0:00

Today on the AI Daily Brief, what

0:02

the impacts of the tariffs are likely to

0:04

be on the AI industry. The AI

0:06

Daily Brief is a daily podcast

0:08

and video about the most important

0:10

news and discussions in AI. To

0:12

join the conversation, follow the Discord

0:14

link at our show notes. Hello friends.

0:16

Three quick notes before we dive into

0:19

today's show. First up, as I've been

0:21

mentioning a couple times, for those of you

0:23

who are looking for an ad-free version of

0:25

the AI Daily Brief. You can now head

0:28

on over to patreon.com/a ideally brief to find

0:30

that. We'll be doing a lot more stuff

0:32

over there. For now, the first thing that

0:34

I was trying to do is just give

0:37

you an ad-free option if that's meaningful

0:39

to you. Second, next week is

0:41

spring break, so we will be

0:43

doing the parental pilgrimage of chasing

0:45

our kids around Disney World, which

0:47

means that we will not be doing normal

0:49

shows on the a ideally brief. I'm really

0:52

excited to have a bunch of cool conversations,

0:54

digging, So you will not not have shows,

0:56

but they will be a little bit different

0:58

than normal. You can bet that if any crazy

1:01

news happens, I'll use AI of some kind to

1:03

get my perspective on it out there. But the

1:05

plan is of now to not have normal shows

1:07

next week. Please give me your thoughts as we

1:09

dive into the world of the mouse. Lastly, something

1:11

I want to gauge people's perspective on.

1:13

The AI Daily Brief community has been

1:16

hugely supportive of and important in the

1:18

super intelligent story. As we build out

1:20

the agent and agent services marketplace, we're

1:23

considering reserving part of our current round

1:25

for investors from this community. However, I'm

1:27

trying to gauge interest. If this is something

1:29

you think we should explore, send me a

1:31

note at NLW at B super dot AI with

1:33

super in the title. Thanks in advance for

1:35

your perspective. And with that, let's get into

1:38

today's show, which is a deep dive only

1:40

no headlines about the tariffs likely impact on

1:42

the AI industry. Welcome back to the AI

1:44

Daily Daily Brief. Right now, the most

1:46

significant and inescapable force-driving markets is,

1:49

of course, the Trump tariffs and

1:51

the potential total restructuring of the

1:54

global economic trading order that they

1:56

represent. Today we are going to look at how

1:58

they are and how they might impact the... AI industry.

2:00

I sort of think that the impacts

2:02

are incredibly wide reaching and incredibly wide

2:04

ranging. There are implications for data centers,

2:06

chip export controls, relationships with China, other

2:09

key ally relationships that relate to AI,

2:11

there's gonna be impact on AI startups.

2:13

I think there'll be an acceleration of

2:15

AI-based job transformation. So let's get into

2:17

all of it. And let's start with

2:19

some of the more obvious areas. One

2:21

of the big glaring issues that hit

2:24

the headlines earlier this week was around

2:26

GPU supply. The administration had provided an

2:28

exemption in the tariffs for semiconductors, but

2:30

not for finished GPUs. It's a little

2:32

unclear whether the carve-outs were intended to

2:34

allow invidia to bring in their crucial

2:37

hardware without paying the tariff, but for

2:39

now it looks like GPUs just got

2:41

a whole lot more expensive. invidia is

2:43

still looking at big price increases any

2:45

way you slice it. They largely delivered

2:47

GPUs for AI applications as full server

2:50

racks which require memory, storage, and dozens

2:52

of other electronic electronic components. The entire

2:54

supply chain flows flows through through China.

2:56

So essentially all of these manufacturers will

2:58

be caught up in the spiraling escalation

3:00

of tariffs. I literally have two podcasts

3:02

that this is relevant for, and I

3:05

can't even keep track of the tit-for-tat

3:07

acceleration that we're seeing between the US

3:09

and China when it comes to adding

3:11

more tariffs on, what's more, China when

3:13

it comes to adding more tariffs on.

3:15

What's more, China has been targeting their

3:18

retaliation at electronic supply specifically, including export

3:20

controls being imposed on crucial rare earth

3:22

minerals. Their latest escalation at least that

3:24

I can. announced just recently to match

3:26

the 104% levied by the US. You

3:28

may be thinking to yourself at that

3:31

level the tariffs may as well be

3:33

a ban on trade and you wouldn't

3:35

be all that far off. Given that

3:37

even in a world where there weren't

3:39

tariffs, there were huge GPU shortages, and

3:41

this key infrastructure was one of the

3:43

major challenges for AI companies, having all

3:46

of the inputs across the entire supply

3:48

chain be effectively twice as much, can't

3:50

be doing anyone any favors. Beyond the

3:52

GPU price squeeze? The US is also

3:54

in the middle of a data center

3:56

construction boom, a boom that is meant

3:59

specifically. to address some of these issues.

4:01

The tariffs could add massively to the

4:03

cost of raw inputs like steel, concrete,

4:05

and aluminum, and build-out costs could escalate

4:07

with networking and cooling equipment also being

4:09

hit with the additional charges. Matthew Middleset,

4:12

a technology policy researcher at the Cato

4:14

Institute, said, the AI future is now

4:16

being taxed. Aside from construction, the ability

4:18

to get the gigantic energy supply needed

4:20

to power the AI revolution could also

4:22

be in jeopardy. The US imports around

4:24

a fifth of our solar panels from

4:27

China, and another fifth from countries in

4:29

Southeast Asia who are all facing heavy

4:31

tariffs as well. To the extent that

4:33

renewable energy is being used to power

4:35

data centers, the cost just went up.

4:37

The bulk of data center projects are

4:40

being powered by gas turbines, which are

4:42

largely sourced from Germany or Japan. These

4:44

crucial inputs are already in scare supplies,

4:46

so manufacturers will pass on all the

4:48

tariff costs to an already expensive piece

4:50

of equipment. To give a sense of

4:53

how tenuous the power supply build-out could

4:55

be, the administration has reportedly drafted an

4:57

executive order to expand coal production in

4:59

order to meet data center demand. Now

5:01

outside all of these practical effects, on

5:03

the opposite side of the trade wall,

5:05

chip exports are likely to be a

5:08

difficult task for US-based companies as well.

5:10

In the first quarter, Chinese firms rush

5:12

to order 16 billion worth of invidious

5:14

chips to get ahead of tariffs. Despite

5:16

being limited to underpowered chips, China still

5:18

represents around 13% of invidia sales, or

5:21

even more if you're skeptical about demand

5:23

out of Singapore or other countries in

5:25

Southeast Asia. Invidia might be able to

5:27

cut the U.S. out of their logistics

5:29

and deliver directly, but the company would

5:31

then risk being demonized by the administration

5:34

for working around the tariffs. Invidia has

5:36

high margins, so can arguably afford to

5:38

absorb the additional costs, but the strategically

5:40

important company is increasingly a political football

5:42

in the competition between the U.S. and

5:44

China, and could be uncomfortably forced to

5:46

pick sides. There's also something broader going

5:49

on here as well. Up until now,

5:51

despite the rhetoric... Thriving open source communities

5:53

in both China and the US have

5:55

been building on each other's work, accelerating

5:57

things as that happens. However, the AI

5:59

Cold War has been getting colder more

6:02

recently. We'd already seen reports, for example,

6:04

of top Chinese-born AI scientists returning home,

6:06

and things like that are only set

6:08

to accelerate as tensions rise. What's more...

6:10

to the extent that one sees the

6:12

battle between China and the US when

6:14

it comes to AI supremacy, as fundamentally

6:17

about whose models are used around the

6:19

world, the US by effectively isolating itself

6:21

from everyone, certainly seems to be creating

6:23

an incentive for other countries to become

6:25

AI vassal states of China and not

6:27

us. All right, AI daily brief listeners,

6:30

today I'm excited to tell you about

6:32

the disruption incubator. One of the things

6:34

that our team sees all the time

6:36

is a lot of frustration from enterprises.

6:38

There's a fatigue around small incremental solutions,

6:40

a concern around not thinking big enough,

6:43

tons of bureaucratic challenges of course inside

6:45

big companies, and frankly we just hear

6:47

all the time from CEOs, CTOs, other

6:49

types of leaders that they want to

6:51

ship some groundbreaking AI agent or product

6:53

or feature. In many cases they even

6:55

have a pretty well thought out vision

6:58

for what this could be. Their teams

7:00

are just not in an environment conducive

7:02

to that type of ambition. Well it

7:04

turns out our friends at Fractional have

7:06

experienced the exact same thing. Fractional are

7:08

the top AI engineers specializing in transformative

7:11

AI product development, and to answer this

7:13

particular challenge, they have, with perhaps a

7:15

little bit of help from super intelligent,

7:17

set up what they're calling the disruption

7:19

incubator for exactly this type of situation.

7:21

The idea of the disruption incubator is

7:24

to give a small group of your

7:26

most talented people, an overly ambitious mandate,

7:28

something that might have taken one to

7:30

two years within their current construct. Send

7:32

them to San Francisco to work with

7:34

the team at Fractional. And within two

7:36

to three months, ship something that would

7:39

have previously been impossible. The idea here

7:41

is that you are not just building

7:43

some powerful new agent or AI feature,

7:45

but you're actually investing in your AI

7:47

leadership at the same time. If this

7:49

is something interesting to you, send us

7:52

a note at Agent at B-super.i with

7:54

the word disruption in the title and

7:56

we will get right back to you

7:58

with more information. Again that's Agent at

8:00

B-super dot AI with disruption in the

8:02

subject line. Today's episode is brought to

8:05

you by Vanta. Vanta is a trust

8:07

management platform that helps businesses automate security

8:09

and compliance, enabling them to demonstrate strong

8:11

security. practices and scale. In today's business

8:13

landscape, businesses can't just claim security, they

8:15

have to prove it. Achieving compliance with

8:17

the framework like SOC 2, ISO 2701,

8:20

HIPAA, GDPR, etc. is how businesses can

8:22

demonstrate strong security practices. And we see

8:24

how much this matters every time we

8:26

connect enterprises with agent services providers at

8:28

Super Intelligent. Many of these compliance frameworks

8:30

are simply not negotiable for enterprises. The

8:33

problem is that navigating security and compliance

8:35

is time-consuming and complicated. It can take

8:37

months of work and use up valuable

8:39

time and resources. Vanta makes it easy

8:41

and faster by automating compliance across 35

8:43

plus frameworks. It gets you audit ready

8:46

in weeks instead of months and saves

8:48

you up to 85% of associated costs.

8:50

In fact, a recent IDC white paper

8:52

found that Vanta customers achieve $535,000 per

8:54

year in benefits, and the platform pays

8:56

for itself in just three months. The

8:58

proof is in the numbers. More than

9:01

10,000 global companies trust Vanta, including Atlian,

9:03

Cora, Cora, and more. For a limited

9:05

time, listeners get $1,000 off at vented.com/NLW.

9:07

That's va-n-t-a.com/NLW for $1,000 off. Today's episode

9:09

is brought to you by Plum. If

9:11

you're building agentic workflows for clients or

9:14

colleagues, it's time to take another look

9:16

at Plum. Plum is where AI experts

9:18

create, deploy, manage, and monetize complex automations.

9:20

With features like one-click updates that reach

9:22

all your subscribers, user-level variables for personalization,

9:24

and the ability to protect your problems

9:27

and workflow IP, it's the best place

9:29

to grow your AI automation practice. Serve

9:31

twice the clients in half the time

9:33

with Plum. Sign up today at useplum.com.

9:35

That's use-p-l-u-m-b.com./NLW. Indeed, one of the key

9:37

areas to watch will be how third-party

9:39

nations react to the fragmentation of foreign

9:42

relations. During the recent escalation of chip

9:44

exports, countries like India and Israel were

9:46

excluded from a list of close allies

9:48

with open access to U.S. tech. China,

9:50

meanwhile, has made it a national priority

9:52

to make open-source AI as widely and

9:55

cheaply available as possible. One of the

9:57

major flashpoints for this potential global reorder

9:59

is of course the Middle East. The

10:01

Gulf states have been pushing hard to

10:03

establish themselves as an AI hub, and both

10:05

geographically and strategically the region straddles

10:08

the US in China. Even before any

10:10

of this tariff stuff happened, Gulf countries

10:12

were already walking a thin line. Golf

10:14

company G42, for example, has been in

10:16

the spotlight for more than a year now.

10:18

Microsoft wanted to take a minority stake

10:21

in the company, and at first there

10:23

was big consternation about it, which forced

10:25

G42 to effectively pick sides and pick

10:27

the US. And then ultimately the Commerce

10:30

Secretary for the Biden administration, Gina Raimondo,

10:32

was involved in the final deal that

10:34

came together at the end of last year.

10:36

Now when it comes to the tariffs, Saudi

10:38

Arabia and the UAE, were both included in

10:40

the basket of countries subjected to the

10:42

10% baseline tariffs. Relative to others then,

10:45

we might see them as relatively

10:47

unscathed. Still, the geopolitical centering in

10:49

general could force the region to pick

10:51

aside, and that doesn't necessarily point back

10:53

to the US. Another dimension where I

10:56

think tariffs themselves. First of all, let's zoom

10:58

out to Wall Street. AI has been propping up markets

11:00

for two and a half years now. The launch

11:02

of ChatGPT coincided with the beginning

11:04

of the hiking cycle, and even

11:06

as markets were tanking with the

11:08

unwind of Zirpara policies, enthusiasm about

11:10

AI kept things afloat. AI's ability

11:12

to sustain the market, however, has looked

11:14

more and more shakyy over the last three

11:17

to six months. And certainly it would

11:19

not be surprising to see all

11:21

of the volatility around tariffs accelerate

11:23

the pullback that we're already seeing

11:25

from tech companies when it comes

11:27

to data center and general

11:29

AI infrastructure investment. Now that

11:32

doesn't just matter for Wall Street

11:34

it has downstream effects. Already we're

11:36

seeing IPO delays which puts even

11:38

more pressure on an already beleaguered VC

11:40

space. Without IPOs and exits, the

11:42

venture ecosystem doesn't have money

11:45

to return and reinvest in new

11:47

startups. And it's also the case for angels

11:49

who without liquidity have limited means to invest

11:51

in the next generation of startups. The information

11:54

took this specific example of Charles Hudson's precursor

11:56

ventures who very bluntly described what he was

11:58

likely to have to do over the next

12:00

half decade or so. He said that secondaries, basically

12:02

selling stock of private startups to other investors, will

12:04

represent 75 to 80% of the dollars that LPs

12:06

get back in the next five years. Effectively, Charles

12:08

is anticipating having to stay in private markets to

12:10

get liquidity because of the difficulty of the larger

12:12

exit situation. I'm already seeing in my conversations with

12:14

funds, especially funds that are trying to raise new

12:16

funds right now. LPs are clamming up, which means

12:18

venture firms are having a harder time raising new

12:20

funds. which means that everyone is going to be

12:22

doing more sitting on their hands, which reduces access

12:24

to capital for all companies in general, inclusive of

12:26

AI, even if it remains the hottest category. One

12:29

more little piece of evidence around some maybe volatility

12:31

in the space. Own McCabe the CEO of Intercom

12:33

tweeted, I've been receiving about one new AI acquisition

12:35

opportunity in my inbox every day recently. Today already

12:37

I've got two. Not sure what it means, if

12:39

anything. Jason Freed from 37 Signals writes, me too,

12:41

and we don't buy companies never have. And I'm

12:43

not really an investor either. Could all be BBS

12:45

or signaling the obvious? Almost all these AI think

12:47

companies have no path to survival. Now it is

12:49

a longer conversation around how much this represents the

12:51

natural consolidation of the AI industry a couple years

12:53

after the post-chat GBT boom, or whether this is

12:55

a leading indicator of troubled waters. Now even outside

12:57

everything with tariffs, AI was already adding some weird

12:59

complication to the VC model. Ethan Mollick recently tweeted,

13:01

startups take five to seven years to exit on

13:03

average, more for biotech. Most of the VCs seem

13:05

to believe that American AI advancement will happen in

13:07

that time frame, but I'd love to hear more

13:09

about their vision for the world in five to

13:11

seven years and how their portfolio firms maintain advantage.

13:13

I totally get how right now is an amazing

13:16

time to be a startup with the huge multipliers

13:18

that LLLM's can provide the huge multipliers that LLLM's

13:20

can provide the founders can provide to founders. The

13:22

VCs get how right now is an amazing time

13:24

to be a startup with the huge multipliers that

13:26

LMLs, what is, what is, what is, what is,

13:28

what is, what is, what is, what is, what

13:30

is, what is, what is, what is, what is,

13:32

what is, what is, what is, what is, will

13:34

happen, what is, what is, what is, what is,

13:36

will happen, what is, will happen, what is, what

13:38

is, will happen, what is, will happen, will happen,

13:40

and, and, and, will happen, will happen, and, and,

13:42

and And finally he wrote, Reader, they did not.

13:44

Another dimension of this, we also have this new

13:46

phenomenon. of companies seed strapping, basically raising one round

13:48

of funding and then trying to turn to profitability,

13:50

using the new efficiency gains and opportunities that AI

13:52

represents. All of this I think was already happening

13:54

and was likely to lead to some amount of

13:56

transformation of the venture capital business, but I actually

13:58

think that tariffs are going to radically hasten this

14:01

transformation. My logic is this. AI was already making

14:03

teams reconsider how much capital they need it. Now

14:05

is elfies freeze up and VCs also start to

14:07

slow down. Portfolio companies and entrepreneurs are going to

14:09

accelerate their move to a defensive cash-efficient posture posture.

14:11

Those that make that transition successfully, many of them

14:13

will probably decide that actually they don't need venture

14:15

the same way they might have expected before. And

14:17

so in this way, I actually think that the

14:19

natural tendency of VCs to turtle up right now

14:21

is going to hasten their own decline. That's not

14:23

to say that big companies that want to go

14:25

after Blue Ocean's opportunities won't still need capital, they

14:27

will. And that's not to say that VCs can

14:29

adapt their can adapt their model. They can, but

14:31

it's very clear that the venture capital scene of

14:33

today looks very different than the venture capital scene

14:35

of tomorrow, and I think tariffs are going to

14:37

significantly accelerate forces that were already happening. Lastly, of

14:39

course, there are the job implications. And this is

14:41

not just for startups, but for every company. Yesterday's

14:43

show was all about the Shopify AI memo, which

14:46

as I argued, I don't believe was just strictly

14:48

about a soft hiring freeze. And yet still, the

14:50

implications of it were functionally a soft hiring freeze.

14:52

Some even speculated that that was the actual point

14:54

underneath, and it was a market-palatable way of doing

14:56

that without spooking investors. AI was already creating some

14:58

very dynamic conversations internal to companies around how they

15:00

think about staffing going forward. With recession predictions being

15:02

updated by the minute this week, those conversations have

15:04

to be accelerating the relatively normal downturn was for

15:06

the tech sector in 2023, and this could be

15:08

much much worse. Now of course layoffs traditionally aren't

15:10

a particularly welcome option for corporate leaders. They tend

15:12

to be seen as a sign of weakness and

15:14

slowing growth. However, if a recession or downturn does

15:16

come to pass, it'll be the first one where

15:18

AI is a viable replacement for some amount of

15:20

that human labor. I've talked to... how I

15:22

think we have to get

15:24

through the efficiency phase

15:26

of AI of AI, companies treat

15:28

it primarily as a

15:31

way to cut costs to cut

15:33

opposed to harness new

15:35

opportunities. new I think that's

15:37

going to happen a heck

15:39

of a lot faster

15:41

because of everything going on

15:43

with tariffs of their downstream

15:45

impacts like recession. and their

15:47

None of this is for

15:49

certain. like Part of what

15:51

makes it such a

15:53

difficult environment of that it's

15:55

really not clear to anyone

15:57

exactly what the end

15:59

game here is. Because we

16:01

don't know that, it's

16:03

hard for anyone to make

16:05

clear decisions. is. And so

16:07

we're going to be

16:09

operating in a period of

16:11

instability for the foreseeable

16:13

future. The impact on AI

16:15

is, of course, just

16:18

one small set of impacts

16:20

a a radical of of that

16:22

these these new policies But even

16:24

But even over here

16:26

in our little corner of

16:28

the world, there are

16:30

clearly going to be some

16:32

big changes that come

16:34

from all of this. all of

16:36

For now, though, that

16:38

to do it for today's

16:40

it Brief. Aye Daily you listening

16:42

or watching as always. or

16:44

And until next time,

16:46

peace. time, peace. you