the west. Actually second fastest. I

stole that from a friend. Good guy. Uh

if he if he hears this because he

probably listens to it. Appreciate you.

Um, I think that that is a posturing

mechanic. In the same way that Trump

needs to say really mean things to LBT

to keep him on his toes in the in the in

the hardcore job, you probably also need

um you probably also need Lip Boutan to

be like, I will kill this child if you

don't give me money. Okay, that's like

that's

like gun to the head. It's like 14A.

I almost want to kill that child. 10A,

no chance.

40, 50, maybe even $60 billion. Like you

need a you need some oomph in it. Um

pretty shocking, I think. But I think it

is a posturing mechanic

>> and it's smart.

>> Yeah,

>> smart.

>> That's that's what I would do.

Art of the deal. So, okay. Is that all

the topics today? This is a long

podcast.

>> Wait. Yeah, I was going to say we got

Weey here. We need some hot We need some

hot Weey takes, dude. We's like my

secret. He secretly gives me takes,

dude. I was all doomed up on Weey We and

I's DMs back in the tariff days.

>> What even is a tariff these days? What

even is a tariff? Who knows?

>> They keep getting delayed.

>> Oh, we can talk about 232.

>> 232 for Wait,

>> uh tariffs

>> 100%.

>> Yeah, the 100%.

>> Oh, they're not real. They're doing that

is that thing.

>> I I forgot that we joined and then

instantly completely stopped reading

news or semi or earnings or anything and

it's still just me in the [ __ ]

trenches, dude. And I'm like, god dang

it, dude. You got to keep up with this

[ __ ] man. Keep up. Um, you're doing

that.

>> You don't understand.

>> Yeah. Uh, no.

>> I live doesn't read news.

>> No, no, no. I I read news. I look I've

my mindset has changed since joining

semi analysis. I just realized like

we're in like the pink pony club

economy, you know? Shit's just going to

work out. You don't have

>> You're not worry about it.

>> What

some reference I don't

>> He's not allowed to sing Pink Pony Club

on this [ __ ] podcast. I will stop. I

will draw a line right here. This man is

>> Chappelle has got her back. All right,

everyone needs to chill out and just let

this hot girl summer economy just, you

know, keep going.

>> Yeah. an AIDC for every man, woman, and

child.

>> AIDC for everyone.

>> Dude, it's so it's actually kind of

genius because it's so capital intense.

We are actually investmentled economy

now.

>> Yeah. This this is the

>> this this I mean we we are in the

greatest experiment ever and I'm just

like I'm just here with Doug reporting

on the experiment results. You know,

>> sometimes they're good, sometimes

they're GPT5. You know, it's it's it's

>> but you know,

>> great inside by you way. I really I

really enjoyed that.

>> We're just along for the ride and like I

don't know. I think it's a great place

to be.

>> No, it definitely

>> going to say something else

>> with me.

>> Yeah. Did you say something Doug said

something about the 232 [ __ ]

>> Oh, tariffs.

>> Oh, tariffs. Yeah, I guess I'm the only

one who's like tariff pill. like always

information pled which is a feature I

guess uh after all this time. So,

they're supposed to be doing 100%

tariffs on semiconductors, but this does

not apply to anyone who is investing in

the American economy. Um, what that

means, what that list is, no one knows.

But that means it doesn't apply to

Apple. That doesn't apply to most of the

hyperscalers, that doesn't apply to

Nvidia. That pro I don't know what AMD

going to apply to.

>> Who's who's going to be dumb enough not

to figure out a way to get Trump a

golden

>> random small companies?

small companies that can't afford to

invest like random Taiwanese

chip coc.

>> Yeah. UMC.

>> Yeah. Or like or like fabulous small fab

companies like the ones that come out to

you know cuz you know I'm a shitco

enjoyer like the silicon labs of the

world the like you know what's max

linear like those guys that are like you

know they make about a hundred million

dollars of revenue. So they make 30

whatever we can even say bigger $100

million of revenue 20 $30 million of of

EBIT. They don't have any money left

over to invest massively. So what they

can do is they can probably I mean and I

think this is where the shifting all the

orders comes from. Now Intel has a long

tale of customers potentially or TSMC

Arizona. So I mean that that's kind of

the hard part I think. Um

yeah, everyone, every big semiconductor

company has already announced an

investment. Every big hyperscaler has

announced an investment. All the little

guys are kind of the ones caught in the

crossfire.

>> But you can change that with a little

bit of goldplated something.

>> There's a lot of things to goldplate in

that White House in the new convention

ballroom.

We got rid of the rose garden for a lot

of donation opportunities. And honestly,

it's it's surprisingly okay, I this is

in a group chat outside of there was a

really good conversation I had where

it's like some of the things that are

happening for for this is like

surprisingly cheap, you know, like I

would argue it's a fiduciary duty for

you to grift a little harder. Like some

of these donations we're talking about

is like $5 million,

you know, $5 million in a super PAC gets

you uh Yeah, sorry. Trump delayed a

Biden era rule that sought to prevent

abusive Medicare pra uh pricing

practices from private companies after

Trump super PAC MAGA Inc. received a $5

million donation from one such such

company that chose uh that that stood to

lose billions out in earnings due to the

rule. Like guys, grift harder. Like it's

for sale. I I I think that that's

something I keep thinking about. Um I

mean I I know it but I really just don't

think anyone appreciates how commercial

the president is. We're ready for

America Inc., baby. We're open for

business. 15% rev share.

>> Yeah, he he probably would have taken

five or 10.

>> No, I think he he he's like, you know,

hedge fun manager makes

>> the tariff is the tariff baseline,

right? So, that's a good it's a good

number for him.

>> That's true. That's true.

>> I'm trying to think what number I would

take. Probably like 50 or 75% and the

money going to funding BIS or buying

Taiwan weapons.

>> Are you sure? I hear US A has left the

>> US could use some help.

>> Drones are drones or malaria vaccines.

This is a tough uh

>> it's almost impossible choice, dude.

Almost impossible choice. It's drones.

Um drones dropping malaria vaccines onto

people in Africa.

>> All right. Involuntary. This is how how

we're going to catch up to all the

nonvaccine taking people is just have

drones. I feel like there's

>> conditions into them.

>> Yeah, just shoot them with drone vax

bullets. Um I feel like there's

>> I feel like there's a there's an incel

joke to be there like involuntarily

vaccinated versus voluntary vaccinated.

I don't know. I don't know. I'm just

workshopping this out loud. you know,

Invax, Volvax, I don't know, but we can

we can

>> US you're free to not be vaccinated in

Africa under the under the firepower of

our drones. You must be vaccinated.

>> 100% maintenance. Um,

>> yeah.

>> Doug, give us your uh what's your what's

your uh Hawaii take?

>> Where are we in Hawaii?

>> Hawaii. Okay, so actually was talking

about this with someone. Everyone's

like, "Oh, it's overrated." It really

isn't. It's pretty nice. Hawaii is

great, dude. The the island en uh

radiates good energy. It's pretty chill.

I wish I could live there forever, but

then I don't think I would actually be

down to live there. It's like also

pretty sparse in some ways. Um I read

half of Scholes of Time. I'm still

working on the other half. Um it's good,

right? I got Yeah, it's pretty good. I

caught up on a lot of reading. I've been

really reading the heck out of and this

is, you know, this is two Jordan

Schneider book recommendations on the

podcast. Uh, Scholes of Time, History of

Hawaii. Pretty good. Uh, and then, uh,

Search for Modern China. Is that

correct?

>> Yeah,

>> that is it. Dude, it's such a good book.

>> That's a great book.

>> Uh, book.

>> I'm going to be honest with you. It

feels like a contender for number one

this year. At least top three. So,

>> I'm I'm 150 pages into the one you told

me to read.

>> I don't even know. It's a little too

like pop psyche for me,

>> but I the ideas are very powerful, but I

just don't I don't like the rhythm of

chapters where it's like here's an idea

and here's the study and here's the

other study

>> and here's the conclusion.

>> I think the thing that sucks is like so

he's a Harvard professor and he got like

SCP into naming it.

>> Say what the name of the book is.

>> It's uh um Secrets of Our Success. Uh

>> Joseph Hinrich.

>> What the Joseph Hinrich? Yeah, it's my f

it's like literally one of my favorite

books of all time. Um, but it sounds

really popsai. The way it's written is

very case. It's like here's an example.

Here is the meaning from the example.

Here's another example. Here's the

meaning from the example. So, it gets

repetitive. Um,

>> for dumb people.

>> Yeah.

Like me.

>> This book looks really good. I should

read it. the book. It is interesting and

and there there are a handful of the

anecdotes which he raised which are

sticking with me like the

>> like how all the

>> just like when the the marooned white

people always

>> all die. only marooned people who ever

the only marooned white people who

survive in like the 16th and 17th

>> century are the ones that like have the

Inuit take pity on them because like you

know we're the same human beings but

like they have the quote unquote

technology to know to like how to hunt

the seal and how to cure the thing so

you can eat it without it killing you

and that's like 10,000 years of

experimentation to get you to that place

>> I got okay my contrarian take is um

maybe God isn't n't so bad. That's like

literally that's actually my takeaway

from the book

just because it's uh the oldest. So So

okay, now we're going into a straight

up, you know, if you're listening for

semiconductors and AI, you probably

should stop right here. Um so a lot of

things that I think about the book. So

there's a lot of really good anecdotes.

I like the antidotes because they

actually have stuck with me over time.

Um it's like reading the Sapiens book

except for I don't remember anything

about the Sapiens book. Not a single

thing except for Dumbar's number. I

think that's the only thing I remember

from the entirety of the book. Um, but

there's like a lot of little things that

kind of make sense that like, okay, so

over a long period of time, essentially

the the thesis of the book is that

culture and our shared culture and norms

um like passes on community knowledge

that uh compounds to where we are today.

A really good example is like spicy

food. You're like, "Dude, why would

spicy food matter at all in culture?"

Well, in, you know, in the equator

region, it's slightly antimicrobial and

that compounds over your entire life. So

that if you're taught a learned

preference to prefer spicy food, you're

going to live longer. And so that that

trait gets passed down and kind of

taught. And so it gives all these

ridiculous examples of all these little

things. It talks about the norms and

some of some of the ways that even get

reinforced within our culture to pass

these things down. And so it kind of

makes you ask and qu question like a lot

of things that have um you know a lot of

the ancient wisdom so so you say is like

there's a lot of deep powerful truths in

it and the one that I keep wondering

about and the one that like a lot of us

fight within a modern secular society is

like religion. uh God is dead, we've

killed it, blah blah blah blah blah. But

dang, did religion really serve a

powerful community bonding? Like it it's

a really really good way to to organize

a lot of people without like a you know,

it's like almost like a currency for

trust in a very large group of people

without like a no one can audit it

either. It's it's it there's a lot of

really powerful stuff um that I think

makes you feel happier and works really

well for the human brain. It's a really

successful cultural pattern that has

worked for since since the time

beginning of time in recorded human

history. Um, and you know, like a happy

Mormon in Utah who's soaking seems to

have more meaning than the average

extremely bombed out Gen Z Gooner. you

know, um these are two ends of the

spectrum, but like there's a lot of uh

there's a lot of meaning and community

purpose that might be found through

religion that I think we might have

thrown away that cultural pattern too

soon. And the problem is if you don't

believe in it, I I don't believe in it

personally. It's pretty hard for you to

like, you know, fake when you know the

the Wizard of Oz isn't real, but you

look at the whole spectacle and you're

like, dang, that was a really powerful

tool that we kind of threw away. And

that's one of my my takeaways.

>> Doug, I've actually thought about this

like I feel like the society that the

capitalist society is probably one of

the worst position if you actually

believe in ASI is is on the come

like if you think about like how we've

replaced any kind of like fundamental

belief right I think you know US is bas

Yeah. we replace it with capitalism,

right? Like it's like the it's it's the

true religion. So much so that we deemed

like other economic systems as like the

enemy

>> unreligious like capital and freedom,

right? That's that might as well be the

Bible.

>> Yeah, it might as well be the Bible

here. But like it also could be the

worst system for

>> that that makes the Dude, that's

actually

>> that's a good brain way. That's a good

take. That's a good take. It's a great

take.

>> I earned my spot on this podcast. Thank

you guys. Thank you.

>> So, uh, so like for example, it's just

like like all this like like for

example, if there's ASI and you're like

a godloving society and ASI is just

chilling. It'd be like this this

probably works really well.

>> It's just another extension of your

belief, right? It's like, oh, we always

knew someone was guiding us. Now he has

a name Cortana or Siri, right?

>> No, no, but

>> Cortana, that's who you're choosing?

You're going to say Cortana? Oh my god,

brother. Um,

>> the god of productivity,

>> dude. Um, the Halo series is just an

offspin of the future, you know. Uh, no,

but but I think um, yeah, I I I don't

know. I I think it's a really really

powerful way to organize societies

and it works really well. Also,

coincidentally enough, he wrote a whole

another other book. This is a Joseph

Hinrich book as well called Weird.

It's another [ __ ]

jabated title. Weirdest people in the

world or some crap. But it talks about

how like Christianity specifically kind

of dismantled some of the family based

hierarchies that we used for trust

before. Uh the thought of you going to a

marketplace and just being like, "Yeah,

I'll buy it from Joe over here instead

of John over there with no information

is a very Christian idea versus if you

Christianity didn't exist and it's like

all family based relig uh family based

stuff. You're like buying it from your

third cousin always because you know you

could trust them." Um, and then my

favorite stat from that book, which

always horrifies people, but like 90% of

all marriages in the 20th century were

cousin marriages. Um, not cousin

marriages, is a relatively new

invention. Uh, past and past and, uh,

enforced through Christianity. So,

that's also a good book if you like it.

But I I think Secrets of Our Success is

my favorite book. And then I I I do like

that one as well.

>> Book review hour with

>> I just Search for modern China book.

I'll listen to it tonight.

>> It's the best book. It's truly the It's

It's the one China book to rule.

>> Jonathan Spence. Rest in peace. Rest in

power.

>> That's from China Talk. A US A funded

organization also funded by Sorry, dude.

I I I

>> the Jonathan if if I could be funded by

the Jonathan Spence uh foundation.

>> He's like he's like, "Sign me up. China

Talk is open for is open for

sponsorship. Um, no all the way through.

>> It's a really good book. I really

>> got to embrace the grift.

>> Yeah, dude. It's for sale.

>> Don't forget.

>> Don't forget. Okay. Well, uh, let's end

this Transistor Radio with another

episode without Dylan. He at least tried

to show

>> never came back. It's pretty hilarious.

He texted he texted the group chat and

he said my internet is is okay we'll do

a direct quote coming my internet is

[ __ ] question mark WTF to which

Jonathan respond

WTF and you're like okay cool you going

to do something about that no so that's

uh our episode of transistor radio

I got a text from let's send it by

>> yeah bye bye