Endgame with Gita Wirjawan - Mochamad Asri: Talenta Indonesia Jadi Lead Architect di NVIDIA

Episode Date: October 6, 2026

Membahas AI secara lebar dan dalam. Mulai dari hal yang sebabkan AI haus energi, peran perkembangan teknologi cip dalam pengembangan AI, hingga komparasi open-source dan close-source yang sejatinya bisa berkembang berdampingan.Topik selengkapnya:0:00:00 - Intro0:01:55 - Keluarga terdampak Krismon 980:04:40 - Perjalanan pendidikan: Tokyo dan Texas0:23:31 - Kenapa AI punya problem energi0:29:51 - Edge AI0:38:48 - Moore’s Law0:40:55 - Mengapa open-source lebih baik untuk bersama0:51:06 - Efisiensi di bidang energi juga bisa bantu AI0:54:43 - Kompleksitas AI0:59:55 - AI dan cip 10 tahun ke depan1:04:12 - Cara tingkatkan minat STEM1:09:02 - Kualitas guru pengaruhi kualitas pendidikan1:22:31 - Mengapa ‘brain drain’ salah kaprah#Endgame #GitaWirjawan #NVIDIA------------------Buku yang saya tulis, ‘What It Takes: Southeast Asia’, sekarang sudah tersedia dalam Bahasa Indonesia dan Inggris. Dapatkan di sini:https://books.endgame.idSudah baca? Tinggalkan review-mu di sini:Goodreads------------------Episode lain yang mungkin Anda sukai:When Businessman, Ex-Policymaker, and Engineer Collide to Talk About AINiki Luhur: How AI-Enabled Organized Crimes & Systemic Scams WorkDr. Yasantha: AI vs AGI & Homo Sapiens’ Next Chapter

When I was to Austin, there was a culture gap, actually, but... ...secarist, yeah, pa. Deru Kuiwa uttariru. If there's a paku that's been onjol, it's in palu, so, so, we're gonna be able, because we're...

...they're very, Fereux, extreme. Individual. So about you? Yeah. What do you think? Meeting, that, in Japan, we're just, yeah, take turns, though,

for, for, to say, please, Let's say, can, in your mind you what you're saying? Wow. Free for all, ma'am. Yeah. Who who want to mong?

I'm just going to mhm, yeah, right? In the system of computer, compute is very cheap. It's a very cheap. The more than consuming energy 200-kali from compute,

that's the data, why this is relevant? It's consequence if we have if we have Who's who who want to use the camera smart glasses if it lasts only for one minute? Hello,

friends, I'm very much to tell you know that book my book It Takes, Southeast Asia, is already in the basis Indonesia and language English. This book this can be developed through books.endgame.

Dot and ID and the toco book the best to the show. Hello, So, my friends, we're to come on Muhammad Asri as a lead architect in Vindia.

Asri, thank you. Can't come. I'm, P, were there to discuss with you with Bapa. You're talking from the two of two of the

from Succa Bumi and Acheh. Yeah, in Jakarta, and then The rest of your life story. Yeah, yeah. So, I'm my, I'm from

Sukabumi, Pak. Yeah, and my, My father, my father, from Acheh, and they're in Jakarta, and they're living in

long, and I'm going to the other than from Jakarta. Now, um, the, he was, the work for

and Ibu, before Chrisman, 1998. After Chris Mon, she was in P. She was in P. And two years after that,

BAPA Meningal. So, So it was very, very hard for our family, because Ibu, because Ibu, not have put up-appenance, yeah, and yeah, we're figuring out life, yeah, right, so for me personally, it was very hard too, because I,

not have a father figure, like. And I really... Umur, how old are you? 12 years? Yeah, 12-year, yeah. And I always remember, I always enjoy doing in per-matematica with my dad

because he was very good with numbers, yeah, and very good being present. So, when I'm working P.R. It's, I'm the person that, I'm the most important in the world, because not only I enjoy solving the problem, but also, with present in the present. So, so, that, we, as a family, trying to figure out how to survive, that, ma. And, Ibu, you have to try some billion.

I think I'm going to do you know, I'm often times, I'm going to 1st p.ggy, for, uh, to make you to do you, to do you want to

to dole to the time to help, pay payah school, and my and k'a'am and

we, and we are, back to come in time 5, 20, to run to

from from the because if if you're there matcett yeah, it's a

matcett, yeah, so, that's that's sometimes I'm some of something,

what, so be a big big, so I can disappoint her. Yeah. SMA

LAPA D. I'm 8, Jakarta, sir, sir, sir, BANjur,

sir, but, BANjur, if we're Bancur, if we're If we're banquered, we're like,

Libur, right? Yeah, Samala, Panjakarta. Then, then we're doing to get

to do you in the world or in the country, who's what in the world so, so I'm

so much that I'm saying, no, no, there, no, no,

yeah, part, part, the time, that's the same way, just, but,

just, because, the partasasas like I'm the, human time, even for

college, it's something that's a lot as far as financial. I'm, I've said said, I remember that,

when I'm, maybe, we're going to get to be doing, we're not, we can't support you

on financial, because, you know, we're because, you know, but. So,

after that, I, I, mindset, I have, I have to support,

this, for this, school, like, if I'm, that's a person,

I'm going to play, but I'm always playing ball. After I'm listening to, I'm going to slow to be able to

it, sir, it's more more than, for, yeah, serious, I have to find B.

B.S. And, you know, the one of BASISU that I came to is a BASISU PEMPANTA

Jepan, name-Namani BAKUSO, Department of Education, and They're them and I'm

small in SMA to learn in Japan with stipend and tuition fee that's

so much so much but the sure you need to lose test that acceptance rate

that is very right right I'm I'm okay let we

try this I'm I'm not my murid the high school

because why you know what Olympia do I'm playing ball just,

ma. Because it's a hobby, but I say, well, if there's a case, why not? Let's try it.

If you know, if we're, if we're, we're not, we're trying a way, we're we're trying. We're trying.

We're able, I'm going, I'm going, I'm still remember when there's the telephone from the Dubas

that, um, he, you're, you're to be able to continue to talk to

the day bachisisiscus of Japan. I'm I still not can't back, but I'm

there's a bell the phone, man, line, that, can, I'm, I'm,

I'm sorry, to come to back, back. I'm, we're we're we're not, we're getting,

the, B.S. S.A. Japan, her, that's, that's, the number

of you, try, try, did double-check. So, so, so skeptical, that,

like, yeah, because, yeah, yeah, yeah, yeah, yeah,

yeah, yeah, yeah, we're, I mean there's also, there also, Olympia, mathematics, physical, and much,

much, but sometimes miracles do happen, yeah, right? And, and then, to Japan, yeah, I mean,

got, doors after doors, open, and that I'm not, did not. Ample about, Kooliator? Computer Science,

ma'a, yeah, yeah, Jursan Technic Computer. Yeah. Yeah. And then, in Basa Japanese or?

It was in BASA Japan, yeah. Yeah. Yeah. So, so, program

is we dis-school in the language before one year. Before we we're

in campus campus umum in Japan. Because when we're there's curriculum English

English, but now now, now, I don't know, but that was in

different, so, so in one year, we're we're we're to be

under can't be can't can't be, yeah, man, but,

learning Japanese is one thing, yeah, but With the language Japanese. That's another thing,

Pa. Yeah, I'm still I'm thinking that's, I'm still every day, at least 10 kanji,

that we have to have in and know how the way analysis of what you know,

that's like caching capasan mannari, we're and, oh, this,

why, we have have been haphalin can't yeah, yeah, frustations

sometimes, despair too, yeah. But, like, like they're saying,

what doesn't kill you, makes you stronger. Not not always, but you know, oftentimes, you know, yeah.

So, yeah, it's all right now be able to learn in Japan even though even it's not too, to the

Tokyo Institute of Technology. That, yeah, expectations we have to have one level in cognitive understanding in Japanese language,

so, I can't take class as far as than the other of Japan, so, so, that's in

language in the language, what you're what you're what you're on what you different

than what you see that in Indonesia or in America? Yeah, well, first, at I'm

that I'm struggling, ma, so, there's, there's that, Sensor or dosain that, you can, Nulis,

Paget BASA in whiteboard, yeah, right, that, it's quick, speed, yeah,

right. And, kind of, because we don't level on the young, we're still

struggle, bach canjiny, but, that, that, that, that, that,

then, then, then, there, there, CHAPT or Google Lens, but, what,

it, um, the, um, , the, um, paper dictionary,

so, so we have so we're so, we're up a parhalamat until

get to can't it's time for the time, but to be the book camus,

until when we got to get used because the writing back, because the

document to the next chapter. Yeah. So, wow. That's, that's, wow,

that's, yeah, yeah. It's really better. You can't make sure. Yeah,

yeah. It's, this is, what, yeah, I'm glad I did it, but I'm glad I did it,

you know, yeah, I'm glad I'm doing to my friends people, I'm saying, yeah,

I'm not I can't I'm not I'm gonna not I'm gonna make sure, yeah, come to come to come,

so, yeah, it's important, but I'm part of the challenging times like this. This,

this, yeah, this I'm after this I'm overal about about how about how

the country and bank that to do not even in the Japan, too, in a lot of decades, it's,

it's, doing talenta indigenous, yeah. Yeah. So, industrialization Japan,

to, Prascha, in the world, that, upon talent local. Conan,

80-an-p-pents to not-the-o- -up-the-the- -up-the-the- Now that, can, it's been a different from, or with the other other than

other than talent from the world. America, he's be gantung from the world. Yeah. Australia, be gantung from out of the world. Singapore, it's from out of it. Now, that, you know, intensities

in Japan, that? Yeah, ma'am. I, uh, I'm I'm, I'm, in the United States of Japan is very, intense, in the art of

tequoon and really, high quality. Begin'y, they're not look for a shortcut. If something needs to be very high quality, then, better,

let's try a way that to go to without shortcut can't right-kir. So, dedication to make

work for good-becky-beck coming from Indonesia, where we're good enough is good enough, kind of kind of, kind of, doesn't work. In Japan,

in Japan, the highest quality, that is the abhangered from their own, their own, so, the culture is that's been built like that. Maka,

when, culture has been made up to the best, that's, you know, but the standard is, it's up,

and, I think, that's really, really, really, because of the key. Because Japan, can't really, can make sure can

make sure can technology that's the leading electronic companies. Toshiba, used to be a producer memory, back then. And it's really,

quality assurance that's it's just to be chagued. You can argue, there's plus and minus it, yeah, sometimes in, when we're overlapped with business, we have

to know trade-offing to what, not have to the that's the most. the cost analysis not not much, yeah, maybe we need to be in a compromise, yeah, but,

but, it's just to be a culture and buddaya, it's just to be a culture Japan that's really, really amazing, that's... What's what makes them make, that's,

the good, the best, what can be sure, what can be made the culture to, from room, to can't,

to, institutions social, all-in-as- people, that's a good question. If, If you know, to go about with the

it's like there are two, yeah, bah, yeah, yeah, that's that's the highest quality,

yeah, and the second, is like there responsibility, bah, that you that's,

that you, the, that's, there's responsibility, yeah, can, maybe, if,

simple, it's like, film Spider-Man, man, with great power comes great responsibility, yeah,

so, So, so, there a rasa about, this has a responsibility with power that's high quality. So, we have to deliver the highest quality.

It's in the per-baga tatarant communities and people, always, want public servant, want the

kawai-swast, they want to deliver the best quality. Even, even if, if, if, if we're what we're

sacksicae in China or Korea, they've got a bit of their from the side of the etica, and stuff, yeah, yeah, then,

then what's the other, why not in Japan? Yeah, so, there, sequence of, why, I'm,

I'm, at the time, at the time, I, um, to class

the class tamu, or the dosan tamu. Now, the docent tamu this is from America. It's,

he was manpan big IBM, who was of course, one of the best companies

in the world, if not the best. Yeah. And now, so, he was

professor at University of Michigan. And he told how Silicon Valley, the ecosystem and the

best engineers in the world. I said I'm not really, I'm still if it's really cool, if you're to Silicon Valley. Because of all technology that's impactful in the

world, that's, it's in the Silicon Valley, in internet, computer, PC, and that's much, much. And then, it's been, because it's like, it's good,

if it, if it's to America, that, and, and, yeah, plen-plank, you, get, to, you, try,

yeah, try, we're trying how how much how much to America. I'm not

know what BASIsa like what before before before. And I'm not crazy,

rich Asian. Maybe just crazy. Well, now you're rich. Well, that's just really. That's

that's a idea. We're, yeah, I'm trying can't be able to come down to America. But,

can it's still, but, but, from from the time to the year-year-upat. From there, to inspire inspiration,

and, um, I'm, the year, I'm going to planing strategy. And, one of the strategy that,

to get Biasisua to America, um, we have to be able to do research, especially for PhD, yeah, right? So, yeah, so,

when I was, it was, I'd publication in top conferences or journal. So, I'm

my case I'm going to say, I'm going to say, I'm saying, I'm want to be a topic

that's potential to be potential to do you, and it, if it, if it's, it's a

publication that, I'm, I'm not peduli, that's what, I'm, I'm,

I'm, I'm, I'm, it's, it's, it's, yeah, he suggested this topic

topic that's potential until we're working when we're doing the topic topic is optimization microprocessor

Pachep and that we're actually we submit there's conference called hot chips at Stanford

that Augustus this is Augustus that's primarily industry for to showcase chip of their

just there's there's there posterns al-a-l-Dol-Las And it's the paper we accept in there, yeah, it's a good, sir.

And the docent I'm also agate, wow, it's really, it's really, it's also, yeah? Yeah, yeah. Yeah, it's, it's a model, that, oh, yeah, it's a good start, yeah, because, first, it's good conference. The second, the other,

the other people in the bidan I know conference that, so, so, Professor Brasser, that, oh, yeah, you've been published there. And then, I'll continue S2 in Japan and M.S.2 that's all full throttle

that has to add more than publication. So, so, chance to get BASXVACUS is more than better.

Yeah, then, you know, but, but, you know, if, can America, there's a

GRI, TOFEL, it also also, uh, do, uh, I'm,

uh, uh, ap other than, um, what, costa-cata, causea-cata, that,

yeah, out, out, out, all of all, still to want to muntah,

but, but after with the other than the problem that

and al-hmm about paper did get to in the conference and journal,

I think we can submit this that we can we can we can't try at least apply

to to be some universities in America and al-hmm it's not,

it's with publication illmiya it's it's so much it's

there's and plus biasisua because if if you're just not by visa,

I'm not able back, yeah, yeah, shuckor, yeah, the way,

yeah, ma, why, why, Austin, man, man, man,

man, ma, ma, ma, ha, in Austin, at the when it was

there, um, computer architecture track for chip design, pa. That was really, professor professor's professor. And connect the industry

also, too, okay, that's, and I, and I, was very impressed with, what, yeah, faculty, where,

many of them who are manan, even, uh, the idea of so, so, so,

and, uh, and, yeah, and, I'm, I've seen, I'm, yeah,

and, um, and, , From, there are other options. That's the weather or Indonesia. Sometimes, not warm just hot.

Hot, yeah, that's right. Summer, it's a matter. Yeah, right. Yeah, it's right. Yeah, it's got to say, ma'am. And, yeah, I had one of the best times in my life. How many times in my life? I'm in the PhD, this.

Yeah, six years. And this, uh... I've got to get, don't, meadow in Austin like how much. Oh, I know. Yeah. Yeah. Wow, I know, ba'am.

The best barbecue in man. The best taco, where Indonesian food truck where? I know, you know, but, yeah.

But, when I was to Austin that, there, there, there, there,

yeah, I'm, yeah, I'm, yeah, I'm, from the system

building, Japan, man, and America, it's a different story, right, but,

first, in, philosophies, yeah, yeah, yeah, Japan, it,

that's kind society really bener harmonious society, so, so, so,

group consensus and harmony is number one. There's a pre-bhasa in Japan, Deru Kuiwa Uttareru, which means

a nail that stands up will get hammered down. If there paku that menonjol, it can be palu so,

so that's right. Because group consensus is important, because we are collective society.

we're big because we're together between. They're extreme.

Individuals. So about you? Yeah. What do you think? Individualistic. Yeah, man.

So, so, I'm not quite quite kind of again, but, you know, when back,

you know, wow, like meeting, that, in Japan that's we're used, yeah,

take turns, though, for, but, you'll say, like, should, say,

and, and, you're, like, wow, free for all, pa. Yeah.

Who's who who want to say, right? And that's a lot of shock, yeah,

right, but first when I was I'm just head DA, you know, in teaching assistant,

there's one introduction to computing, yeah, right, yeah, that's, not,

not, it's not kept time time, take turns, yeah, if all, if all,

if you, if you, if you, if you, speak up, computer sense, that, did,

RML, yeah, did what, yeah, did what, yeah, , INS, sir.

INS. Yeah, INS. Now, now, now, what's the, now, what, it's been up. Oh, it's been

to be born. Yeah, yeah, yeah. It's been able to go to be. Yeah, it's in. Javan, I'm in fact,

200,000. Wow. So, a jute a, yeah, a a year, yeah, Tesla,

to, Dell, to, Samsung. Yeah. You're favorite to where, if you're if you're

in Austin. Sixth Street. Of course. Yes, of course. Noton, I don't know what I'd like to be happy, happy. Yeah, yeah, yeah. Thesis of you're what?

Tessis your own. Tessis I was, how we can orchestrasy or demurement or pemminehata, so that it's energy efficient, Pa. Probably statement is like this, in the

in the computer, Now, it's compute is very cheap. Computations menabhack, mengurangy, or percolonial,

it's very cheap. The costum the energy 200-kali from compute, is the demindh from data,

from data from from the data that from, dipindah to unit computation. It's 200

times need to need energy more, yeah, right, yeah, this,

this, the end-ug- -un-un-un-in-y-ion-y, if if you're like, if we're not orchestrasy as smart, we're wasted energy, because this is relevant? Yeah, consequence is if we have

battery-powered device, you know, that we're not at all the time with consumption power, yeah, who want to make make a camera, if it lost only for one minute? It doesn't make sense, right? That's why I'm going to

make sense, how make system orchestration of the data different compute elements, so, orchestrasy data that is the the most optimal and energy efficient. That's too, ma'am.

Sorry, so, so, so, so, that's not as high as much as it's musty. It's a got a wrong parker,

in context of the energy. This is because transportation data, that's just true that's proportionally

that's more more. They're movement bottleneck. So, analogies like this, pa. Yeah, right? This,

this, this, can't let's process a thousand people per day, or not that, in process

administration, or anything, yeah, who are who want to get to get to get to

get to run, Sudermant Tamrin, or anything. Suddermantamri, can't allow, just to process 100

people per day. Yeah, Yeah, right? It doesn't make sense, because there catimpangangant on the speed

that's the ability and the speed and the speed that we have to move endh to move in this to get to be

back. Now, taxi's my message I'm inside. This, if if miscellalak

in the demindhanda data, we have we're just focus to so on that.

Yeah. So, I'm going to take, I mean, Today we have a lot of about how energy. It's very important for the

important of the penitain of the narrations narration AI. Yeah, yeah. Yeah. Yeah. Mugnaitiga, cloud deep sea, chat, GPT, that's conan 10 to 50

times, dibandinging on Google search the penguanaerginess. It's more than is the the data, so, not because

the capacity compute is it's because in general, it's the data movement is

much more much much more because because data center there is there uh...

uh... uh... so, uh, data model it is the data model it can't because

back in just, Just that model, the capacity not main, yeah, can, trillion parameters. If at a byte, it's

not even contained in one GPU, but it's to redistribusically to the other than it's redistribusically in the

cluster, yeah, to pinnhine to compute it is puth time and energy, yeah, so,

that's what, the bottleneck now, now, when, uh, when we're, um,

that AI consume energy, yeah, it's from, there are three sides, first, capacity. Because of capacity, because one GPU that's not can contain model that's so that bester to give up to GPU on the line. With to give it to GPU line to comeputation, it's

has gathered and scatter, can't, after that, gather, data, so, that it's, it's, it's, it's, it's, it's, it's, it's, you, can, part. The second, the problem bandwidth, yeah, can. Bandwidth is, uh, what, what's,

what we can't be able to pumpa, in one minute, that, if we're talking, within GPU, it's, it's, it's, it's been very bandwidth. It's, but it's different, it's the it's better, that, is it's the way, it's it. If we're going to go, yeah, after we go to,

go to, go. Jalantikos, yeah, can't, yeah, can't, yeah, that's, right. So, that, it's problem, that now now, the

company try to solve, like, how much there. There's there.

Photonics, interconnect, yeah, right? Um, it's all the way, uh, a way

we're doing optimisation data movement bottleneck this. The other. The third, is latency.

Why? latency matters? Latency is the relationship, back. If you're type chat GPT

or, or, whatever provider to ask something, if response that

only that's a week and practical, right? Yeah, right?

Ligency is important because as far user, we have used to look to see how

we're back back-pugate. Now, the AI, machine learning, back-propagation or

perceptron neural network, that concept is there, from 50-an. Not a concept

back-maticus. Why now can't take-off? Because we have hardware and system that capable

to deliver latency. That's which is designed by the company

that. Yeah. If the back in 50s, if the respond that's back in the month then,

who's the one way who will make up? Nourique. Yeah. Now, in milliseconds,

microsecond, the respond that's coming. I want to ask, this, the under

electrification in the people, the country, this matter of Indonesia. Seren I'm

saying I'm Semi-R-1-Hour-P per capita. Yeah. Tionk, America, Singapore, Japan, that's 10,000 kilowatt hour per capita. It's maybe a batas for the country to be able to beamongue, narracy AI. How, for the countries that bebeamonged in the

world that electrification is, 1,000 to be pungu, not the people, not the Ration, because the riskuagueness, because we can't make electrification to 10,000 in a second, this is a big amount. If we're just being

5,000 megawatt per-t per-a-a-half, we have to make in the air to be 1,000 gigawatt, and to build 600-gawatt to build 600,000 megawatt with capacity of bangun 5,000 megawatt, we need to be able to 120 years. Right, right.

For the people, how to men's enderhanacan narration AI, so, that's in the time to be able to beman not penantone? Yeah, this is a very good question.

So, if we're going to from layer AI, that's, if you can, if you're going to, there's five layers, but the layer that the bottom that's the energy, like I said, as well I said.

Because AII is run it's like electricity, man, if not want, not, if energy supply that's not quite, yeah, yeah, you can't

so, yeah, so, and, the capacity electricity, is, U.S. because they look

China, capacity production is much, it's more, so, how they can may

so, the data centers that can be more, right? Now, what's the question is how with the country that's

capacity that's not as a major maus it's a real challenge. I'm not going to sugar-recode it, but,

but, really real challenge. But we also have a little there's a there's a I, where we're called H-AI,

or efficient AI, some people people, model, where we can compress model, it's, but, and more energy-consuming. Why can't like that? Yeah, because trade-offs is like this.

If model big as open-a-I or anthropic, it can't do not do you make general intelligence, whatever can't, that's, that can. Now, if it's just like generalists. Yeah, can, many networks that have been put up so they can generalists. Age AI,

it will be specialist. They're targeted okay, let's say, for industry agriculture. Let's say, intelligence for

industry, fabric, that, that's not general intelligence that's a level model and a level,

model, so, so, benefit is, we don't need, we don't need consumption

electricity that's a lot of scale data center, but enough intelligence to get some productivity boost. Yeah, right? Now, that branch is now, takeoff, let me, slowly, and then, it's from,

the problem privacy, yeah, right, because if data center-based AI, may not want to, data proprietary, it's to be given to get to send to that. Now, community this is well,

if we want, established secure AI, we have run locally, which is where devices is not as powerful data center, yeah, can. But for run locally, automatically, we have been memackas model intelligence this, so he not consume the energy

that energy that's energy, because we don't have energy budget in devices, that's been a realignan, so, how it can be able I, in fact I, in a certain negatna gaugabe be agenpae for ketchup. Of course, that doesn't have to be either or.

Energy, if it can't, if it's, it's, but while waiting for this, because acceleration gulfs, not stimpal, yeah, we can play with the alternatives, like small language model, H.A.I. So, so, that, it's a. It's, yeah, it's a. Or, small language model,

it's not to becapar or it's or with a or with or with curumitance for transportation data, like the illustration

before yet, it's not that data is that's so, so if if,

that's a assumption that memorioning that, it's not the data center, because that particular user.

Okay. For data center, it can, serve how many 10,000 or million or million, manusia, so,

capacity has been really, better, right? If H.A. is a particular use case, yeah,

can, that's really, really, is the specific, the second, the user user data

center. From there, requirement for bandwidth and, and, the capacity that's

not as a center, pa. If, even, the, bandwidth and capacity more,

it can Also, HAI this is to be made, so, it's distributed system, talk to each other. Right, right? Yeah, can't. BUSAID. BUSA. BUSA.

B. And that's, there's, it's, it's, that's the agreement A. and H.A. Network A.

NATW. Okay, not, if we're together, win-win, that. But, can, it's, this,

this, concept, concept, of the media-smr, small modular. Yeah, but yeah, yeah, actors. In a way, it's very connected, yeah. Because, yeah, um, if, if it's hypothetical case just,

yeah, there data center in a disa, that, for the attention, that, for the, yeah, can, if we don't have data center that's a gigawatt, that,

yeah, maybe there's a data center that's a little in a city, that's, it's from where, Right, right, and SMR is a potential use case, can, Pat, to be able to be able to be sure that's being a 10 megawatt,

and then modularization, to 20, to 30, to 30. And that, that's, almost, past, the power of language model to age. Yeah, I see. So, yeah, yeah, yeah.

Not the typical LLM. Not the typical big language model. Yeah, yeah, yeah, yeah. So, it's similar, yeah, Pallogyna, if we look, if we look at 20 years or 15 years that year, that processor that we use,

chip that we use in our phone our that's using, that's used, for server computer. Wow. Yeah, ma'amologi is like that, sir.

What is, what, what, facilitation improvement that's dramaticous? If, like at the server, mainframe,

but now it's already in HP. But, other than design, but what's what's facilitation? Many factor, yeah, first, design, especially, yeah, architecture, how we come up with new architectures,

so that it's more energy efficient. That's pasty, yeah, can. But it's a gapungan from several things. Artitecto, the two, technology, yeah, can, more slow, yeah, so every time,

we can make a transistor more small, yeah, can. So, what, 20-tang-law is, to make impossible, to make up to the phone, now, it's possible, that's possible, right? That's been,

it's still, now, yeah, right, and it's 20, 30-tankal, maybe 100, 90, almost 200 nanometer,

yeah, so, progress it's very very, yeah, right, From the gapungan factors, factor, factor, that's made,

um, we can, um, muster on the server, as powerful, uh, power it, it's,

it's also powerful as as a server's chip 10, 20 years ago. So, combination that, yeah. With what you see what you

see, in Silicon Valley, yeah. More slow, is still, still, is it's still, be partown

for 20 time to the two years. F. That's a million-dollar question. question, but at least until

we're still still to some extent. There's there, there, uh, pernourning

uh, yeah, in a two two and a two and a time, but, um,

but, um, the way that's we're making we scale, as a small size transistor that it,

it's more investment in, um, equipment, equipment, to, um,

to make transistors as small that, so, so, per-kechle transistor is now cost functioning is exponential. It's not linear like that you like that, there's a-smell,

we have to make machine lithographiness, there, that's, that's challenging for morsel to continue. The second, the equation,

there's one part of the equation. There's other that's the denard scaling. Yeah. It's, in, in the,

Narccaleing that, if we're going to downsize the uproar transistor our, that we also will downsize power, proportional with the the current transistor we can. So, power budget is constant, that's that one of the one of the ones that is to the HP.

Now, now, with the transistor so be that's so much a freelance anymore. Because sub-threshold leakage that's the cost of it's been consequence what, the consequence is, if it's the heat that is

it's very hot and impractical, not, not much, DENR scaling is stopped, now more slow, with DEN-R scaling hand-in-hand is what we've been seeing the past 30 years,

now, if more slow, maju, but the DNR scaling has stopped, can not be good good again, this I'm going to try This I want to try to go un-buckus in context

in the country be able to beaubank, and in the world, this, can, negara, the other than the people that per-orang in, in $13,250 per-per-one, this is the $7 million. Yeah.

Which, the general south. The important of their, because the dais-belin it's is to be able to use energy and efficient. The kedgaratiness that is not capacity

computants that more but just true the of the data that is more this can beobati with the if they're

if they're if they're want to be the manrasy AI now. After the end we're hearing we're hearing

platform AI, especially in America, it's be gasser from open source to close source. This is in batas intuition to come with

the to getaulatant data. Yeah, yeah, yeah, yeah, yeah, yeah, to come

to overweigh the deolaten data. Yeah, yeah, yeah, yeah, to solosicin it to energy, to get the hearing from E.C.I. Then, the two, to dewlotan data. Mungit many of the platform to

be gerser from open source to close source. So, yeah, ma'am. So, there multiple perspectives that, I think I, valid, about this issue, yeah. If you're from the Jensen, he supports both, close and open. Two days he was published letter in public to to support open source.

X. Yeah. Yeah. Yeah, for example, for next. Yeah, yeah. So, yeah. So, yeah. So, yeah.

So, yeah. So, if we're, if we're going AI to make progress, we have give access to everyone, and, that, because, yeah, with everyone get exposed with AI, it's more rich ecosystem, yeah, pa. So,

so, um, anna, or penit, people who are people who are people, that's about,

with open model, yeah, can't, soing- collectively, we become better, that, um,

it's argument from open source, yeah, from, from closed source and the argument is, well, this is a bit

be bad if all the people to public. Because

if you know, there's people who can't people who people who for things

harmful that for cyber security attack, much, blah, so,

so, yeah, okay, there, there's different different, yeah,

but I think like this, but, it's either or, yeah, we've

we've seen the same play with Windows and Linux. One, closed-source, Windows, one, open source, Linux.

There are several cases where closed source, it's, yeah, there's market that, you know, Windows, yeah. But Linux, it's also, be coming,

that's open source, you can, be used in the area centers, and many, much, so, yeah, right, so, so,

don't, be matting one or the other other than, be able to bebeam because Linux,

open source, it will have been market the market itself. Researcher, researchers rely a lot on open source, pa, yeah, not only just in the other, even the country and major major maju, but especially in the country that's becoming. That we

we need open source, because it is it's a platform for optimisation and also commodification model that, so that it's fit with a cariffing or context local, and that, that's very important. So, it's very important. So, If I believe, yeah, two-douan if you can't move, but open-source, don't be looked as

second-class citizen. In my personal view, if we, as a global society, it's, um, ingin, uh, to be better, collectively we have to be better. Not just just

U.S., China, or the country that maju, but collectively, the countrymen of reembalch, also, the government-in-gand-in-dang-tangue. But, actually, open-source, too,

is more-muror. Yeah. From from the close source, in the batas intuition. I could be wrong about this. Yeah, so, yeah. So, it's not necessarily the case. The-that-is-your-case,

that's not a case. That's open-source, it's, is they're able data data and also recipe-in-na. Right, right? Pintu-inthunner-divok.

Yeah. So, if you want to move, gratis. Yes, yes. Yes, yes. From that angle, yes, from that angle. Yeah, from that angle. So, so,

So the person is skin because of the government, but to access data that is going to be a good-gaping, not there's gatekeeping, not use to pay toll. But if you're closed-source, it's a lot. Right, right. Intuition just.

Yeah, no, no, it makes sense, that's how it works, yeah. Because there's how it works, yeah. Because there's no, there's a barrier cost, so, open source it will, yeah, can't help case-case, where, yeah, like researchers to have been

become, for to make sure to make technology their and it's

almost impossible if you're almost impossible because of money and we're much and we're and we're

that wealth inequality exist, right now I'm want to take that if if

platform this this is more much close source this this is kind of

again in batas intuition that it's the same, it can beaqqas information.

Two, also, the other, this is how, this, to make make make that the data to be in each, each,

or, at each, at all of any, right, so, so far, the topic this,

there, and it's one of the big topics these days, right, I've ever heard UAE or Saudi Arabia, that's made deal with with some of America, chips in America,

not that it, AMD, NVIDIA, to make infrastructure in the country, but why it's important? Why, they're making it. It's important, that's, that's the way, that's what I said,

the lawlata, yeah, can, they want, all data data sensitive in the country, it's, the process in the domain of their,

not have to not to give to the country to make it's a motivation to make deal with the

subsurances to make deal with subs and the chips, so the country their country so that

data that don't need to be able to be that. That's the smangatement to be

there, yeah, can, anyone with data data-sensitive and much much make sure

have been local data center that, that data not can to be where

where that it's so it's non-negotiable about and I'm think

the time it's going to many other other other that

men mirror model like this Hong Hong Kong to

still still still still to to make for

to end technology open source yeah than America

I yeah I know I know that it

will be because they're going to because they also have to bottom line that positive. Yeah, yeah, yeah, yeah. Or not whether or they're going to keep

un-unut, open source, or the predate-and-source, or close-source. But if they're they're just, men-nour-mellue merang-cull

the predatation that's more than to the countries and other banks in the world. In the world, yeah. The use the most it's more than more than it's a more than it is a more than

for me, for sure, that's all of the same, and design architecture to the whole thing. So, yeah, so much, the philosophy or the smanget of

the time, in Nvidia, we have to be, we have to be architecture for anyone. So, so, we have to deliver

very good performance. performance, that's, right? Yeah, that's the letter of Jensen, also, came-in-source, yeah,

smacetka, whatever model we're doing model agnostic. We want, we want, we're running on our platform, right?

We believe, you know, it's a very good chips and architecture, and that, now, now, the problem open and closed

debate is, everybody, everyone needs to sit together. And, this is there really point point valid, yeah, and,

especially open source, that's the law, if we're even really, we're able to access to the world, and also, the world,

and also progress, because with open society, it's, rate progressing, is very, if, if, if, if, if, if you're

all the GitHub, and much, all the open source that, wow, that, that, out of, exponential,

here, contribution, resist the researcher. So, I think we have to be open source. Don't even if it's not yet to be an actariccan because purely closed source and motivation business.

I want to ask, this is Lin Wen Feng who is in Tokyo. This is the mosty it's been inspired for many people in the world, especially in the

country of bebeang or in the country miskin. But back to the omongan If you had, if you're going to be, it's still there's

there's there's if they're exposed with a situation that's so what so what

is the to to be deep seek moment or non-linear or non-linearity It would be very challenging, for

sure. Yeah, yeah, very challenging. Because like I said the same energy, if not quite, it's very challenging, for model model that to be be able to be able to be. Just as I've said, the core of engineering

is problem solving. In fact, if we talk about deep seek, why deep sick that can be able? And people are people hebo. They to dance around their limitation and constraint, right? He's not to give the hardware the world's the most

the world's from the US so that bandwidth between chief that's small, yeah,

right, he said, okay, we live with this, now what? What can we do? Right, that's,

right now, they come up with tricks that where, yeah, maybe people people, not be able

there, because they don't, there, there, like, there, like,

there, there, there's, there's, um, what, what, Sorry, the matter of necessity, the matter of invention.

Because they're but too, how, how much, how much with the under-terbatasance hardware, like that. Right, right. I'm not sure. I'm going to be argumentation. But,

the better than we have energy. If you're talking about, electrification's already $10,000. So I'm seeing it, the of the our own episode or moment Spudnik or moment or episode

non-linear, like Deep SIG, that's the catersatessen systemic-nick, yeah, that's energy. We're able to

make sysat whatever, but still will be abratum. Because electrification. I agree, sir.

Yeah, bottom-lake-not-n-n-lake-not- not-maw-n't-n't-n't-nless. We come up with new energy sources that, you know, maybe can dobroop capacity our capacity

our electricity we, if if the growth like like now, yeah, unfortunately, like it or not,

We'll hit barrel neck, that's why. I mean, I mean, I'm not necessarily invention, I'm afraid there's doberation too from line energy, like,

how, amen. Yeah, right? Because this is pipeline, you can only be as good as your weakest point, you know, this, this iser

together, that's the answer, that's together, let's be with them, barren. Because this, I believe, I,

more challenging or big problems that you're trying to solve, it's more butchew collaboration and cross-pollination lintas pittal,

but not just only AI engineer that's just not just only energy engineering. Everyone has to do with us

together, we're doing, we're doing, we're going to one and other. Because problem we we're trying to solve, this is so much bigger

problem is, it's more, it's more, we need to work together and cross-pollination and collaboration. I'm looking

Indonesia, has, can be able minimum 30,000 per year,

not can only with 5,000 megawatt. If we want to be the narration

AI. And AI is, you, yeah, people at the company in

the company, like, to make using AI. And, and,

the trend after this, it's not generative again, and A.

A. A.O. A.O. A. A. A. A. energy is more than generative.

Because a loop, Merencannacanate, concre-enecta-can, recircuecutioning and re-investigation that's intuitively, the impugnaugn't disproportionately, much,

right, right, yeah, we're really, we're going to if, if, design chip that always, you know,

need to be at least, you know, so-and-so-X energy efficient, more energy efficient. Because it's, because it's, how energy can't be able to

turn? Yeah, energy has to run. Performance, of course, still, still, performance per watt.

Right. The metric is better than, because what, energy matters. The lucky set, sir. So,

so, team that, must be thinking, how much how it, in level chip, if it'sal, energy that's

true, yeah, in level cluster, data center, that's there's center, that's, there's,

optimisation they're from the from the air cooling so liquid cooling yeah, that's safe energy and water, that's the time,

yeah, so, yeah, so, yeah, in the layer and line, it's, engineering and so, engineering,

and so, energy consumption, this, but, that's, that's, the level macro, we're,

we're, to boost energy supply our, but, why, why, Jeffson paradox.

And, And then the energy it's down. Oh, we need more for the tasking to be bigger. That's natural human manusia,

ma. Yeah. in the long in the long, to increase computation or capacity computation

or making efficiency transportation data. For the of the important

of the energy So, so there there nuance in there,

context, yeah, and proportion of the the part of the

context of the country that's beenbeam because compute is over-biasa tingy, that's, right? Yeah, right. So, yeah, right. So, if not there's supply data that that much, that, like, that. Yeah, that's not.

Yeah, that's... Yeah, but, that's... Yeah, but, but... Yeah, but... Yeah. But, there's also, in H.A.I., some cases in H.I. Computeing that, it's not too waugh

too, so... So, that data not has to supply so much, and it's... ...and it's... ...and, you, It's gaiting, really, nuance of what, see,

what's the domain is the, yeah, right? So, because as a architect, like chip architect, we, I'm trying to, team, to understand,

the requirement from the software, algorithm, and also hardware. People are you software guard or a hardware guy? I'm a guy.

What do you want? Yeah, sir, right? Because we're not. You know, what, I'm glad. Yeah, exactly,

I think I think I think it's important to look at in the way because if we just try optimisation

we're able to puttah in big picture yeah we're stuck in local optimally we're in the end

we build system we have understand system yeah because I'm like I'm going to people algorithm in the

in the kind of hardware yeah because I have been because I'm the gembatan yeah

I'm who's like hardware like this, software, like this, software, like this. Now, back to contact

to the next to it's not really, yeah, it's got to again, the domain of the amount of the

amount of the need to be so and so-and-so. Yeah, right, from there, it's derived again,

it's down, so, so, to support computation requirements like that, how bandwidth

that need to be for the capacity that we need to we're we're down from there,

so, so, then, who's if you can be able to it's not back again.

So, yeah, right, so it's really, really, it's really depending

nuance of yeah. Yeah. We can keep chips too, um, constraint in

data center and contrast in self-driving cars, that's bad, yeah,

right, can't be generalize some of the other. You know, you're going to look at

the design chip, it, the, the, the, the, where the

application to the mobile phone, or to data center, to transportation to, what, for people, in the world.

Well, there's been, many, but many, many domains that we can unlock, with,

with AI, particularly, um, Bapap, maybe I've seen read, article,

article, that many people who try apply AI to medical science, drugs discovery, curing cancer,

Yeah. So, pa. And I personally am very excited when he, it's able to unlock how things before in the new medicine. Yeah.

What's the do you doggia? Yeah. Exactly. Protein unfolding. Exactly. Exactly. Exactly. And that's true reinforcement learning.

Who would have thought? Yeah, right. Exactly, yeah. Yeah. Exactly, yeah. Right. For the computer, yeah. For sure, for.

For chemistry. Yeah. For me. Yeah. Yeah. Yeah. Um, um, on course,

we're still in the era data center, because we're still building foundation. Tremont, like all CEO, big tech, said, the next wave will be physical AI, robotics,

and much, but when will we get there? We, not know, so, at least, for the time being, we're really really heavy

in the era data center in this, at some point, we'll get to physical AI. But, Not anyone knows how long. There's a prediction of much, but, Indian, we don't know,

it's got to be the curve of growth here, but, um, but, um, but, um, inrude I, um, this very exciting era.

Why, because, because this, AI, this enabled us, to, to, embaltani idea and execuice,

so, gap-ne is so much yeah, and productivity. Yeah. Now, I personally, I,

I feel like to know yeah, well, even though, it's not the way, because I'm from the principle,

for unsolved problem that's a problem that best to cross-discipline expertise. Not-munk- can't-

you can't. So, naturally, I, too, problem-driven, problem what what I'm going to solve? I'm-

know what I have to solve problem this. Not-munk- that's just one. I need to talk

to biologists, and the other and the other, yeah, right, yeah, this is, this can

be able to end up us with initial assessment, like, like, like, what,

like, what, we can be okay, try, see, the state of the art.

Yeah, right. Now, I'm, my, this, we're, we're,

more and more system thinkers instead of just specific silo of like, oh, yeah, you're hardware

guy, you're a software guy, yeah. Yeah, yeah. I have one panut

in part in my bedang in the person named Herbert Simon he's one of the computer science got a touring award

where it's a Nobel Prize to compare science and he also got Nobel Prize in economy later later

after we get an Oscar because system stinkler because Grammy Grammy yeah

yeah yeah yeah if you're if you're people who know the

yeah yeah too can be if they did ask with people what, what's, what's, what's,

what's been acuant or panutant? Yeah, there are two book who are very contrast, one

same to the other that's the outliers. That's right. 10,000 hours. Early specialization.

But, one other other is the range, which is by the judge. Yeah, yeah, yeah, yeah.

Yeah, that's a yeah, yeah, he's aughts specializer. Yeah. It's,

these, some of the different to be able to thinker. I'm more like range from the outliers.

Because it, to do not to get back to be more to be systems thinker. Now, this AI,

this can, can be able to be able to hallucinacy that's very very, or, or the

hypnotice. Halusinati, yeah. Yeah. If, if the, if notusiness it,

that's from from the IQ's 14 yeah hallucination yeah alusination but

if you're hypnotis not from people who the people who quality the quality

also the quality I'm going to back to topic our our and you

the product of good business but but Asia

Tengara it production system yeah the year that

just 750 R Tong Tionk 4. Tuck America is only 800,000.

I see, the more we've got to askerkan people who are young, stem, the more of the power of narration AI to the right to the right.

This is not zero sum to the non-STAM. But we're conversi the stemmed in Indonesia. This is a small. The same is the Asia, the Kengarer, the most

is the same thing is Malaysia. They can't conversion 40 to 43, and 1⁄2% of from the other 100, get 3,

nambil stamp. Singapore, Kala, 36. Indonesia's just 22%. How much, we're going to get to before,

we're really, we've been talking about energy, panang lebar on the biya transportation data, panang lebar

on small language model, but So, so, the per-daiouration AI, we're becoming become we're making

people, not pen, how? Yeah. How? Pugh, that's a million-dollar question, ma'billion dollar question,

because, because, talent is the bedrock yeah, from, from system technology AI, this,

um, if talent, quality, we're going to be able to be you said in the end we're using, we're only again, we're only again, we're

or penantone, yeah, buttool, yeah. And this, I'm still mernung and try to try data, that,

why, the output is the endah, like that's purely because not there's interest, yeah, or purely,

maybe, um, not capable, you know, the children who can um, injures to STEM. I'm a big believer of a growth mindset.

I think every kid is pretty capable. We just we're going to unlock the potency. If we're going to from that and we can see output of MASSERRENDA, it's a matter of interest,

isn't, ma, it's a matter of interest. Now, now, how much it's, we're more more more

to produce student STEM. There are some of the first time. I think exposure is very

important, if BAP, if you've ever to Stanford, it's a mattolea CS-151

Tinkat Purtain, Frontier Systems. This are not an-an-anat- first, that, you

get-in-a- class that. Every week, the people, the conversation is not kind of

Kaling, sir Alpian, Jansson Wang, Satiana Dela, Andre Carpetty.

Now, people who people who, you know, back, you know, back. You know,

school, back, mind of the mindset. This is, there, can't get to be

there, there, video in YouTube there, but, yeah, if,

if, if you can, if you can, it's possible, it is possible. It's not

a pipe dream. That's, I sure, I'm interested to STEM that's more than

more opinion, I'm opinion, you know, I'm personal stories I'm going to America because there docent tamu

who was it who could be at Tokyo Tech and about how asic-ne from Michigan

Yeah, from Michigan yeah, yeah, so. So, I think, wow,

this is very cool. Now, how we're We're making narration and ecosystem that in the United States, so so, not a lot of the people that,

it's just a interest, like, so, but if it's there, and, and motivation, that's the other that's going to, yeah. I always say to

the why is strong enough, the how will follow. We're just to be bangun why-in-it-it-it-you-law, so, if you know, if you're going to Manda Aceh, to Merokee.

Currong-le-le-per-per-send from the Kepalahsha. Not-punate-punuched, not-punuched, you've got to becky. The people-tua have been a pen-diction. But, majority, 90%,

from S-Bank, not-Pen-N-N-Broquet. So, it's hard to us, we're not for us, that a child that is that is put in-cult-and-a-it-inspired, with inspiration, like imagination, like, ambitia, like, and,

and, succour, succour, to do you latif, to play, lathian only to be able to tinkies, yeah, lathian only muckul bula, like, that's a lot.

And that's, there's a lot. And so, so, in the absence of the people who who, can't-l-l-a-tribute

like that, is a guru, Guru, right? This is true, right, this is I like going to go about guru, which, from 3.5 million guru that

that's the majority not be beckh with compensation that's been so it's really to recruit guru that good quality,

I'm sorry. So, if I, if I'm lures from Tokyo Institute or U.T. Osten,

rather to be a Gaze 2. Muccia, maybe it'll be invidia. Pendapotan the end uptance now, this, without any reformation,

guru, or gaji guru, is difficult. For we can use, how to make sure how to be conversi

the kids, young, to get manta plazeran that more non-relevant with the past the time.

And they're not can't be a bit of , ah, I want to be satiata, Because I don't know concept for Gereida Adela is what? Yeah.

Yeah. Because guru, it's, he has to nontick imagination, Yeah, Sto-Bee-Ambi and serendipity. Serendipity that can be able to like to. Yeah. Yeah, I see, I'm going to-lorn-indul just.

If guru, if you're gaijee $30,000, yeah, maybe you'll come back to Aceh. Nah-jur. I'm going to bea-cambing, ma'ambing, I'm. If you're going to be like, if there's a lot of the people from Sumedang,

he's lulles from school that's great, computer science, PhD, and scathe, and so oner as a guru at Sumedang, 30,000, and then get a parloran, and get a matter 50,000 per-bulan,

he can't be able to come to summedang. Because with 30-juta, he can't put on the table, he can't meet the people, he can't meet

people. It's not even at the Sundang, from therein'clock in Mountain View. Yeah, right, right. But if, if you're a badinging of $50,000 with $2.8,000, Selsa, maybe I'm going to go to Mountain View. Right, ma'u. Many, though,

that, who, who, actually, I want to be resumbang, to comeg. Yeah, so long as he has put up to be on the matronan on the table. This is structural. to move

reformation system ajaran the system after we're I've been over with Kai Fulie

I get to and then he's one one proponent about on the A.

I'men E.A. that year he's embanked one concept with

the power E.A. How can we can talk smart teacher in in the city.

And then with the people are how it can be even 120 smart teacher this

means real person right? Smart teacher real person and there smart teacher prime

AI yeah yeah yeah that can and interpretation uh...

uh... uh... uh... from 120 campus in the allur the world or in

the cabopaten, the province or what and it filtering wisdoms is and the

endawing wisdoms also also. That's one of many other of many ways to

we can puttimbang can gulirken teaching methodology or teaching quality that's more more

big. That's true so, so, so, so, so, very interesting

perspective, because this, but, you, B'Ubbaugh B'uuk that's,

Also, if we look at ratio of guru and murid in Indonesia, it's that's all right-biasa, superbara. One guru that has must handle more than 1,000 students, if we're purely

looking at a number. Yeah, right? So, there's a better not even one class, can't even, 1,000, siswa. And I'll look,

I wonder, apaka, this is assist in there, that, so I'm still a big believer for in-person teaching. Because, I mean, I mean, for

each students, it's a human connection. Oh, yeah, very, right. Yeah, right. But if there are to be able to be able to

quality, yeah, maybe we're, idealing, we can clone. Yes, exactly. But maybe virtually cloning is one

that's one that is very quality there in Japan, and then he'll be not, but, not, but maybe 10 or

100 and that's it's a better back. Another alternative um that I came to

my mind when I'm heard of Pat Gita is is open course I'm not know state of

the art open course in Indonesia like what quality umbrick it's like

what what? Emerging technology if if open source it is the course

that the course that the degree it's top rate yeah So, right? So, you know,

not really in particular location, but all that can't be able course now. Right. So, they're to make sure

E.A. I mean, there's professor at MIT named song Head. It, he's, it,

gate, for efficient and edge AI, yeah, so, after the first the course that

the public, wow, that penetrations were more abysa, because it's all the all.

That's because, yeah, it's because, I can, I think, in the other countries be coming, but how do you make make make make that you from from

from from from that's from that's right, that's right, ideal is there there is a idea,

we're but we can but we can have to be able to there, yeah, right, yeah, yeah, interpretation

in their, to, asperation, expectations, and, if, even, how much, how much, how to build, memup, merwit, and

everything, so that's that's going to be able to that's I, I, I'm, I,

that's the in-person, is the part of the but with the betterbatassan, and I'm looking at

good, not mental, for the country of the sub-aumasance and that, to the future. Yeah. Yeah. Tionk, in 2004,

it's under 1.8 million patent. Asia Tengara, Conan, for 20 years, and in 2004, made daftar can kind of 3,000

patent. In that we've got to look, it's really it's that's the so little, and the kids

in the young in Asia Tengara. Now, if we can look at there are around

which is five-five campus are in Indonesia. Jumla murid's in Asia Tengara, 25,

but two campus that's on two campus two times two in the United,

Tijuana, Tijuana, Tijuana, but it's been able to putt can two, in the Peringkate 20th atas, menopatting at 8 at the Prynx 10thus

for the importance of research. This, this is a structural in a few countries, in Asia, Tengara, to be in Indonesia. It's,

to be able to be a panku. Right, right. How much? I'm not, I'm not. I'm not.

Japan, and Amrik, I see, yeah, it's, there's a problem fundamental, you, sir, system of education, especially in Indonesia. If we're going

to be born Einstein, Einstein, from Indonesia. What I've seen, in a person, you, sir, as far as, you know, but, you know, quality gurue, yeah.

Many gurus, of the guru guru who are smart and capable in Indonesia. Just as a system, maybe it's, they're not able to be the best version of themselves, because system that's not not much of why and critical thinking. I'll give you this,

back. At first I'm at SMP, SMA, it's the many of the questions, of our own, it's a piling ganda. And, and if we're learning mathematics, that there rumus, you know,

the number to the rumus and then what's what's it, but I'm

really, I'm really feel like wow, this is very different. When I PhD program

at UT Austin, I have got to give pre-qual so so before

PhD, to be PhD to be PhD candidate to qualify exam.

Yeah, co-thewing exam. That, examing that's oral, you know, not the case to know

to ask what you have learned, fair game. That's the endang with Professor Punguja,

there's five that, but, but the money is a board. Right, and we

ask, okay, we make this system this. Now, try you design from scratch. How much

step by step by step yeah? Yeah, right? Yeah, ma. Assum what you can

requirement what? Bebered all, so after it's not there. So,

not there's not, no, no, no, you build a system. Right, from scratch. That, that's,

the, ourstack we got, we're, back, from, from, and if you

don't have a solid understanding, is it impossible. Selsa. Selsa, not much, ma,

ma'amak. And, So, this, I think I, it's the thing's the same it's not,

we can turn-k to level of education that's same, sir, in UT Austin

in the introduction to computer system, yeah, right? When we, for class freshman

this, I'm still we think, we can soal exam, it's past essay and pasty open-endid

that, that's been before before before in exam exam before me. The A.

who's who's the A. The A. Yeah, the A lot of original questions. But it's an investment. It's not a cost. It's an investment. We're still

think, waddu, how do you? The problem is, why? The problem is, the original problem. The second,

has also to get great it. If it's too hard to grade it, can't even too much more than this is. For we know how important this is,

for engineering and engineer that's new, but I mean, it's the, you know, maybe it's It's not to have to be it

in Indonesia, where we make make a lot of people who have been able to put in just making

just making roomus, but what can't be there remus it? Intuysinia like what? Try,

turn it from scratch, why can't be rumus that's about problem solving, Pa.

Engineering is not memorization. Yeah. You know, you need to know how to solve the problem from the first

principles. Even, not have to make rumus. Yeah, That's what engineers is about. The thing that's what I'm saying. They're who's about.

They're who's about binary. If you're not null or not one. If you're not to know the answer they, they're going to know. I just know. But if you're not saying non-STAM that's what's the that's the ambiguity or subjectivitist.

So if if people stem, it's a common, it's the more objective. And that it's a more objective. That's the the same-being the efficiency

of the velocity and acceleration that's more be part of job-to-know- I'm going to try to book us

we're talking about brain-train that we're manifestation in Japan. Talenta local not-biasa. We've got

about going to bein-gain. With get-turbcaan engundang talent from all-d- world, there's led

industrialization. This, this, Australia. The first, Japan. The third, this brain circulation. This is done by Hongok. Gila, gila.

Semeniapeneged to 18, since then 7,000, the majority of learning STEM, majority of school, Jailash, Jambi, Wai, Huawei, Opo, Xiaomi.

The last, brain linkage, India, yeah. 50 years, heirim, more than 7, million.

Talentah. majority of the majority of them come to come to come. It's jadellas Sondar Pica that's every give W.A. Yeah.

To temen S.D. Yeah. Toemn S.P.E.N. Spupu's it, in Gujarat, like, in Bangleur, like, in Calcutta, like, in Delhi.

Apapot, this, this, this, this is realisalization to, the negaran, kebansan.

Now, I'm looking, Indonesia, brain-train, sulet. Because we don't have a scale. Many people

But not many people not enough to be enough to make industrialization This is this is a

ratio manufacture per-d-d-d- in 12-a- year, turn from 29% to 18%

we're we're in- we're-gain we're to make-t people who under-un-tank-t to come-to-

come-to-cili- to-cath-cuit- we can-cath- -cath-could- we can-a-catch. And the three, we have We have done brain circulation. Calculation, casar,

we have to give 2,000 people who are Indian Indianian brilliant to learn anywhere in Tionk, like in Japan, like in German, English, America. And not use to be too mabok with idea they have to be able to. They're going to beer-perged-in-V-Betia,

Bia, every day, I'm going to get him WA to pacarer in Sumedang, to her own Tijuana, to Gnanzh, to the friend SD's in Wono-Sobo, in Sabang, and in anypun. The last, brain linkage, I'm not looking at Indonesia, because in diaspora our diaspora we, in the above $300,000, in 50 years.

India, $7,000, $0.00. Now, I'm concerned, pananguant, Yeah, is this panangor or people are making the mind brain gain and brain circulation,

than brain train or brain linkage? Yeah, begin me, I'm looking at you, yeah, in Silicon Valley, yeah,

that, there, many engineers from different different countries, work on, because even kibbleth

technology, is there. Now, I'm back in, number one, engineer that's the first, engineer that's the

from India and other than the other than the other than India. Number two, obviously, China. Yeah, in a population, India number one, China, number two. Population global, number three is America. Fair, that's home turf in America, when you're America,

many people might wonder, number four, this is what, the population of the world? Indonesia, now we're, Now, if we're looking at Silicon Valley, in Silicon Valley, is apocational that? Engineer, many of engineers in Indonesia in the Silicon Valley.

There's, ma'am. We're not enough of Korea. We're called the jumblown. We're the number of the Japanese. Even sometimes Filipinos. Singapore. Singapore, exactly. It doesn't make sense.

We're number four, ma, in terms of population. VP's my question. I said, wow, you're the first architect that we hired in this Ort from Indonesia. Where have you guys been? Because he's taught we're number four in the world.

Yeah, right? What would be your message to Indonesia? I mean, I like playing ball, so, I'm framing simple-a. Like, in context of the sport ball. To make surestscied ball to make surests to the country, it's simple. There's two ways,

part. First, give as much of talent as of ball you to the ligas league that people who are quality in the Europe,

yeah, and especially, yeah, and that's all the people people who play in Europe that's always

the quality of good. The other, there's also, there's also also, okay, give people who are

capable and competent to the Liga Domestic you, this is the Saudi Arabia and MLS, America, Munda Messi and much-macom. So simple

that, Pa. So simple, it's not binary. I'm an engineer, but I don't think in this case we should apply binary thinking. Yeah, can? Give talent, talenta, talenta, Indonesia as many of the many. Bearing them, be able to get ridse again.

Look at how technology the world, it's working, right? Because perspective, it's important, ma. When the perspective, they're out with people who get to decide what's the next iPhone

would look like, it gives you perspective, pa. Yeah, right? When they're gopi-nopi with startup startup founders and just got billion dollar funding, they'll beckxed, yeah, right? And perspective matters.

So, after the a lot of they're accumulation and forced to dream bigger and also think, maybe I too can change the world, that's, man, that's, but, that's, back, exposure that I said, It's from C.S.1.5.1.

Or with the people, or with the same thing, we're getting onjerd and injure we're to Silicon Valley. The second valley. If, if, if there's talenta that we can be

badiacan in the country, let's do it, like, like, like, analogy MLS and Saudi Arabia League,

can get to be. Hey, come to this. We're, we've got to bea ecosystem it. So, soing-in-ta-tal-local

that's not to be to be around the people that can't feel like to be going to be going to be

going to be being put in we're going to be able to be able to get to be not going to go back

value of their can be more more than ungul again when they're such as much yeah yeah

yeah the endini you're intuement brain circulation with brain what name gain

per bough yeah any final message feel free to say it in Japanese ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha ha and I'm gonna hangout and I'm gonna be black gamutting

yeah yeah yeah yeah yeah yeah so what we're discussing with I just

um I'm mean to say to the young young men young young

Indonesia, that they're the next future of Indonesia. By success or

not in Indonesia, this is in the hand their right. And we

have that advantage in terms of population. But, but it doesn't

be a capability intellectual. I'm I'm in young young young,

you, we're same same, we we're, we're, we're

up up, uh, uh, we're um, from, um,

we're, yeah, so we're we're being a generation, and generation that's usil

to mopeck and innovation, yeah, okay, like, we'll, look, and,

and, but, yeah, don't every day, so, like, so,

you know, so, so, so, so, many, it's, many,

at YouTube, because it's also very interesting and that's going to make the value their value

that's more more more than will be and will be the end up the end up the next

for their and yeah I really hope that we will get there with generations new

we're in another new than in person just just in the same

just Nanty, maybe two hours again, but try it in 30 minutes. Hmm.

Um... Blank, Blank. Arigato, Goda, Mazta. Thank you. Thank you,

thank you, last minute. Thank you, thank you, thank you, Muhammad Asri, lead architect from Indydia.

Thank you.