WEBVTT

00:00:00.000 --> 00:00:03.459
Imagine the race to build the future of AI. You

00:00:03.459 --> 00:00:06.179
might think it's all about, you know, brilliant

00:00:06.179 --> 00:00:08.099
algorithms or maybe those groundbreaking new

00:00:08.099 --> 00:00:11.380
chips. But what if the real advantage, the thing

00:00:11.380 --> 00:00:14.199
that really matters, is something way more fundamental,

00:00:14.500 --> 00:00:16.760
like an electrical plug? Yeah, like literally

00:00:16.760 --> 00:00:18.780
about having enough juice to keep the lights

00:00:18.780 --> 00:00:22.920
on. And then some. Maybe less about the code,

00:00:23.019 --> 00:00:25.760
more about the current, right? Welcome to the

00:00:25.760 --> 00:00:28.420
Deep Dive. Today we're looking closely at a really

00:00:28.420 --> 00:00:31.460
fascinating, honestly a bit unsettling observation

00:00:31.460 --> 00:00:33.859
from some top AI researchers who just got back

00:00:33.859 --> 00:00:35.820
from China. We're going to unpack how something

00:00:35.820 --> 00:00:38.840
as basic as foundational really as energy infrastructure

00:00:38.840 --> 00:00:42.060
is. Well, quietly deciding who's actually going

00:00:42.060 --> 00:00:44.460
to lead this AI revolution. It's a real eye -opener,

00:00:44.799 --> 00:00:47.240
kind of shifts your whole view. we'll compare

00:00:47.240 --> 00:00:49.880
the energy strategies the very different paths

00:00:49.880 --> 00:00:52.060
taken by the US and China see how that impacts

00:00:52.060 --> 00:00:54.640
AI development on the ground and Then think about

00:00:54.640 --> 00:00:58.079
what this this gap means for all of us. Okay,

00:00:58.100 --> 00:01:01.399
so let's dig into this observation a group of

00:01:01.399 --> 00:01:03.780
pretty respected AI researchers from Silicon

00:01:03.780 --> 00:01:05.900
Valley they recently toured China's tech hubs

00:01:05.900 --> 00:01:09.900
and They came back with this one sentence that

00:01:09.900 --> 00:01:12.739
should really you know, make people pause. And

00:01:12.739 --> 00:01:15.079
it's a powerful sentence, yeah. Everywhere we

00:01:15.079 --> 00:01:17.280
went, people treated the availability of energy

00:01:17.280 --> 00:01:19.620
as a given. Just think about that for a second.

00:01:19.879 --> 00:01:22.760
Can you even imagine that mindset here? In the

00:01:22.760 --> 00:01:25.719
American AI scene, it's just not how it is. It's

00:01:25.719 --> 00:01:28.219
almost unthinkable here, isn't it? But over there,

00:01:28.280 --> 00:01:30.420
for those developers, just solved, done. And

00:01:30.420 --> 00:01:32.280
that simple statement, when you really let it

00:01:32.280 --> 00:01:35.120
sink in, it exposes a pretty uncomfortable truth

00:01:35.120 --> 00:01:37.599
about this whole tech race narrative. Yeah. And

00:01:37.599 --> 00:01:39.180
what's really wild, maybe a little alarming,

00:01:39.319 --> 00:01:41.739
is the implication, right, that the biggest technology

00:01:41.739 --> 00:01:44.299
race of our time might not be won by the smartest

00:01:44.299 --> 00:01:46.760
algorithm or the fastest chip. It could actually

00:01:46.760 --> 00:01:48.959
come down to the electrical grid, the literal

00:01:48.959 --> 00:01:51.340
foundation. So while the U .S. has been, you

00:01:51.340 --> 00:01:54.840
know, caught up in debates fighting these ideological

00:01:54.840 --> 00:01:58.500
battles over energy policy, China has been quietly,

00:01:59.000 --> 00:02:02.280
methodically building this incredibly solid foundation

00:02:02.280 --> 00:02:05.909
for its tech future, almost invisible. but powerful.

00:02:06.370 --> 00:02:08.729
Right. They're AI devs. They can innovate with

00:02:08.729 --> 00:02:10.490
this freedom, you know. They don't have to worry

00:02:10.490 --> 00:02:13.090
about the power bill spiking or if the grid can

00:02:13.090 --> 00:02:15.469
even handle their next big model training run.

00:02:16.129 --> 00:02:18.310
Meanwhile, you've got American teams seriously

00:02:18.310 --> 00:02:20.310
considering building their own private power

00:02:20.310 --> 00:02:23.090
plants or constantly having to plan around potential

00:02:23.090 --> 00:02:25.930
outages. It's a different world. It really suggests

00:02:25.930 --> 00:02:28.379
this race. which we often talk about in terms

00:02:28.379 --> 00:02:31.240
of, like, software and ideas, might actually

00:02:31.240 --> 00:02:33.240
be decided on a very physical level, an area

00:02:33.240 --> 00:02:35.419
that arguably America hasn't focused on nearly

00:02:35.419 --> 00:02:37.319
enough. So this isn't just about the chips of

00:02:37.319 --> 00:02:39.500
the code, then. No, it's really about the fundamental

00:02:39.500 --> 00:02:42.039
power grid. That's the bedrock. Okay, this is

00:02:42.039 --> 00:02:43.960
where it gets really interesting. The difference

00:02:43.960 --> 00:02:46.719
isn't just scale, though the scale is huge. It's

00:02:46.719 --> 00:02:49.979
philosophy. Our U .S. grid, it operates on what

00:02:49.979 --> 00:02:52.740
experts call an incredibly thin reserve margin,

00:02:53.080 --> 00:02:56.030
often just, what, 15 %? Yeah, 15 percent above

00:02:56.030 --> 00:02:58.530
peak. It's like running a restaurant that's always

00:02:58.530 --> 00:03:01.689
like one big table away from running out of food

00:03:01.689 --> 00:03:04.710
entirely. Every heat wave, every unexpected spike,

00:03:04.710 --> 00:03:06.969
it pushes the system right to the edge. That's

00:03:06.969 --> 00:03:09.770
why we see outages. It's designed for efficiency,

00:03:09.990 --> 00:03:13.449
maybe, but not resilience, not foresight. China,

00:03:13.509 --> 00:03:15.569
on the other hand, totally different approach.

00:03:15.770 --> 00:03:18.629
They went for strategic abundance. Their generation

00:03:18.629 --> 00:03:21.110
capacity is nearly double their peak demand.

00:03:21.370 --> 00:03:24.270
Just let that sink in. And that's not waste,

00:03:24.449 --> 00:03:26.930
right? It's a deliberate strategic investment

00:03:26.930 --> 00:03:29.889
for the nation. This massive energy surplus means

00:03:29.889 --> 00:03:32.289
industries that need tons of power, especially

00:03:32.289 --> 00:03:34.550
AI data centers, they aren't just tolerated,

00:03:34.590 --> 00:03:36.849
they're welcomed, they're actually useful. Right.

00:03:37.150 --> 00:03:39.689
What we see is a huge strain here. A new data

00:03:39.689 --> 00:03:42.050
center can actually help balance their grid over

00:03:42.050 --> 00:03:44.270
there. It soaks up excess power. And the scale

00:03:44.270 --> 00:03:47.419
is just... Mind boggling. China's increase in

00:03:47.419 --> 00:03:49.479
electricity demand each year is more than all

00:03:49.479 --> 00:03:51.759
of Germany uses in a year. And just a couple

00:03:51.759 --> 00:03:53.919
of their provinces generate as much power as

00:03:53.919 --> 00:03:56.319
the entire country of India. It makes you rethink

00:03:56.319 --> 00:03:59.419
what national infrastructure even means. And

00:03:59.419 --> 00:04:01.439
they did this through serious long term planning.

00:04:01.550 --> 00:04:04.349
building thousands and thousands of kilometers

00:04:04.349 --> 00:04:07.629
of these ultra -high voltage UHV transmission

00:04:07.629 --> 00:04:09.930
lines. These things are amazing. They carry huge

00:04:09.930 --> 00:04:12.069
amounts of power over long distances really efficiently,

00:04:12.689 --> 00:04:16.089
moving it from, say, remote dams or solar farms

00:04:16.089 --> 00:04:18.970
right to the tech hubs. Meanwhile, back in the

00:04:18.970 --> 00:04:21.810
U .S., just upgrading one old transmission line

00:04:21.810 --> 00:04:24.209
can take, what, a decade? Maybe more. You get

00:04:24.209 --> 00:04:26.649
stuck in regulatory battles, lawsuits, local

00:04:26.649 --> 00:04:29.839
opposition. It's just a quagmire. So it's not

00:04:29.839 --> 00:04:32.060
just about making enough power. It's also that

00:04:32.060 --> 00:04:34.220
we struggle to actually get the power where it's

00:04:34.220 --> 00:04:38.040
needed most. So China plans for abundance, basically,

00:04:38.100 --> 00:04:40.560
while the US operates closer to the edge. Essentially,

00:04:40.759 --> 00:04:42.639
yeah. One builds surplus, the other struggles

00:04:42.639 --> 00:04:45.120
with limits. And now generative AI comes along.

00:04:45.279 --> 00:04:48.379
And it has created this unprecedented thirst

00:04:48.379 --> 00:04:50.180
for energy, as people are calling it. It's not

00:04:50.180 --> 00:04:52.569
just about computation anymore. Oh, absolutely.

00:04:53.029 --> 00:04:55.410
Training, just one of these big models, these

00:04:55.410 --> 00:04:59.310
LLMs, large language models. AI trained on huge

00:04:59.310 --> 00:05:03.009
amounts of text and data, like GPT -4. That single

00:05:03.009 --> 00:05:05.709
training run can use as much electricity as hundreds

00:05:05.709 --> 00:05:09.170
of homes do for an entire year. And that's only

00:05:09.170 --> 00:05:11.110
the training part. Right, that's the key point.

00:05:11.389 --> 00:05:13.550
Because the ongoing part, the inference, that's

00:05:13.550 --> 00:05:16.110
when the AI is actually working, answering questions,

00:05:16.149 --> 00:05:19.509
generating stuff for billions of users that uses

00:05:19.509 --> 00:05:22.100
even more power. Constantly. Day in, day out.

00:05:22.160 --> 00:05:24.100
McKinsey did the math. They figure the world

00:05:24.100 --> 00:05:26.660
needs to invest something like $6 .7 trillion

00:05:26.660 --> 00:05:29.620
in new data center infrastructure by 2030. And

00:05:29.620 --> 00:05:32.040
when you see China's energy situation, well,

00:05:32.040 --> 00:05:34.120
they look uniquely ready for that kind of demand,

00:05:34.180 --> 00:05:36.439
don't they? Definitely. And here in the US, we're

00:05:36.439 --> 00:05:39.519
already seeing this tension, this battle for

00:05:39.519 --> 00:05:41.519
energy between the tech giants and just regular

00:05:41.519 --> 00:05:44.079
people. Yeah, like in Ohio. Reports say families

00:05:44.079 --> 00:05:46.759
are paying, on average, $15 more a month on their

00:05:46.759 --> 00:05:49.379
electric bills. Why? Partly because these huge

00:05:49.379 --> 00:05:51.480
data centers came in and sucked up so much power.

00:05:51.439 --> 00:05:53.800
it drove up the wholesale prices for everyone.

00:05:54.040 --> 00:05:56.439
Wow. And in Virginia, you know, data center alley,

00:05:56.939 --> 00:05:58.420
where so much of the world's internet traffic

00:05:58.420 --> 00:06:02.060
flows, the main utility, Dominion Energy, actually

00:06:02.060 --> 00:06:04.639
had to pause new data center hookups for a while.

00:06:05.000 --> 00:06:06.740
They were genuinely worried the grid just couldn't

00:06:06.740 --> 00:06:09.120
handle any more load. And think about California,

00:06:09.560 --> 00:06:12.519
Texas. They keep issuing these flex alerts, right?

00:06:12.920 --> 00:06:15.560
Asking people, begging them, please use less

00:06:15.560 --> 00:06:18.000
electricity so we don't have blackouts. Meanwhile,

00:06:18.199 --> 00:06:20.300
the big computing facilities, the AI engines,

00:06:20.579 --> 00:06:23.160
they're running full blast. It's a stark contrast.

00:06:23.399 --> 00:06:25.800
You know, I have to admit. hearing about how

00:06:25.800 --> 00:06:28.839
fragile our own grid is becoming, I still wrestle

00:06:28.839 --> 00:06:30.680
with how we let our foundational infrastructure

00:06:30.680 --> 00:06:33.379
get to this point. Yeah, it's kind of a sad reality.

00:06:33.720 --> 00:06:36.120
How did we get here? Well, our grid, the American

00:06:36.120 --> 00:06:37.639
grid, it was mostly built for a different time,

00:06:37.800 --> 00:06:40.120
right? Industrial era. It's just buckling under

00:06:40.120 --> 00:06:42.620
this digital revolution that it was never really

00:06:42.620 --> 00:06:45.040
designed to support. It's a fundamental mismatch.

00:06:45.160 --> 00:06:47.600
So is our aging grid simply not built for this

00:06:47.600 --> 00:06:49.720
digital era? Yeah, it's buckling under these

00:06:49.720 --> 00:06:52.920
unexpected digital demands. Mid -roll sponsor,

00:06:53.220 --> 00:06:56.160
Reed. Okay, so this huge difference in the grids.

00:06:56.399 --> 00:06:58.660
As we've been saying, it's not just about wires

00:06:58.660 --> 00:07:01.360
and power plants, is it? It feels like a symptom

00:07:01.360 --> 00:07:05.279
of something deeper, rooted in really basic differences

00:07:05.279 --> 00:07:07.819
in how the two countries run things, their economic

00:07:07.819 --> 00:07:10.800
models, their governance. Yeah, totally. On the

00:07:10.800 --> 00:07:13.240
U .S. side, you often hear about short -termism,

00:07:13.360 --> 00:07:16.860
right, and shareholder primacy. CEOs, entire

00:07:16.860 --> 00:07:19.399
industries really, focused intensely on those

00:07:19.399 --> 00:07:21.740
quarterly earnings reports. And when that's the

00:07:21.740 --> 00:07:24.459
main focus, big long -term projects like power

00:07:24.459 --> 00:07:26.759
plants, new transmission lines. They just don't

00:07:26.759 --> 00:07:28.540
look very appealing. They cost a ton of money

00:07:28.540 --> 00:07:30.860
up front, take ages to build, and the payoff

00:07:30.860 --> 00:07:33.139
is way down the road. Yeah. That's a tough sell

00:07:33.139 --> 00:07:34.660
when you're worried about next quarter's numbers.

00:07:34.839 --> 00:07:36.899
Exactly. And Silicon Valley's venture capital

00:07:36.899 --> 00:07:39.060
world, pretty much the same logic, scaled up.

00:07:39.339 --> 00:07:41.579
Billions pour into software companies, SaaS companies,

00:07:41.920 --> 00:07:45.060
hoping for that 10x, 100x return in like... five

00:07:45.060 --> 00:07:49.360
to seven years. No VC fund has the patience for

00:07:49.360 --> 00:07:51.420
a hydroelectric dam. It just doesn't fit the

00:07:51.420 --> 00:07:53.860
model. And the results, some argue, is what they

00:07:53.860 --> 00:07:56.160
call an extraction economy. Instead of plowing

00:07:56.160 --> 00:07:59.459
profits back into R &D or long term capital spending,

00:07:59.819 --> 00:08:02.339
capex, you know, for future growth. Corporations

00:08:02.339 --> 00:08:05.100
often prefer stock buybacks, inflate the share

00:08:05.100 --> 00:08:08.360
price, pay out big executive bonuses. It's like

00:08:08.360 --> 00:08:12.180
the phrase someone used, preferring to be emperors

00:08:12.180 --> 00:08:14.860
of a dumpster fire, rather than putting in the

00:08:14.860 --> 00:08:17.240
slow, hard work of building something sustainable.

00:08:17.819 --> 00:08:20.839
Ouch. Right. China, though, operates very differently.

00:08:21.379 --> 00:08:23.180
The government, through these big five -year

00:08:23.180 --> 00:08:26.339
plans, directs money into sectors they see as

00:08:26.339 --> 00:08:28.800
strategically vital, often years, even decades,

00:08:29.139 --> 00:08:31.180
before the demand fully shows up. And you have

00:08:31.180 --> 00:08:33.820
these massive state -owned companies, SOEs, like

00:08:33.820 --> 00:08:36.519
the State Grid Corporation. Their job isn't just

00:08:36.519 --> 00:08:38.620
profit. They have a national mission to fulfill.

00:08:38.700 --> 00:08:41.730
It's a different mandate. They accept that maybe

00:08:41.730 --> 00:08:44.009
some big projects won't pay off immediately or

00:08:44.009 --> 00:08:46.629
might run under capacity for a while. But that's

00:08:46.629 --> 00:08:49.529
seen as a necessary cost, a strategic bet to

00:08:49.529 --> 00:08:51.309
make sure the capacity is there when the country

00:08:51.309 --> 00:08:53.470
needs it. Yeah. Well, it's basic long -range

00:08:53.470 --> 00:08:55.740
planning, isn't it? Absolutely. And that difference

00:08:55.740 --> 00:08:59.100
helps explain why here in the U .S. you see powerful

00:08:59.100 --> 00:09:02.500
lobbies like oil and gas actively working against

00:09:02.500 --> 00:09:05.200
renewable energy development, even when solar

00:09:05.200 --> 00:09:07.419
and wind are getting cheaper. They're protecting

00:09:07.419 --> 00:09:10.039
those short term quarterly dividends. And while

00:09:10.039 --> 00:09:12.470
that's happening here, China's just. building

00:09:12.470 --> 00:09:15.250
the infrastructure for the next century. So it

00:09:15.250 --> 00:09:17.710
really is a clash between short -term gain and

00:09:17.710 --> 00:09:19.629
long -term national interest. Essentially, yeah.

00:09:19.850 --> 00:09:22.669
Quarterly reports versus strategic national planning.

00:09:23.090 --> 00:09:24.730
And then there's another layer, another barrier,

00:09:25.009 --> 00:09:28.230
particularly frustrating for the U .S., the politicization

00:09:28.230 --> 00:09:30.980
of... Well, almost everything. Renewable energy

00:09:30.980 --> 00:09:33.559
isn't just discussed in terms of economics or

00:09:33.559 --> 00:09:36.100
strategy here. It often gets tangled up in the

00:09:36.100 --> 00:09:38.399
culture wars. Right. We're still having fundamental

00:09:38.399 --> 00:09:41.340
debates about climate change. So proposing a

00:09:41.340 --> 00:09:44.379
new wind farm or a big solar project, it often

00:09:44.379 --> 00:09:47.059
gets framed as a political statement, an ideological

00:09:47.059 --> 00:09:49.539
battle, rather than just a practical decision

00:09:49.539 --> 00:09:52.240
about energy or security. China just isn't caught

00:09:52.240 --> 00:09:54.159
up in that same way. They don't seem to treat

00:09:54.159 --> 00:09:57.330
renewables as some kind of moral issue. For them,

00:09:57.409 --> 00:10:00.429
solar and wind, they just make sense. Economically,

00:10:00.629 --> 00:10:02.830
strategically, helps with energy independence,

00:10:03.129 --> 00:10:06.169
cuts reliance on imported fuel, simple pragmatism.

00:10:06.450 --> 00:10:09.090
So they just build them at a stale in speed that's

00:10:09.090 --> 00:10:12.409
honestly staggering. And pragmatically, if renewables

00:10:12.409 --> 00:10:15.590
can't quite keep up with AI's crazy growth spurt,

00:10:16.090 --> 00:10:18.570
they'll fire up. reserve coal plants. They see

00:10:18.570 --> 00:10:21.269
coal as, you know, old tech to be phased out,

00:10:21.289 --> 00:10:23.649
not some kind of pure evil. It's a backup. It's

00:10:23.649 --> 00:10:26.169
that pragmatic focus, isn't it? Let's them concentrate

00:10:26.169 --> 00:10:29.110
on getting results, building capacity instead

00:10:29.110 --> 00:10:31.090
of getting stuck in endless ideological fight.

00:10:31.429 --> 00:10:33.629
Yeah. And compare that to the U .S. permitting

00:10:33.629 --> 00:10:36.070
process for any big infrastructure project. It

00:10:36.070 --> 00:10:38.629
can take forever. NEPA, the National Environmental

00:10:38.629 --> 00:10:40.929
Policy Act, started with good intentions. Sure.

00:10:40.929 --> 00:10:42.909
But now it's often used by all sorts of groups

00:10:42.909 --> 00:10:45.649
to sue, delay, stall projects for years, sometimes

00:10:45.649 --> 00:10:49.580
decades. While China, with its more centralized

00:10:49.580 --> 00:10:53.240
system, can push through massive projects like

00:10:53.240 --> 00:10:55.360
the Three Gorges Dam and the time it takes us

00:10:55.360 --> 00:10:57.320
just to finish the paperwork, the environmental

00:10:57.320 --> 00:11:00.379
reviews. Does this mean ideology really hinders

00:11:00.379 --> 00:11:02.820
progress in the US? Yes. I think it's fair to

00:11:02.820 --> 00:11:05.600
say partisan divides definitely stall practical

00:11:05.600 --> 00:11:08.139
infrastructure development. So this energy difference,

00:11:08.320 --> 00:11:11.019
it's not just some abstract economic issue. It

00:11:11.019 --> 00:11:13.860
hits AI engineers and researchers right where

00:11:13.860 --> 00:11:16.419
they live, in their daily work. It affects their

00:11:16.419 --> 00:11:18.740
freedom to create. Oh, absolutely. Let's paint

00:11:18.740 --> 00:11:21.360
a picture. An American AI developer, maybe at

00:11:21.360 --> 00:11:23.740
a startup in Palo Alto, they're spending a lot

00:11:23.740 --> 00:11:26.279
of their brain power just optimizing code to

00:11:26.279 --> 00:11:29.019
use less energy. Their access to computing power,

00:11:29.100 --> 00:11:31.419
to those essential GPUs, graphics processing

00:11:31.419 --> 00:11:34.759
units, the workhorses of AI. It's limited by

00:11:34.759 --> 00:11:37.080
electricity costs, by grid availability. Right.

00:11:37.080 --> 00:11:39.759
So if a big model training run fails, it's not

00:11:39.759 --> 00:11:41.919
just lost time. It's a serious hit to the budget.

00:11:42.700 --> 00:11:45.019
bill they might struggle to pay. Exactly, so

00:11:45.019 --> 00:11:48.000
they might use an AI tool like Claude maybe to

00:11:48.000 --> 00:11:51.039
help them optimize. Crafting these really detailed

00:11:51.039 --> 00:11:53.919
prompts like, okay how do I reduce the GPU memory,

00:11:54.240 --> 00:11:56.879
cut computational costs, make data loading faster,

00:11:57.399 --> 00:11:59.759
convert this model to run more efficiently during

00:11:59.759 --> 00:12:03.440
inference. That's when the AI is live doing its

00:12:03.440 --> 00:12:06.610
job. The whole focus becomes efficiency. conservation,

00:12:07.049 --> 00:12:09.649
doing more with less. American developers end

00:12:09.649 --> 00:12:11.870
up using their talent to solve problems caused

00:12:11.870 --> 00:12:14.549
by shaky infrastructure instead of purely pushing

00:12:14.549 --> 00:12:16.789
the frontiers of AI itself. Okay, now picture

00:12:16.789 --> 00:12:20.070
an AI lab in Shenzhen. Energy is plentiful. It's

00:12:20.070 --> 00:12:22.809
cheap. Those researchers, they can chase the

00:12:22.809 --> 00:12:25.009
wildest, most resource -hungry ideas they can

00:12:25.009 --> 00:12:27.450
dream up. No hesitation. They can train bigger

00:12:27.450 --> 00:12:30.570
models way bigger on more data. Experiment with

00:12:30.570 --> 00:12:32.669
super complex ideas without constantly checking

00:12:32.669 --> 00:12:35.330
the electricity meter. Failure is just part of

00:12:35.330 --> 00:12:37.730
the process, a necessary step in discovery, not

00:12:37.730 --> 00:12:40.090
a potential financial catastrophe. Right. So

00:12:40.090 --> 00:12:42.409
a researcher there might use a powerful local

00:12:42.409 --> 00:12:46.149
model, say Baidu's Erniebot, for something truly

00:12:46.149 --> 00:12:48.850
ambitious, like maybe prompting it to generate

00:12:48.850 --> 00:12:52.009
an entire 3D virtual world from a classic novel.

00:12:52.250 --> 00:12:54.809
Simulate 400 characters with unique personalities,

00:12:55.429 --> 00:12:57.789
create storylines that emerge on their own, even

00:12:57.789 --> 00:13:00.610
try to recreate smells and textures. The focus

00:13:00.610 --> 00:13:05.190
there is just scale, ambition. When energy isn't

00:13:05.190 --> 00:13:07.470
the bottleneck, when it's just there, the only

00:13:07.470 --> 00:13:10.610
real limit left is imagination. Slight pause,

00:13:10.950 --> 00:13:13.509
tone of wonder. Whoa. Seriously, imagine that.

00:13:13.600 --> 00:13:15.899
Creating a whole virtual world smells and all

00:13:15.899 --> 00:13:18.100
just because you can. Because the power bill

00:13:18.100 --> 00:13:21.039
isn't even a factor. It's like having infinite

00:13:21.039 --> 00:13:23.480
Lego bricks of data to stack. What does that

00:13:23.480 --> 00:13:25.480
unlock? It unlocks a huge amount. It means they

00:13:25.480 --> 00:13:27.820
can often just brute force problems, try things

00:13:27.820 --> 00:13:30.139
out, push boundaries in ways their Western counterparts

00:13:30.139 --> 00:13:32.700
have to find clever, often smaller scale workarounds

00:13:32.700 --> 00:13:35.539
for because of those energy limits. So energy

00:13:35.539 --> 00:13:37.759
constraints directly limit the scope of innovation.

00:13:38.059 --> 00:13:39.980
Absolutely. It fundamentally shifts the focus

00:13:39.980 --> 00:13:42.240
from pure discovery towards optimization and

00:13:42.240 --> 00:13:45.330
making do. So this whole AI race, maybe without

00:13:45.330 --> 00:13:47.710
anyone really intending it, it's become this

00:13:47.710 --> 00:13:49.909
massive stress test for different national development

00:13:49.909 --> 00:13:52.570
models. And the results so far, especially looking

00:13:52.570 --> 00:13:55.470
at the US, well, they show a system facing some

00:13:55.470 --> 00:13:58.039
serious challenges. failing in some key ways.

00:13:58.519 --> 00:14:00.879
Yeah, and it's crucial to say the issue isn't

00:14:00.879 --> 00:14:04.299
a lack of brilliant people or innovative ideas

00:14:04.299 --> 00:14:07.179
in Silicon Valley, right? The talent is there.

00:14:07.179 --> 00:14:09.360
It's the foundation they're trying to build on.

00:14:09.539 --> 00:14:12.179
It's, as someone put it, a decaying physical

00:14:12.179 --> 00:14:15.460
foundation. We've sort of seen this pattern play

00:14:15.460 --> 00:14:17.679
out before, haven't we? US companies getting

00:14:17.679 --> 00:14:20.620
really good at optimizing for individual wealth

00:14:20.620 --> 00:14:23.600
extraction, while competitors abroad focus on

00:14:23.600 --> 00:14:25.879
building systemic advantages. Right, we tend

00:14:25.879 --> 00:14:28.919
to sell things like tax cuts, deregulation, focusing

00:14:28.919 --> 00:14:31.360
on maybe short -term boosts, while China is making

00:14:31.360 --> 00:14:34.039
these huge long -term bets on collective capability,

00:14:34.620 --> 00:14:37.820
beating the system itself. And this energy situation,

00:14:37.980 --> 00:14:41.720
this AI race, it's really exposed to truth that

00:14:41.720 --> 00:14:44.960
maybe many Americans don't want to face, that

00:14:44.960 --> 00:14:48.519
maybe our entire model is broken in some fundamental

00:14:48.519 --> 00:14:52.169
ways. It's hard to compete long -term with nations

00:14:52.169 --> 00:14:54.870
that can actually plan beyond the next election

00:14:54.870 --> 00:14:57.669
cycle that can invest strategically for decades.

00:14:57.870 --> 00:14:59.789
The grid tells the story, doesn't it? It doesn't

00:14:59.789 --> 00:15:03.409
lie. It shows that decades of maybe short -sightedness,

00:15:03.649 --> 00:15:05.629
prioritizing immediate profits have left the

00:15:05.629 --> 00:15:07.990
nation struggling to power its own future innovations.

00:15:08.690 --> 00:15:10.850
China saw this coming, and they basically solved

00:15:10.850 --> 00:15:13.789
their energy supply problem years ago, strategically.

00:15:13.830 --> 00:15:16.240
And here. Well, here we're sometimes still arguing

00:15:16.240 --> 00:15:18.940
if spending money on basic infrastructure, something

00:15:18.940 --> 00:15:22.100
everyone needs, is somehow socialism. It feels

00:15:22.100 --> 00:15:23.639
like we're stuck in the wrong debate sometimes.

00:15:23.740 --> 00:15:25.759
We really could be watching the beginning of

00:15:25.759 --> 00:15:28.379
a major shift in technological power, a shift

00:15:28.379 --> 00:15:30.580
that's going to define the next hundred years.

00:15:30.919 --> 00:15:34.039
And unless America really changes course, fundamentally

00:15:34.039 --> 00:15:37.120
rethinks how it balances that immediate individual

00:15:37.120 --> 00:15:39.299
game against long -term collective investment,

00:15:39.840 --> 00:15:41.240
we're likely going to keep falling behind in

00:15:41.240 --> 00:15:43.639
these crucial future races. Yeah, the big question

00:15:43.639 --> 00:15:45.519
isn't really, can we catch up anymore? It might

00:15:45.519 --> 00:15:49.179
be. Are we even capable of trying when the system

00:15:49.179 --> 00:15:52.039
itself seems to reward the very behaviors that

00:15:52.039 --> 00:15:54.840
got us into this situation? Does this mean the

00:15:54.840 --> 00:15:57.399
core system needs reevaluation, not just tweaks?

00:15:57.799 --> 00:16:00.659
Yes. I think it points towards needing a fundamental

00:16:00.659 --> 00:16:03.299
shift in our national priorities. So the big

00:16:03.299 --> 00:16:07.730
idea here, crystal clear, the AI race. It's not

00:16:07.730 --> 00:16:09.529
just about the smartest software or the fastest

00:16:09.529 --> 00:16:12.409
hardware, not just brilliant engineers. It is

00:16:12.409 --> 00:16:14.590
profoundly shaped by something much more basic,

00:16:15.269 --> 00:16:17.509
the energy infrastructure humming or sputtering.

00:16:17.820 --> 00:16:20.720
beneath it all. Yeah, China's big bet on energy

00:16:20.720 --> 00:16:22.840
abundance, it creates this playground for almost

00:16:22.840 --> 00:16:26.000
unlimited AI ambition. Scale, experimentation,

00:16:26.279 --> 00:16:29.039
go nuts. Meanwhile, the US, with its older, more

00:16:29.039 --> 00:16:31.500
fragile grid, forces its innovators to constantly

00:16:31.500 --> 00:16:33.519
think about efficiency, about conserving power.

00:16:33.860 --> 00:16:35.860
Which, you know, channels their amazing brain

00:16:35.860 --> 00:16:38.269
power differently. And that difference, that

00:16:38.269 --> 00:16:41.049
disparity, it really shines a light on two very

00:16:41.049 --> 00:16:43.029
different national philosophies, doesn't it?

00:16:43.250 --> 00:16:46.210
Short -term individual profits versus long -term

00:16:46.210 --> 00:16:49.049
collective strength and planning ahead. So as

00:16:49.049 --> 00:16:51.169
you go about your day today, maybe chew on this

00:16:51.169 --> 00:16:53.830
thought. What if the most important innovation

00:16:53.830 --> 00:16:56.409
we need right now isn't actually a new algorithm?

00:16:56.620 --> 00:17:00.200
What if it's a new way for us as a society to

00:17:00.200 --> 00:17:03.399
prioritize the long -term collective good over

00:17:03.399 --> 00:17:05.819
that immediate short -term individual gain? Yeah,

00:17:05.819 --> 00:17:07.480
think about how the choice is being made about

00:17:07.480 --> 00:17:09.420
energy right now or maybe the choice is not being

00:17:09.420 --> 00:17:12.539
made how they're shaping Not just our tech future,

00:17:12.640 --> 00:17:14.940
but our everyday lives for years and years to

00:17:14.940 --> 00:17:16.740
come. Thank you for joining us for this deep

00:17:16.740 --> 00:17:19.480
dive We really hope this gave you some fresh

00:17:19.480 --> 00:17:21.460
perspective something new to consist until next

00:17:21.460 --> 00:17:23.920
time outro music