Latest / a16z AI Policy Brief / AI is How America Builds Again

Transcript

AI is How America Builds Again · a16z AI Policy Brief

0:00
SPEAKER_01: Welcome back to the AI Policy Brief.
0:02
SPEAKER_01: Today,
0:03
SPEAKER_01: Erin Price-Wright,
0:03
SPEAKER_01: General Partner on our American Dynamism team,
0:06
SPEAKER_01: joins the podcast to discuss how AI is reshaping the physical world,
0:09
SPEAKER_01: from manufacturing and energy to defense and robotics.
0:12
Our conversation centers on how AI can help us build things
0:15
that we couldn't build before.
0:17
SPEAKER_01: She talks about what that looks like in practice,
0:19
SPEAKER_01: building factories,
0:20
SPEAKER_01: improving mining,
0:21
SPEAKER_01: making automation more accessible to more people,
0:23
SPEAKER_01: and bringing more advanced manufacturing back to the United States.
0:27
SPEAKER_01: Erin also talks about what the war in Ukraine teaches us about AI.
0:30
SPEAKER_01: and about how data centers can lower the cost of power in the long term.
0:33
SPEAKER_01: She argues that if we get the policy environment right,
0:36
AI will create new kinds of jobs and strengthen the industrial
0:39
base needed for the
0:40
SPEAKER_01: next era of technology.
0:41
SPEAKER_01: Enjoy the conversation.
0:42
SPEAKER_02: Aaron, welcome to the podcast.
0:50
SPEAKER_00: Thanks.
0:51
SPEAKER_00: It's so fun to be here.
0:52
SPEAKER_02: It's great to have you on.
0:53
SPEAKER_02: So first things first, can you talk a little bit about what you do?
0:58
SPEAKER_00: I am,
0:59
SPEAKER_00: Erin,
1:00
SPEAKER_00: I am one of the general partners on our American Dynamism team,
1:04
SPEAKER_00: which means,
1:05
SPEAKER_00: you know,
1:05
SPEAKER_00: I invest in companies that are supporting the interests of the U.S.
1:10
SPEAKER_00: and our allies.
1:10
SPEAKER_00: So we invest,
1:11
SPEAKER_00: you know,
1:12
something like half of our portfolio companies sell directly
1:15
into the government,
1:16
SPEAKER_00: primarily Department of War,
1:17
SPEAKER_00: but also public safety,
1:18
SPEAKER_00: GovTech more generally.
1:20
And then the other half of our companies are really supporting
1:22
the build out of the
1:23
SPEAKER_00: broader Western industrial base.
1:24
SPEAKER_00: So how do we bring the latest tech into problems like energy,
1:28
SPEAKER_00: manufacturing,
1:29
SPEAKER_00: supply chain,
1:30
SPEAKER_00: logistics,
1:32
SPEAKER_00: aerospace,
1:34
SPEAKER_00: and,
1:34
SPEAKER_00: you know,
1:35
many other areas that affect kind of the day-to-day lives of
1:39
Americans in this
1:40
SPEAKER_00: country.
1:41
SPEAKER_02: So there's a ton in there.
1:43
SPEAKER_02: What's the unifying idea?
1:44
SPEAKER_02: Like, how do you get from energy stuff to defense to GovTech?
1:48
SPEAKER_02: Like, you listed a bunch of different things.
1:50
SPEAKER_02: What's the unifying concept?
1:51
SPEAKER_00: Yeah, totally.
1:52
So I think,
1:53
you know,
1:54
all of us on the team probably think about it a little bit
1:57
differently.
1:58
So I spent my career before investing at Palantir,
2:01
and I'm just really interested in taking...
2:04
SPEAKER_00: Very hard,
2:05
SPEAKER_00: complicated,
2:06
messy problems that we've been solving one way for a really
2:08
long time and figuring
2:10
out how to bring tech and in particular AI to thinking about
2:15
how we might do those
2:16
SPEAKER_00: things differently.
2:18
And so I think the intersection of the AI and the physical
2:20
world has been my lens
2:22
SPEAKER_00: for how I think about a lot of my investing here.
2:26
um it's a really important and interesting moment in AI it's
2:29
also a really
2:30
important and interesting moment in the kind of
2:32
re-industrialization of the United
2:34
States and these two waves are coming together at sort of just
2:37
the right time and I
2:39
say like I joke like I don't know if I am a VC because I'm an
2:43
optimist or if I'm an
2:44
optimist because I'm a VC and I just listen to all these like
2:47
really exciting
2:47
SPEAKER_00: founder pitches all day long but I think uh
2:51
SPEAKER_00: AI Policy Brief a16z
3:13
I'll give you one example that I love because I think it sort
3:16
of captures a pretty
3:17
SPEAKER_00: broad spectrum of what we're talking about here.
3:21
SPEAKER_00: We have a portfolio company called Mariana Minerals.
3:24
SPEAKER_00: It's founded by a former longtime Tesla engineer.
3:28
He was at Tesla for a decade working across Gigafactory and
3:34
building out the
3:35
lithium plant in Corpus Christi and the battery metals and
3:38
mineral supply chain.
3:39
did lots of really important things working very closely with
3:42
Elon Musk and he
3:43
recognized which has you know it's been a topic of conversation
3:47
over the last year
3:48
SPEAKER_00: that we are dangerously dependent on China
3:53
for our critical minerals and materials that are important for
3:57
powering economy and
3:58
SPEAKER_00: not just our defense base,
4:00
SPEAKER_00: but our broader industrial base.
4:01
SPEAKER_00: So the materials that go into the things that we use every single day,
4:06
SPEAKER_00: you know,
4:06
60 something percent of them are probably mined in China and 90
4:10
something percent
4:11
SPEAKER_00: of them are refined in China.
4:13
SPEAKER_00: So all roads kind of lead through China when it comes to things like
4:16
SPEAKER_00: AI Policy Brief a16z
4:39
and he took his experience at Tesla which is this real factory
4:42
mindset like how do
4:43
we how do we construct everything as a factory and is bringing
4:46
it to verticalized
4:48
mining and refining so the way he's thinking about it is like
4:52
how do we deconstruct
4:54
the process of what it means to build and operate a mine and
4:57
refinery into its kind
4:58
of core components and understand how to turn those core
5:02
components into like mini
5:04
SPEAKER_00: factories and
5:05
using technology and how do we basically like inject technology
5:08
into every single
5:10
process of owning and operating a mine which is just not how
5:15
current mining
5:16
SPEAKER_00: companies work so everything from how do you use
5:21
LLMs to basically handle a lot of the complicated coordination
5:27
between all of the
5:28
SPEAKER_00: different parties that you have operating at a mine.
5:31
SPEAKER_00: You have EPCs,
5:33
SPEAKER_00: you have lots of different types of contractors,
5:36
SPEAKER_00: you have internal employees,
5:38
SPEAKER_00: you have folks working in the pit,
5:39
SPEAKER_00: you have technologists and geologists.
5:42
All of these people have kind of complicated and overlapping
5:45
responsibilities.
5:46
SPEAKER_00: And one of the big challenges,
5:47
like one of the reasons why it takes 20 years to permit and
5:51
build a new mine today
5:52
is it's just a lot of different people to coordinate across
5:56
with regulators,
5:57
SPEAKER_00: etc.
5:58
So they've come and said,
6:00
okay,
6:00
well,
6:00
how do we use AI to make this problem better?
6:04
How do we use AI to make sure that we have all the data we need
6:08
when we need it?
6:09
SPEAKER_00: How do we use AI to make sure we're
6:11
Quickly Responding to all of the requests that are coming from
6:14
regulators on
6:15
SPEAKER_00: various parts of the permitting process.
6:17
SPEAKER_00: How do we use AI to make sure all of these complicated contractor groups,
6:24
SPEAKER_00: you know,
6:24
SPEAKER_00: who have sort of overlapping needs are communicating effectively.
6:27
SPEAKER_00: And when something updates in one part of the plan,
6:31
SPEAKER_00: it sort of flows down through to every other part of the plan.
6:34
SPEAKER_00: AI Policy Brief a16z
6:59
SPEAKER_00: And then all the way down to the refining.
7:00
SPEAKER_00: Like,
7:01
how do we use the latest in reinforcement learning and control
7:04
systems to actually
7:05
SPEAKER_00: control the chemistry for the,
7:08
SPEAKER_00: you know,
7:08
SPEAKER_00: for how we're extracting copper from,
7:12
SPEAKER_00: you know,
7:12
SPEAKER_00: the slush that comes out of a copper mine.
7:15
SPEAKER_00: And those are really expensive chemicals.
7:17
SPEAKER_00: That's, like, mainly the driver of cost in mines.
7:19
So if we can use AI to do all that better,
7:22
you know,
7:22
you operate your mine more efficiently.
7:25
So I know that's,
7:25
like,
7:26
a very long answer,
7:27
but I think it just goes to show
7:29
if you're doing something complicated and physical we're
7:31
talking about interjecting
7:33
AI at basically every single stage of that business and turning
7:36
it from something
7:37
that's like not economical to do into something that's very
7:41
economical to do and
7:43
SPEAKER_00: doing it more safely more cleanly and just like overall better
7:47
SPEAKER_02: so is that then the answer to why this time is different like the idea of
7:52
reinvigorating America's industrial base reinvigorating
7:56
American manufacturing
7:58
that's not a new concept we've been trying to do that for a
8:00
long period of time
8:01
what why now why why is this different it's a really good
8:04
question I think it's a
8:05
conflagration of both sort of geopolitical considerations and
8:10
tech advances so
8:13
SPEAKER_00: The Ukraine war has been a wake-up call for the United States.
8:17
SPEAKER_00: And if you go on the ground and you see the amount of low-cost,
8:24
SPEAKER_00: attritable mass being used to conduct a large-scale warfare...
8:29
SPEAKER_02: What does that mean?
8:30
SPEAKER_02: What do you mean?
8:32
SPEAKER_00: We're talking about,
8:33
SPEAKER_00: you know,
8:34
the United States develops weapon systems that are millions and
8:38
tens of millions
8:39
SPEAKER_00: and sometimes hundreds of millions of dollars.
8:41
SPEAKER_00: We are really good at developing like these beautiful exquisite systems,
8:46
SPEAKER_00: you know,
8:46
where we have this exquisite supply chain of mom and pop shops
8:50
who have been making
8:51
SPEAKER_00: this specialty part for 50 years.
8:55
SPEAKER_00: It's perfectly engineered.
8:57
SPEAKER_00: It goes into this, you know, beautiful missile or plane or system.
9:02
And we have a handful of them and they are very good at the
9:06
specific things they do.
9:08
Where we've been,
9:10
like,
9:11
what we've been confronted with in Ukraine is,
9:14
you know,
9:14
a...
9:16
SPEAKER_00: $500 drone, or maybe a $40,000 drone, let's say.
9:20
SPEAKER_00: And this is a system that is manufactured quickly with cheap parts.
9:26
SPEAKER_00: The design is changing rapidly based on feedback from the field.
9:30
SPEAKER_00: You know, you can, it doesn't matter if you never get it back.
9:33
SPEAKER_00: Like, if you blow up the drone, who cares?
9:35
SPEAKER_00: Because it's a cheap drone and you'll just build a new one.
9:38
SPEAKER_00: And we just,
9:39
SPEAKER_00: I think,
9:40
slowly have realized that we actually don't have the
9:43
manufacturing base to support
9:46
SPEAKER_00: that type of warfare.
9:48
And I say,
9:49
like,
9:49
you don't have a defense industrial base if you don't have an
9:51
industrial base.
9:52
SPEAKER_00: If you're not building things, you can't build a lot of military things.
9:56
SPEAKER_00: You're just, like, you can't build full stop.
9:58
SPEAKER_00: You can't just spin up,
10:00
SPEAKER_00: you know,
10:01
a defense base without having this know-how and process
10:03
knowledge in the economy.
10:05
SPEAKER_00: And so I think that it's been a real wake-up call.
10:08
SPEAKER_00: And it's,
10:08
SPEAKER_00: you know,
10:09
SPEAKER_00: like,
10:09
one of the stories that I'm sure people have heard before
10:11
that's come out a bunch
10:12
SPEAKER_00: is a few years ago.
10:14
Ukraine was buying their drones from China so they were buying
10:18
DJI drones and
10:20
obviously China and Russia are very aligned so China was giving
10:23
Russia this
10:24
SPEAKER_00: backdoor
10:25
to be able to figure out where the drone controllers were
10:29
located who were
10:30
SPEAKER_00: operating DJI drones.
10:32
And so Ukrainian soldiers were getting killed because they were
10:34
using Chinese drones.
10:36
SPEAKER_00: And so I think it was a big wake-up call for Ukraine to realize,
10:39
SPEAKER_00: oh,
10:39
SPEAKER_00: wow,
10:39
SPEAKER_00: we actually have to be able to produce this stuff.
10:41
We need to be able to build things at scale to be able to fight
10:43
this conflict.
10:44
and I think watching that from the United States and not even
10:47
being in a strong
10:48
position to be able to help and support because we don't have
10:50
the kind of a
10:51
treatable mass or the know-how for how to build a treatable
10:54
mass that they need was
10:55
like a big wake-up call for us and that combined with the
10:59
rising geopolitical
11:00
tension around China and the realization that if you pull our
11:04
supply chains back
11:05
for essentially everything we use in the United States all
11:08
roads lead to China so
11:10
SPEAKER_00: the idea of building
11:12
SPEAKER_00: Fully U.S.
11:13
SPEAKER_00: Manufactured Drone with U.S.
11:15
SPEAKER_00: Manufactured PCB Boards and Controllers, U.S.
11:18
SPEAKER_00: Manufactured Brushless Motors, U.S.
11:21
SPEAKER_00: Manufactured Cameras and Optical Systems, U.S.
11:24
SPEAKER_00: Manufactured Sensors, or not even U.S.
11:26
Manufactured,
11:27
but just Allied Manufactured,
11:30
you know,
11:30
the Rare Earth Magnets.
11:32
All of these components,
11:33
once you start going up the supply chain,
11:35
are all coming from China.
11:36
SPEAKER_00: So there's this big moment in
11:38
and large appetite to figure out how to bring more of that back
11:42
and reshore it in the US.
11:43
And frankly,
11:43
there's no universe where we can do that without heavily
11:47
relying on AI.
11:48
SPEAKER_02: So that makes a ton of sense in terms of the sense of urgency.
11:53
It sounds like what you're suggesting is a really massive shift
11:56
in how we produce
11:58
SPEAKER_02: and what we produce and how government procures.
12:02
SPEAKER_02: Are you optimistic about the pace of that shift as well?
12:05
What are you seeing in terms of how responsive the economy and
12:09
the government is to
12:09
SPEAKER_02: this sense of need?
12:11
SPEAKER_00: Sure.
12:11
SPEAKER_00: Well, on the government side, I think there's been a ton of progress.
12:14
SPEAKER_00: We've seen a lot of reform in how procurement works.
12:18
SPEAKER_00: There's extreme appetite,
12:20
very bipartisan appetite to figure out how to modernize and
12:23
reinvigorate the
12:24
SPEAKER_00: industrial base and the defense industrial base.
12:27
SPEAKER_00: So I think things are moving.
12:28
SPEAKER_00: Obviously,
12:29
SPEAKER_00: there's probably more we can do,
12:30
SPEAKER_00: and I'm certain that we could move faster,
12:33
SPEAKER_00: but I'm actually optimistic about the way things are going there.
12:37
SPEAKER_00: And on the technology side,
12:38
SPEAKER_00: you know,
12:39
SPEAKER_00: we have a huge technological advantage with how strong we are in AI.
12:45
And I think that's going to be really critical to figuring out
12:49
how to actually make this happen.
12:51
SPEAKER_00: Because economically,
12:52
SPEAKER_00: financially,
12:53
there's just no world where we can compete on the labor
12:56
economics of China.
12:59
So we're going to need to figure out how to do all this using
13:02
technology and technology.
13:05
and software and the United States is very good at software and
13:08
very good at AI.
13:10
So it's like pretty critical that we figure out how to merge
13:12
those two worlds together.
13:13
SPEAKER_00: So I am optimistic.
13:14
SPEAKER_00: There's a lot of work to do.
13:16
SPEAKER_00: It's not going to be an easy problem to solve,
13:18
but I'm heartened by the fact that so many founders and
13:22
technologists are focused
13:23
SPEAKER_00: on this problem,
13:24
SPEAKER_00: which was like the unsexy stepchild for several decades.
13:28
SPEAKER_00: Like it was just very much not cool to go into these fields and
13:32
SPEAKER_00: Like physical manufacturing,
13:34
SPEAKER_00: even things like electrical engineering and mechanical engineering.
13:37
SPEAKER_00: It was just like not cool to go then work as a mechanical engineer.
13:42
So I'm very heartened by the fact that like the American
13:45
dynamism is sexy.
13:47
SPEAKER_00: I think it's a really good sign.
13:48
SPEAKER_02: So speaking of unsexy,
13:50
I think some people would say there's nothing less sexy than
13:53
procurement reform.
13:53
SPEAKER_02: But I think for us...
13:55
SPEAKER_00: I disagree.
13:55
SPEAKER_00: Disagree.
13:56
SPEAKER_00: I think it's very sexy.
13:57
SPEAKER_00: Yeah.
13:59
SPEAKER_02: For us, it's critical.
14:01
SPEAKER_02: And I think as I've worked more and more with our portfolio,
14:04
I've really understood how much selling to the government is
14:06
such a critical market
14:07
SPEAKER_02: for them.
14:08
SPEAKER_02: It's critical beyond just the revenue.
14:10
It's important to build those relationships with the government
14:12
and the policymakers.
14:16
SPEAKER_02: So when you talk about reform that is important for little tech,
14:20
SPEAKER_02: what are some of the main components that we're really advocating for?
14:25
Yes,
14:26
so I'm specifically talking about procurement within the
14:30
Department of War.
14:32
SPEAKER_00: And,
14:32
SPEAKER_00: like,
14:33
SPEAKER_00: broadly speaking,
14:34
we have to figure out how to level the playing field between
14:37
the large primes who
14:39
have these existing books of business with these very large
14:43
programs of record and
14:45
SPEAKER_00: a very set way of how they've historically worked with the government,
14:48
SPEAKER_00: which is cost-plus models where they get a very,
14:51
SPEAKER_00: very specific list of not requirements even,
14:54
SPEAKER_00: but,
14:54
SPEAKER_00: like,
14:55
SPEAKER_00: features and specs for how a particular thing has to function.
14:59
SPEAKER_00: AI Policy Brief a16z
15:26
It incentivizes being slow and being inefficient because you
15:32
essentially you
15:33
essentially take the cost that it took you to build something
15:35
and like you add a
15:36
margin and so what we've been really fighting for alongside
15:39
many others in the kind
15:40
of defense landscape is reforming that and making and
15:44
incentivizing the US
15:45
SPEAKER_00: government to buy products
15:48
SPEAKER_00: AI Policy Brief a16z
16:18
SPEAKER_00: You know, it's a really big ship to turn.
16:21
SPEAKER_00: So I think there has been broad alignment
16:25
by folks in the Department of War and by lawmakers that
16:29
procurement has to change.
16:31
SPEAKER_00: I mean,
16:31
SPEAKER_00: there's been broad alignment around this for,
16:33
SPEAKER_00: I'd say,
16:34
SPEAKER_00: like,
16:35
SPEAKER_00: well over a decade,
16:36
SPEAKER_00: you know,
16:36
SPEAKER_00: since actually probably several decades.
16:40
But it's a really complicated machine to fix,
16:44
and it's really hard to change those incentives.
16:47
SPEAKER_00: Like, this is complicated legislation.
16:49
So we have seen some great wins in the second half of last year
16:52
on NDAA reform
16:55
SPEAKER_00: It's going to be a long road, but we're definitely on it.
16:57
SPEAKER_00: And it's a very exciting moment because it's also a time when we have...
17:15
All of these mission-oriented founders who want to build things
17:20
for the national
17:21
SPEAKER_00: interest,
17:22
SPEAKER_00: you know,
17:22
who have grown up at companies like SpaceX and Andurl and
17:26
Palantir who have really
17:27
SPEAKER_00: helped pave the way for a lot of this reform and,
17:31
SPEAKER_00: you know,
17:31
have incredible ideas about products that they could build that
17:35
would serve the
17:35
SPEAKER_00: warfighter quickly in with,
17:39
SPEAKER_00: you know,
17:39
SPEAKER_00: the latest and greatest in autonomy technology,
17:42
SPEAKER_00: etc., that they could get to market really fast.
17:45
SPEAKER_00: And it finally feels like,
17:47
SPEAKER_00: you know,
17:47
we're starting to see a willing buyer on the other side of that
17:50
who knows how to
17:51
SPEAKER_00: work with startups and is interested in,
17:54
SPEAKER_00: you know,
17:54
SPEAKER_00: bringing new tech into our armed forces.
17:58
SPEAKER_02: So you're in touch with these companies every day.
18:00
SPEAKER_02: And then a few weeks ago,
18:01
SPEAKER_02: we had our American Dynamism Summit,
18:03
SPEAKER_02: which is focused on trying to increase this connective tissue,
18:06
SPEAKER_02: create opportunities for our portfolio to sell with the government.
18:09
SPEAKER_02: But you also said it's a slow moving ship.
18:11
SPEAKER_02: Are you starting to see the space for opportunity expand?
18:24
SPEAKER_00: So Palantir, I'll start with Palantir as an example.
18:26
SPEAKER_00: So Palantir,
18:28
SPEAKER_00: I might get these specific numbers roughly wrong,
18:31
SPEAKER_00: but,
18:31
SPEAKER_00: you know,
18:31
SPEAKER_00: Palantir took over a decade to get its first,
18:35
SPEAKER_00: you know,
18:35
SPEAKER_00: $100 million program of record contract with the U.S.
18:40
SPEAKER_00: military.
18:41
SPEAKER_00: Anduril,
18:43
SPEAKER_00: you know,
18:43
SPEAKER_00: several of the founders from Anduril come out of Palantir,
18:47
SPEAKER_00: took about four years to do the same thing.
18:50
SPEAKER_00: And Saronic, which is founded by alumni from Anduril,
18:54
SPEAKER_00: A16z
19:24
SPEAKER_00: They're hearing their POCs come back to them and say,
19:27
SPEAKER_00: okay,
19:27
SPEAKER_00: I know we talked about this list of requirements.
19:32
SPEAKER_00: How much faster could you move if we cut that list by 75%?
19:37
SPEAKER_00: Like, how long would it take you to build 1,000?
19:40
SPEAKER_00: How long would it take you to build 100,000?
19:43
So that level of urgency and intensity feels like it's only
19:49
increasing and will
19:50
SPEAKER_00: likely only increase for the foreseeable future.
19:54
SPEAKER_00: as I think the geopolitical situation globally intensifies.
19:59
So you've talked about geopolitical situation and how it
20:02
creates a sense of
20:03
SPEAKER_02: urgency,
20:03
SPEAKER_02: how it's intensifying.
20:06
You alluded a little bit to the changes that AI is bringing on
20:10
the company side in
20:11
SPEAKER_02: terms of like shifting the technology to respond to the moment.
20:14
SPEAKER_02: What are some examples of what you're seeing there?
20:16
SPEAKER_02: How is AI really positioning companies different to meet the moment?
20:20
SPEAKER_00: Yeah, well, you know,
20:22
I think like AI is bringing this kind of like allows you to
20:28
bring sort of a factory
20:29
SPEAKER_00: mindset to really complicated problems like like
20:34
SPEAKER_00: If you can even think about the need to build a factory of factories,
20:40
like suddenly we're in a position where it's not like we have a
20:43
decade to sort of
20:44
SPEAKER_00: build and plan one exquisite factory.
20:47
SPEAKER_00: We're talking about needing to build hundreds of factories,
20:50
SPEAKER_00: which,
20:50
SPEAKER_00: you know,
20:50
SPEAKER_00: that starts to sound like a factory itself.
20:53
And AI is really uniquely positioned to be able to take a lot
20:57
of the complexity and
20:58
the sort of like complex systems thinking that goes into how do
21:01
you build a factory
21:03
and really simplify it so if you think about what a factory is
21:06
you know it's you
21:08
you take a complex system and you break it down into kind of
21:12
its component parts
21:14
SPEAKER_00: and then within those component parts you automate and then you iterate
21:21
And that kind of deconstructing something down into its
21:24
components and automating,
21:26
SPEAKER_00: that's something AI is really good at.
21:27
So we're seeing companies both within the defense space and
21:31
outside of the defense
21:32
space or maybe adjacent to the defense space take a lot of
21:35
lessons from AI to like
21:38
SPEAKER_00: how they're building and planning factories.
21:40
SPEAKER_00: So, you know, I'll give you like a couple of examples.
21:44
SPEAKER_00: The process of site selection is kind of an art process.
21:49
SPEAKER_00: Like the sort of real estate deal making,
21:51
SPEAKER_00: you know,
21:52
talking with local politicians and how do you decide where you
21:58
set up your factory
21:59
SPEAKER_00: is like a very complicated problem.
22:01
SPEAKER_00: And historically, you might hire a consultancy
22:05
SPEAKER_00: and you'd spend,
22:06
SPEAKER_00: you know,
22:06
millions of dollars with them and they would go do this whole
22:10
survey and they would
22:11
engage all these political leaders and they would look at
22:13
things like land
22:14
availability and power availability and environmental
22:18
regulation and all these
22:20
other kind of things and then come back with a proposal several
22:24
years later.
22:25
SPEAKER_00: With AI, you can do things like that extremely quickly.
22:28
SPEAKER_00: So you can do,
22:29
SPEAKER_00: like,
22:29
SPEAKER_00: large-scale surveys of power availability,
22:33
SPEAKER_00: land availability,
22:34
SPEAKER_00: regulatory regime and framework,
22:36
SPEAKER_00: you know,
22:37
permitting regimes in place and pretty quickly hone in on a
22:42
very small number of
22:43
SPEAKER_00: sites that you can immediately engage local lawmakers to say,
22:46
SPEAKER_00: like,
22:47
SPEAKER_00: what would it look like to build,
22:48
SPEAKER_00: you know,
22:49
SPEAKER_00: an energetics factory here?
22:51
SPEAKER_00: Or would you be interested in us building out a drone factory here?
22:55
That takes a multi-year and many millions of dollar process and
22:58
lets you do that in
22:59
SPEAKER_00: a month,
23:00
SPEAKER_00: which is a pretty powerful tech advantage.
23:05
SPEAKER_00: Or how do you go through the process of permitting?
23:10
SPEAKER_00: Most regulations, local regulations, are there for a reason.
23:13
SPEAKER_00: They're probably reasonable and good regulations.
23:17
SPEAKER_00: We want them to exist.
23:18
SPEAKER_00: It's important that regulations exist.
23:21
but the process of like going back and forth with permitting
23:25
bodies and regulators
23:27
as someone who wants to build a factory can be a true nightmare
23:31
like we look at you
23:31
know I look at one of our companies Radiant and they're a
23:35
nuclear company so
23:36
they're going back and forth with various nuclear regulators
23:39
and every time they
23:41
SPEAKER_00: have to submit their paperwork it's like
23:44
Tens of thousands of pages and like if you have like a comma in
23:48
the wrong place you
23:49
know it might just get kicked back to you so part of that is an
23:52
alignment problem
23:53
it's like how do you make sure you're on the same page with
23:55
your regulators where
23:56
SPEAKER_00: like everybody is incentivized to figure out how to make this work
24:00
SPEAKER_00: versus figure out how to say no.
24:02
And part of it is,
24:02
you know,
24:03
a problem that's really well solved with LLMs.
24:05
SPEAKER_00: Like, paperwork is a great LLM problem.
24:09
So all regulators should have AI tools and all companies that
24:12
are dealing with
24:13
SPEAKER_00: permits and regulators should also have AI tools.
24:17
SPEAKER_00: Yeah,
24:17
SPEAKER_00: I mean,
24:18
SPEAKER_00: I could go on and on and on,
24:19
SPEAKER_00: but it's just like every single tiny step of like,
24:22
how do you build a factory is a problem that AI is actually
24:26
really well suited to
24:27
SPEAKER_00: help speed up.
24:28
SPEAKER_00: So then you start thinking about,
24:29
SPEAKER_00: okay,
24:29
SPEAKER_00: like if it takes,
24:30
SPEAKER_00: you know,
24:30
SPEAKER_00: historically has taken five to 10 years to permit,
24:34
SPEAKER_00: build and commission a factory,
24:35
SPEAKER_00: like in the new world of AI,
24:37
SPEAKER_00: that should take 12 to 18 months.
24:39
SPEAKER_00: And that's what we're seeing with our companies.
24:41
SPEAKER_00: They're able to build a lot faster.
24:43
So you've talked to me a couple times about a factory mindset
24:45
and when you say
24:46
SPEAKER_02: factory mindset and AI,
24:47
SPEAKER_02: I keep thinking about the next frontier,
24:49
SPEAKER_02: which I think we think will be robotics.
24:51
SPEAKER_02: Yeah.
24:51
Are you starting to see exciting companies emerge in the
24:54
robotics industry and are
24:55
SPEAKER_02: they part of what you see as this new factory mindset?
24:59
SPEAKER_00: Oh, for sure.
24:59
SPEAKER_00: So...
25:01
You know,
25:01
it's not like our,
25:03
like most factories have lots and lots of robots.
25:06
SPEAKER_00: It's,
25:06
SPEAKER_00: you know,
25:07
SPEAKER_00: if you go into a factory today,
25:09
the likelihood that you'll see some sort of a smart conveyor
25:12
belt system or like a
25:14
SPEAKER_00: robotic arm from someone like ABB is very,
25:17
SPEAKER_00: very high.
25:20
What's exciting about the next frontier of robotics is how do
25:23
you actually lower
25:24
SPEAKER_00: that cost of automation
25:27
Such that you can adopt automation in more parts of your
25:31
assembly line or your
25:33
SPEAKER_00: production process.
25:34
So if you think about a production process like a true factory
25:37
line is a series of modules.
25:40
SPEAKER_00: Like this was the Henry Ford realization.
25:42
SPEAKER_00: Like how do we deconstruct building a car into a
25:45
100 steps and every step,
25:48
you know,
25:48
is its own module in an assembly line.
25:52
And then we're going to put one person on every single step and
25:55
they're going to do
25:55
SPEAKER_00: that same step over and over again.
25:57
SPEAKER_00: So they get really good at that task,
26:00
SPEAKER_00: you know,
26:00
SPEAKER_00: and then the next person gets really good at the next task.
26:03
SPEAKER_00: So that's what a factory is.
26:05
SPEAKER_00: That's what an assembly line is.
26:06
SPEAKER_00: That was like the big, you know, realization in the 1920s that Ford had.
26:12
and so now when you go to a factory you usually see you know
26:15
it's still the same
26:16
thing you see a series of production steps that are kind of
26:19
broken out by task and
26:21
then some of those tasks the very expensive tasks are usually
26:26
already automated so
26:27
you'll probably have you know for some very complicated task
26:30
you're going to have
26:31
SPEAKER_00: an ABB arm you know putting a part somewhere screwing something in
26:36
SPEAKER_00: A16z
27:00
and then you have to actually program it and it's you have to
27:03
hire a system
27:04
integrator it's like this really outdated control system that
27:08
you have to like very
27:09
precisely program exactly the movement and there's usually
27:12
reasonably low tolerance
27:14
you know for for for any anything different so it knows exactly
27:18
what it's going to
27:19
do you program it to do exactly that and all in all to automate
27:23
a step on a
27:24
SPEAKER_00: manufacturing line is probably close to two hundred thousand dollars
27:29
and so then if you think about and like and that's probably
27:31
something you should
27:32
think about annually so then if you think about or every time
27:35
your production line
27:36
SPEAKER_00: changes so if you think about like the cost benefit analysis
27:41
SPEAKER_00: AI Policy Brief a16z
28:00
AI and Robotics and where we're going is it really dramatically
28:04
lowers the cost of
28:06
automating a module so you get lower cost less precise hardware
28:11
that has higher
28:12
fault tolerance because instead of these extremely precise
28:16
control systems you're
28:18
using robotic learning you're using vision models you're using
28:21
other types of like
28:22
SPEAKER_00: modern AI concepts to be able to learn how to do a task
28:27
SPEAKER_00: So that,
28:27
SPEAKER_00: you know,
28:27
SPEAKER_00: the cost of setting up isn't $200,000 So maybe now,
28:31
SPEAKER_00: today,
28:32
SPEAKER_00: the cost of automating a module on a factory line is $100,000 And sure,
28:37
you need to hire some very strong techs to take care of your
28:40
robotic fleet But
28:41
you're also changing the cost-benefit analysis of automation on
28:45
a factory line
28:46
and so I'm not talking about like replacing workers who are on
28:49
existing factory
28:51
lines that's usually not how manufacturers are thinking about
28:54
this they're usually
28:55
thinking we're thinking about it in the sense of like what
28:58
things could we not
29:00
SPEAKER_00: build in the U.S.
29:02
before that we could build now so where are things where we
29:07
actually can make like
29:08
a reasonable cost trade-off if a factory that historically
29:12
would have required us
29:13
SPEAKER_00: to hire 500 people to staff
29:15
you know now we can hire 250 people to staff and a hundred of
29:19
those jobs might be
29:20
much higher paying jobs because we're talking about robotic
29:23
techs quality techs you
29:25
know like robotic programmers and robotic engineers instead of
29:28
like someone just
29:29
screwing a bulletin but overall now suddenly something makes
29:33
economic sense that
29:34
didn't make economic sense before so that's how I would think
29:36
about autonomy and
29:38
SPEAKER_00: robotics like on factory floors and the real difference that it's making
29:42
You're presenting a really clear vision,
29:45
I think,
29:46
for this future world that we might have.
29:48
It sounds like there are lots of reasons that you're explaining
29:51
that we're taking steps.
29:51
SPEAKER_00: We're in it.
29:52
SPEAKER_00: It's not the future world that we might have.
29:54
SPEAKER_00: It's happening now.
29:56
SPEAKER_02: A world that is happening now,
29:57
SPEAKER_02: but the transition to it,
29:58
SPEAKER_02: it does seem like,
29:59
SPEAKER_02: as you described,
30:00
SPEAKER_02: it's like a slow-moving ship.
30:01
SPEAKER_02: And so I assume there are some tech barriers,
30:04
SPEAKER_02: there are some economic barriers,
30:05
SPEAKER_02: there are also policy barriers to fully realizing the world that you're
30:08
SPEAKER_02: articulating.
30:09
SPEAKER_02: When you talk to the companies that you work with,
30:12
what are they saying in terms of policy barriers or also policy
30:16
accelerators like
30:17
what are they seeing in the policy environment that is either
30:20
making it harder for
30:20
them to to achieve this vision or making it easier for them
30:25
yeah I mean it's full
30:27
stop fundamentally too hard and slow to build a new factory in
30:31
the US so that's one
30:33
thing it's certainly getting better and timelines are
30:35
compressing but as it stands
30:37
SPEAKER_00: today when you start looking at
30:39
SPEAKER_00: AI Policy Brief a16z
31:01
SPEAKER_00: You know,
31:02
SPEAKER_00: we don't manufacture those machines in the United States,
31:04
SPEAKER_00: so maybe some of them are manufactured in China,
31:07
SPEAKER_00: but a lot are manufactured in places like Korea and Taiwan and Germany,
31:11
and that's the importation of those machines is still a very
31:17
kind of politically
31:18
SPEAKER_00: complicated,
31:19
SPEAKER_00: like,
31:19
SPEAKER_00: it's a,
31:19
SPEAKER_00: you know,
31:20
SPEAKER_00: we have tariffs,
31:20
SPEAKER_00: it's,
31:21
SPEAKER_00: you know,
31:22
SPEAKER_00: there's still a lot of uncertainty and understanding,
31:24
SPEAKER_00: like,
31:24
SPEAKER_00: how to actually bring those machines as quickly as possible to the U.S.,
31:27
SPEAKER_00: how to get trained on them.
31:31
SPEAKER_00: And then if you look at the robotics and automation piece in particular,
31:37
the tricky thing now is that most of the robotic components
31:42
that you might use to
31:43
SPEAKER_00: bring autonomy to a factory line,
31:45
SPEAKER_00: those are all made in China.
31:47
SPEAKER_00: Like 99.9% of the robotic supply chain is China.
31:51
SPEAKER_00: So,
31:52
SPEAKER_00: you know,
31:53
there is no company in the US right now that's making actuators
31:57
at any kind of
31:57
SPEAKER_00: scale.
32:01
SPEAKER_00: Robotic Arms to Move.
32:02
Maybe you could buy one from Germany for like 10 times the cost
32:07
as you could buy one from China.
32:10
SPEAKER_00: And all of those components that go inside robotics are all,
32:13
SPEAKER_00: you know,
32:14
SPEAKER_00: fully Chinese supply chains.
32:15
So we have to figure out how to jumpstart our U.S.-based
32:19
robotic supply chain.
32:21
SPEAKER_00: That's really important, and we need industrial policy to do that.
32:24
SPEAKER_00: On the other hand,
32:26
SPEAKER_00: Aaron, what about AI and manufacturing specifically?
32:57
SPEAKER_00: Yeah,
32:57
SPEAKER_00: I mean,
32:57
another example of where we're seeing AI make a really big
33:01
impact in manufacturing
33:02
SPEAKER_00: is actually in this sort of design for manufacture process.
33:06
SPEAKER_00: So China,
33:07
SPEAKER_00: for instance,
33:08
SPEAKER_00: has this benefit of having these like very tight knit ecosystems where,
33:11
SPEAKER_00: you know,
33:12
SPEAKER_00: you have Shenzhen,
33:13
SPEAKER_00: You're a design engineer.
33:15
SPEAKER_00: You sit a block away from the manufacturing engineer.
33:18
SPEAKER_00: You know, you just go try stuff out.
33:20
SPEAKER_00: And because your friend tells you this is hard to build,
33:23
SPEAKER_00: you know,
33:23
SPEAKER_00: next time you design something,
33:24
SPEAKER_00: you,
33:24
SPEAKER_00: like,
33:25
SPEAKER_00: don't design it that way.
33:26
SPEAKER_00: That's great.
33:27
SPEAKER_00: We have to figure out how to abstract that knowledge in a more,
33:31
SPEAKER_00: in,
33:31
SPEAKER_00: like,
33:32
SPEAKER_00: our American way to build,
33:34
SPEAKER_00: like,
33:34
SPEAKER_00: what is our American version of Shenzhen.
33:36
and like my I would posit that the American version of Shenzhen
33:40
uses software and
33:42
AI to accomplish a lot of what the like physical ecosystem of
33:45
colocation in China
33:47
has given them so like how do we use AI to design better things
33:52
that are easier and
33:53
cheaper to manufacture and AI is really good at solving
33:59
anything that looks like a
34:00
SPEAKER_00: coding problem like you know you look at tools like
34:04
Cursor and Claude and you know these are some of the early
34:06
success cases of AI
34:08
where we're seeing like massive economic value be generated but
34:11
what's cool is like
34:12
SPEAKER_00: anytime you can represent something as like structured logic with
34:18
sort of limitations that you're trying to and constraints that
34:22
you're trying to
34:22
SPEAKER_00: optimize around.
34:23
Like you can represent that as code essentially and AI is
34:27
really good at it.
34:28
SPEAKER_00: So AI is very, very good at design problems.
34:31
SPEAKER_00: AI is very good at scientific design problems.
34:34
AI can design chips better than humans can design chips and AI
34:37
can design,
34:38
manufacture
34:39
Manufactured Goods Better Than Humans Can At This Point So
34:44
That's One Of The Use
34:44
Cases That We're Especially Excited About And Whether We're
34:47
Talking About Chips Or
34:48
PCB Boards Or Entire Components Or Entire Factories Like AI
34:53
Actually Just Being
34:54
Able To Design A Factory Or A Data Center From The Ground Up
34:58
That's Like A Super
34:59
Exciting Use Case Where AI Is Actually Producing Designs That
35:03
Are Better And Easier
35:04
To Manufacture And Cheaper To Manufacture And Faster To
35:06
Manufacture Than People Are
35:09
SPEAKER_02: There are a lot of reasons why I was excited about this conversation.
35:12
One of the main ones is you bring so much expertise across so
35:16
many different
35:18
subject matter areas where there's some kind of there's some
35:22
set of assumptions in
35:23
SPEAKER_02: the policy conversation.
35:24
SPEAKER_02: And I think the business reality is is potentially really different.
35:28
Maybe the main one that I'm thinking of right now is is the
35:31
conversation about data centers.
35:33
SPEAKER_02: So in the policy conversation now, data centers are increasingly becoming
35:38
SPEAKER_02: AI Policy Brief a16z
35:46
SPEAKER_02: I love this question I love data centers It's probably not a surprise
36:14
You know I think that like this is we are as a society we are
36:19
so lucky that
36:22
suddenly there is all this capital and attention and political
36:27
will on behalf of
36:28
tech companies to figure out how to build data centers because
36:33
they represent
36:34
something that I think where the lessons that we learn and the
36:37
process knowledge
36:38
SPEAKER_00: that we learn
36:39
SPEAKER_00: will translate to economic growth that extends so far beyond AI.
36:44
SPEAKER_00: Like, it's ridiculous.
36:46
SPEAKER_00: You know,
36:47
SPEAKER_00: data centers are attracting a lot of capital,
36:50
SPEAKER_00: so there's a ton of investment going into them.
36:53
And this is investment going into teaching us how to build mega
36:58
projects again in
36:59
SPEAKER_00: the U.S.,
36:59
SPEAKER_00: which is a really good thing.
37:01
Like,
37:01
we haven't built,
37:02
we haven't had any large-scale growth in this country,
37:06
physical growth.
37:07
SPEAKER_00: in a really long time.
37:09
SPEAKER_00: And so, you know, our wheels are very creaky.
37:11
SPEAKER_00: So when we're talking about reindustrialization,
37:14
SPEAKER_00: like building of data centers in the U.S.
37:16
SPEAKER_00: is such a boon to reindustrialization in the U.S.
37:19
because we're going to be teaching contractors and EPCs how to
37:23
build like high tech
37:25
SPEAKER_00: sites.
37:26
SPEAKER_00: And sure,
37:27
SPEAKER_00: like maybe those jobs specifically on data centers are short lived,
37:30
but I guarantee you those regions that attract data centers and
37:34
build data centers
37:35
SPEAKER_00: in their regions,
37:35
SPEAKER_00: they're going to attract
37:37
the next set of factories that are part of the
37:39
reindustrialization movement in the
37:41
US because they're going to have the workers who are trained
37:44
and skilled in
37:45
building these kind of high tech centers and sites that are
37:48
going to make companies
37:49
SPEAKER_00: want to build there.
37:50
And they're going to have the kind of regimes in place to be
37:53
able to do that.
37:54
I think the other thing that's really important about data
37:56
centers is they're
37:57
putting into focus a problem that's like been able to live sort
38:01
of under the
38:02
SPEAKER_00: surface
38:03
SPEAKER_00: for many decades, which is our aging electric grid.
38:07
And they're going to mean that we have kind of the capital and
38:10
the will and the
38:12
desire to figure out how to modernize something that we need to
38:15
modernize anyway.
38:17
Like our grid is an incredible complex machine that hasn't
38:21
really grown in several decades.
38:23
And we've been able to get away with it for a long time because
38:26
sort of efficiency gains in
38:28
Things like our refrigerators and industrial motors and other
38:32
processes have sort
38:33
of offset any sort of electric load growth that we've seen on
38:37
the grid over the
38:38
SPEAKER_00: last 30 years.
38:39
SPEAKER_00: And, you know, that's over.
38:41
SPEAKER_00: And that's over not because of data centers.
38:42
SPEAKER_00: That's just over because we're seeing,
38:44
SPEAKER_00: for the first time,
38:45
SPEAKER_00: AI aside,
38:46
a large-scale growth in electricity demand in the United States
38:50
for the first time
38:50
SPEAKER_00: in 30 years.
38:52
And we're going to need to figure out,
38:53
we're going to need to solve that problem.
38:55
We're going to need to solve that problem whether or not we
38:57
build data centers.
38:58
And the nice thing about building data centers is it puts a lot
39:01
of capital behind
39:02
SPEAKER_00: solving that problem.
39:03
Like figuring out how we improve delivery of electricity on the
39:07
grid is something
39:08
SPEAKER_00: that I for one would really like Google to help pay for.
39:12
SPEAKER_00: so that the consumer doesn't have to bear that cost alone.
39:15
SPEAKER_00: So if we can figure out how to do that,
39:16
if we can add transmission to the grid in the process of
39:19
building data centers and
39:20
then use that additional capacity to really drive
39:23
reindustrialization of the US,
39:24
SPEAKER_00: I think that that's an incredible
39:26
SPEAKER_00: outcome.
39:28
SPEAKER_00: And I completely understand the controversy.
39:32
SPEAKER_00: I get it.
39:33
SPEAKER_00: Like,
39:33
SPEAKER_00: people are worried about jobs,
39:35
SPEAKER_00: and they don't hear a positive narrative for AI,
39:38
SPEAKER_00: and all they hear is that it's going to increase their electric base.
39:42
And I think we really need to work hard to counteract that
39:46
narrative with the
39:47
SPEAKER_00: general public because I think it's not actually true.
39:50
SPEAKER_02: Yeah,
39:50
SPEAKER_02: I mean,
39:50
it sounds like you're articulating a case for why the benefits
39:53
of data centers
39:53
SPEAKER_02: outweigh the costs,
39:54
but I would love to get your sense on the specifics of the
39:57
allegations about costs.
39:58
SPEAKER_02: So, like, will people...
39:59
SPEAKER_02: There are many more data centers.
40:00
SPEAKER_02: Will people see their electricity rates rise?
40:03
SPEAKER_02: And what about water usage?
40:05
SPEAKER_02: And what about noise?
40:06
SPEAKER_02: And all of those kinds of issues.
40:07
SPEAKER_02: How do you think about those kinds of costs?
40:09
SPEAKER_00: I'll maybe focus on electricity in particular, just to go deep there.
40:13
SPEAKER_00: And I'd say,
40:13
SPEAKER_00: like,
40:13
there is a similar argument if you go down the line for a lot
40:16
of the arguments
40:17
SPEAKER_00: against data centers.
40:18
SPEAKER_02: You know,
40:18
SPEAKER_00: once they're on the grid and connected to the grid,
40:22
like it's much cheaper for utility to serve a data center
40:25
customer than like you as
40:27
SPEAKER_00: a consumer in your home.
40:28
Because you think about the highest cost of power to serve is
40:32
the low utilization user.
40:34
And like you in your house,
40:35
you turn on a light switch,
40:37
you run your dishwasher.
40:38
SPEAKER_00: On average, you're using like five to 10 percent of your peak power.
40:43
And then suddenly maybe like in the afternoon you might use 25%
40:47
of your peak power
40:48
SPEAKER_00: and then suddenly in the afternoon on a hot day you're using 100%.
40:51
SPEAKER_00: So your average utilization as a consumer is extremely low,
40:57
which means the cost to serve you by a utility is very high
41:01
because they have to
41:01
SPEAKER_00: essentially plan for your peaks,
41:03
SPEAKER_00: but you're very rarely using that.
41:06
SPEAKER_00: On the other hand,
41:08
SPEAKER_00: A data center customer is very high utilization.
41:11
SPEAKER_00: So data centers are like something like 80-90% utilization.
41:15
SPEAKER_00: They're using a lot of power all the time, super predictable.
41:19
SPEAKER_00: They're very, very cheap for utility to serve.
41:22
SPEAKER_00: So when you add a lot of data center loads to the grid,
41:25
SPEAKER_00: assuming,
41:26
SPEAKER_00: you know,
41:27
they help bring some of that generation online and help pay for
41:30
that generation
41:30
SPEAKER_00: online,
41:30
which all of the hyperscalers have indicated that they're
41:33
extremely willing to do
41:34
SPEAKER_00: and in fact are already doing,
41:37
SPEAKER_00: They're bringing more cheap power onto the grid,
41:39
SPEAKER_00: which lowers the cost for the average consumer.
41:42
SPEAKER_00: So in an ideal world,
41:44
SPEAKER_00: long term,
41:46
data center loads plus other heavy high utilization industrial
41:49
loads are really,
41:50
SPEAKER_00: really good for the grid and really good for the consumer.
41:53
They will lower the cost of electricity for every single
41:56
consumer on the grid.
41:58
which feels a little bit counterintuitive and so my big worry
42:02
for this moment that
42:03
we're in is that there's going to be such a backlash against AI
42:06
and data centers
42:07
that all of these data centers are going to go build their own
42:10
grids they're going
42:11
to be completely off the public grid they're going to build
42:13
their own power
42:14
generation they're going to build their own storage and backup
42:17
and they're going to
42:17
SPEAKER_00: give none of that to the grid itself
42:20
you know because everybody rejects them and makes it too hard
42:22
for them to do it and
42:23
then we actually as consumers get none of the benefit of that
42:26
investment into
42:28
lowering our cost of power so that's like one of my big worries
42:31
with the current
42:32
moment in AI I know I like I'm optimistic that we'll solve it
42:36
because I'm an
42:36
optimistic person but I think that that would probably be the
42:38
worst outcome that we
42:40
could get to for the cost of power for the American's consumer
42:45
and also sort of the
42:46
SPEAKER_00: ability for us to support
42:48
a broad scale U.S.
42:49
reindustrialization,
42:51
which will be energy intensive kind of after this wave.
42:54
There are a lot of people who are concerned about what AI is
42:56
going to mean for jobs.
42:57
SPEAKER_02: Erin, how do you see that issue?
42:59
SPEAKER_00: I mean,
42:59
SPEAKER_00: I like fundamentally see AI as a net job creator,
43:05
especially when it comes to reindustrialization and non-white
43:10
collar work.
43:12
SPEAKER_00: Like we're talking about bringing factory jobs back to the U.S.,
43:17
SPEAKER_00: You know, that haven't existed here in decades.
43:20
SPEAKER_00: And I'm not talking about factory jobs where you have someone,
43:23
SPEAKER_00: like,
43:23
SPEAKER_00: lifting heavy boxes or,
43:26
SPEAKER_00: you know,
43:26
lifting heavy parts and throwing out their back or twisting one
43:31
single screw into a
43:34
SPEAKER_00: socket over and over again all day long and,
43:36
SPEAKER_00: like,
43:36
SPEAKER_00: destroying their wrist.
43:37
SPEAKER_00: Like, those aren't the type of manufacturing jobs I'm excited about.
43:40
SPEAKER_00: Those are gonna be done by robots, thank goodness.
43:42
SPEAKER_00: No, I'm talking about, like, high-skilled, high-tech jobs
43:47
SPEAKER_00: You know,
43:48
SPEAKER_00: doing quality engineering at scale,
43:51
SPEAKER_00: doing process engineering at scale,
43:53
SPEAKER_00: understanding how to,
43:55
SPEAKER_00: you know,
43:55
SPEAKER_00: take care of and service that robotic technicians.
43:58
I'm talking about electricians who are working on really
44:01
complicated systems level
44:02
SPEAKER_00: problems,
44:03
SPEAKER_00: you know,
44:03
SPEAKER_00: on our factories.
44:04
SPEAKER_00: These are not, don't require college degrees.
44:08
SPEAKER_00: But as we build out more industrial capacity in the U.S.,
44:11
I think we're going to see an increasing number of this type of
44:14
high-skilled
44:14
SPEAKER_00: blue-collar work come back to the U.S.
44:16
SPEAKER_02: Aaron,
44:16
SPEAKER_02: thanks so much.