WEBVTT

00:00:00.560 --> 00:00:03.037
Even if the technology goes as fast as I

00:00:03.120 --> 00:00:06.397
suspect that it will, we we don't know

00:00:06.480 --> 00:00:10.957
exactly how fast it's going to drive
revenue. We we know it's coming, but

00:00:11.040 --> 00:00:15.277
with the way you buy these data centers,
if you're off by a couple years, that

00:00:15.360 --> 00:00:18.877
can be ruinous. It is just like how I
wrote, you know, in machines of loving

00:00:18.960 --> 00:00:22.957
grace, I said, look, I think we might
get this powerful AI, this country of

00:00:23.040 --> 00:00:25.517
genius in the data center,
that description you gave comes from the

00:00:25.600 --> 00:00:30.317
machines of loving grace. I said, "We'll
get that 2026, maybe 2027." Again, that

00:00:30.400 --> 00:00:33.277
is that is my hunch. Wouldn't be
surprised if I'm off by a year or two,

00:00:33.360 --> 00:00:37.757
but like that is my hunch. Let's say
that happens. That's the starting gun.

00:00:37.840 --> 00:00:41.277
How long does it take to cure all
the diseases, right? That's that's one of

00:00:41.360 --> 00:00:45.597
the ways that like drives a huge amount
of of of of economic value, right? Like

00:00:45.680 --> 00:00:49.277
you cure you cure every disease.
You know there's a question of how much of

00:00:49.360 --> 00:00:51.997
that goes to the pharmaceutical
company to the AI company but there's an

00:00:52.080 --> 00:00:55.917
enormous consumer surplus because
everyone you know assuming we can get

00:00:56.000 --> 00:00:59.597
access for everyone which I care about
greatly we you know we we cure all of

00:00:59.680 --> 00:01:01.597
these diseases how long does it take you

00:01:01.680 --> 00:01:03.757
have to do the biological discovery you

00:01:03.840 --> 00:01:08.957
have to you know go you have to you know
manufacture the new drug you have to you

00:01:09.040 --> 00:01:12.077
know go through the regulatory process
I mean we saw this with like vaccines and

00:01:12.160 --> 00:01:15.357
co right like it there's just this we we

00:01:15.440 --> 00:01:19.117
got the vaccine out to everyone but it
took a year and a half, right? And and

00:01:19.200 --> 00:01:21.517
so my question is, how long does it take

00:01:21.600 --> 00:01:24.957
to get the cure for everything, which AI

00:01:25.040 --> 00:01:27.757
is the genius that can in theory invent

00:01:27.840 --> 00:01:32.397
out to everyone? How long from when
that AI first exists in the lab to when

00:01:32.480 --> 00:01:34.637
diseases have actually been cured for

00:01:34.720 --> 00:01:37.517
everyone, right? In in you know, we've

00:01:37.600 --> 00:01:42.157
had a polio vaccine for 50 years. We're
still trying to eradicate it in the most

00:01:42.240 --> 00:01:45.837
remote corners of Africa. And you know,
the Gates Foundation is trying as hard

00:01:45.920 --> 00:01:49.837
as they can. others are trying
as hard as they can but you know that's

00:01:49.920 --> 00:01:53.277
difficult again I you know I don't
expect most of the economic diffusion to

00:01:53.360 --> 00:01:57.197
be as difficult as that right that's
like the most difficult case but but

00:01:57.280 --> 00:01:59.437
there's a there's a real dilemma here

00:01:59.520 --> 00:02:01.997
and and where I've settled on it is it

00:02:02.080 --> 00:02:04.957
will be it will be it will be faster

00:02:05.040 --> 00:02:06.957
than anything we've seen in the world

00:02:07.040 --> 00:02:09.837
but it still has its limits and and so

00:02:09.920 --> 00:02:13.277
then when we go to buying data centers

00:02:13.360 --> 00:02:15.517
you know you again again the curve I'm

00:02:15.600 --> 00:02:18.477
looking at is okay. We you know we've

00:02:18.560 --> 00:02:21.997
had a 10x a year increase every year. So

00:02:22.080 --> 00:02:27.037
beginning of this year we're looking
at 10 billion in in in annual in you know

00:02:27.120 --> 00:02:30.477
rate of annualized revenue. At the
beginning of the year we have to decide

00:02:30.560 --> 00:02:33.837
how much compute to buy. Um

00:02:33.920 --> 00:02:39.037
and you know it takes a year or two
to actually build out the data centers to

00:02:39.120 --> 00:02:40.637
reserve the data centers. So basically

00:02:40.720 --> 00:02:44.717
I'm saying like in uh 2027 how much

00:02:44.800 --> 00:02:47.837
compute do I get? Well, I could assume

00:02:47.920 --> 00:02:52.717
um uh that uh the uh revenue will

00:02:52.800 --> 00:02:54.717
continue growing 10x a year. So it'll be

00:02:54.800 --> 00:02:58.557
uh one uh one 100 billion at the end of

00:02:58.640 --> 00:03:02.557
2026 and 1 trillion at the end of 2027.

00:03:02.640 --> 00:03:05.197
And so I could buy a trillion dollars.

00:03:05.280 --> 00:03:08.317
Actually, it would be like5 trillion
dollars of compute because it would be a

00:03:08.400 --> 00:03:12.157
trillion dollar a year for for five
years, right? I could buy a trillion

00:03:12.240 --> 00:03:14.157
dollars of compute that starts at the

00:03:14.240 --> 00:03:17.597
end of 2027. And if my if my revenue is

00:03:17.680 --> 00:03:20.637
not a trillion dollars, if it's even

00:03:20.720 --> 00:03:23.677
800 billion, there's no force on earth.

00:03:23.760 --> 00:03:27.997
There's there's no hedge on earth that
could stop me from going bankrupt if I

00:03:28.080 --> 00:03:32.317
if I buy that much compute. And and so
even though a part of my brain wonders

00:03:32.400 --> 00:03:34.957
if it's going to keep growing 10x, I

00:03:35.040 --> 00:03:37.117
can't buy a trillion dollars a year of

00:03:37.200 --> 00:03:40.397
compute in in in in in in in in

00:03:40.480 --> 00:03:43.757
2027. Uh if I'm just off by a year in

00:03:43.840 --> 00:03:47.757
that rate of growth, or if the the growth
rate is 5x a year instead of 10x

00:03:47.840 --> 00:03:50.397
a year, then then you know you go

00:03:50.480 --> 00:03:53.997
bankrupt. Um and and and so you end up

00:03:54.080 --> 00:03:56.237
in a world where you know you're

00:03:56.320 --> 00:03:58.957
supporting hundreds of billions not

00:03:59.040 --> 00:04:01.357
trillions and you accept you accept some

00:04:01.440 --> 00:04:04.317
risk that there's so much demand that

00:04:04.400 --> 00:04:09.197
you can't support the revenue and you
accept still some risk that you know you

00:04:09.280 --> 00:04:13.517
got it wrong and it's still slow and so
when I talked about behaving responsibly

00:04:13.600 --> 00:04:17.757
what I meant actually was not
the absolute amount that that actually was

00:04:17.840 --> 00:04:22.157
not um you know I think it is true we're
spending somewhat less than some of the

00:04:22.240 --> 00:04:26.717
other players. It's actually the other
things like have we been thoughtful

00:04:26.800 --> 00:04:30.477
about it or are we yoloing and saying,
"Oh, we're going to do hundred billion

00:04:30.560 --> 00:04:32.557
here or hundred billion dollars there."

00:04:32.640 --> 00:04:36.317
I kind of get the impression that,
you know, some of the other companies have

00:04:36.400 --> 00:04:39.277
not written down the spreadsheet that
they don't really understand the risk

00:04:39.360 --> 00:04:42.477
they're taking. They're just kind
of doing stuff because it sounds cool. If

00:04:42.560 --> 00:04:44.077
you enjoyed this clip, you can watch the

00:04:44.160 --> 00:04:49.680
full episode here and subscribe
for more clips. Thanks.