Why China's AI Strategy Just Might Work — With Grace Shao

Channel: Alex Kantrowitz

Published at: 2025-04-16

YouTube video id: pksvCuoCOWs

Source: https://www.youtube.com/watch?v=pksvCuoCOWs

Was Deepseek an outlier or a sign that
China's AI industry is a force to be
reckoned with and a potential global
leader in artificial intelligence? We'll
talk about it with a leading China
analyst right after this. Welcome to Big
Technology Podcast, a show for
coolheaded nuance conversation of the
tech world and beyond. You asked for it.
We're delivering a China episode all
about China's AI positioning, how
competitive the country is in AI,
whether deepseek was an outlier, and
what might be coming next. It's a show
that we've been looking forward to doing
for a while and very happy to be
bringing to you today. And we're joined
by a great guest. Hong Kong based writer
and analyst Grace Shia is here. She's
the author of AI Prom on Substack and uh
is here to share with us everything
that's going on in China and all the
context that's been missing from the
headlines. So, Grace, great to see you.
Welcome to the show. Thank you so much
for having me, Alex. Great to meet
everyone here online. All right, let's
start off with this question that we've
been wondering about here, which is
whether Deep Seek was an outlier,
whether it was just a smart group of
people who came up with a real AI
innovation or the tip of the spear
showing that Chinese AI is really
starting to be globally competitive and
a force to be reckoned with on the
worldwide stage. So, I'm curious what
you you think about this. Was it an
outlier or was it something bigger?
I think definitely we're seeing that
it's not an outlier like to start off
with like deep sea completely changed
how the world viewed like our scaling
loss right like and that implication
isn't even just on that end but also I
think it's really helped um bring
confidence to Chinese AI space and AI
entrepreneur space and I think now one
by one we're seeing more and more like
highprofile AI entrepreneurs coming out
and a lot of them are quite young
they're born in the 80s and 90s and this
generation of entrepreneurs are very
different. So first of all to start off
with it's really interesting like a lot
of you might have read Leangfong the CEO
of Deep Seek and his whole team they're
actually educated in China. Um so I
think that's one aspect of where we're
seeing people realizing this is a bigger
trend and we're seeing more talent
really coming from these schools. And
then I think another thing which was
really interesting is um I was talking
to friends about it in China and they're
saying look this is actually starting
the rise of what is called the open
source seal like you know people are
really excited to be quoted to be you
know recognized and I think this is
again a really really stark difference
from the last generation of
entrepreneurs Chinese entrepreneurs now
no longer are just driven by money and
making money they're driven by
recognition influence and power and a
lot of them are like visionary start off
with and not businessmen to start off
with and that's a huge difference in say
even the internet era like let's say
Alibaba jack like he talks about making
money making businesses easier a lot of
the businesses and products were
actually driven by the desire to make
money and it there's no shame around you
know I can copy and copy and copy and
then make it bigger bit better right or
I can make a cheaper version whatever is
happening in the west or a business
product out in the west but in China
right now a lot of these AI companies
are driven by actually like missions. So
I think then we're seeing a bigger trend
of AI companies coming through and I
think last but not least is because of
the last couple years of geopolitical
tensions frankly it's not made it easy
for them um and I think there's a
stronger even sense of nationalistic or
patriotic desire to say hey we can do it
we can actually contribute to the global
innovation scene we can lead in
innovation so there's definitely a lot
of that and then I think that that's
definitely something to continue to we
can continue to watch and I expect that
we'll see more and more innovation in AI
space coming out of China. Grace like I
I hear you but I'm also surprised to
hear you say that this is not driven by
money given that Deepseek was created by
a literal hedge fund. I mean it seems
like it would be the definition of
chasing for money. But I think that's
really interesting because we've seen I
actually been reading a lot of the
Chinese media or interviews or even
coverage on him and himself at least
like Leang himself. I think for sure
people are saying he's he's actually
like quite an interesting figure. He
came from an extremely frugal background
like I'm talking about like bricks as a
h like bricks made house like you know
like a village and he was hailed as a
hero when he returned back for Chinese
New Year recently right and I think he
came from this humble beginning and he
was just a really really really smart
nerdy math guy essentially and he went
into quant trading um and it's really
interesting that essentially he footed
the bills himself to do this he made
this pivot into AI a couple years ago
and started training models And so far
right now everything's for free. And
that's another argument like what topic
we can talk about which is this huge
embrace of open source and this
expectation within the Chinese AI space
that LLMs are right now all going open
source right even by I think announced
that they're going to go um open source
in June. Baba is the famous one that was
always open source and free basically in
that sense. Um, obviously deep seas
completely changed a narrative now
making everyone expect that LM's just
going to be for free in China.
Why is the Chinese AI industry open
sourcing all these models? Like let me
tell you in the US I think the view is
that China is open sourcing these
because they want to mess with US
research labs and show the open AIs of
the world that the product that was
supposed to be their moat is no longer a
moat. and take the US AI industry down a
notch. Uh, is our thinking on the on
that line anywhere close to
reality? I think honestly there's a lot
of politically charged narratives going
around given the last couple years of
where the trade war has led us and on a
very personal level for someone like
myself and friends around me who had to
kind of like build their identity and
business and everything on these two
worlds in in assumption of it working
together. That's not good. It's not a
very productive or cohesive narrative
going forward. Now, in a very business,
I think very pragmatic way, let's just
look at it. Um, China actually I wrote
about this extensively with a
collaborator called JS. And, you know,
we we really examined why China had to
kind of go through um or go with a AI
monetization strategy that's really
heavily leaning toward consumer versus
enterprise. And it fundamentally starts
with a really really low SAS adoption.
So in China there's we we all know
enterprise software just never really
took off right like like this is based
on many many different things. Number
one let's to start off with let's just
look economic structure um China is
still not a very heavily depend is not
heavily dependent on knowledge workers
compared to the US where like 60% of GDP
is driven by knowledge workers. Actually
if you think about China's vast like
significant economic growth over the
last three decades it's very much driven
by labor. So it's still in manufacturing
sectors, it's still in, you know, um
even domestic consumption, but like it's
not really what we think of as white
collar jobs and because of that um it's
just not had that really strong um
structural ecosystem for enterprise to
kind of flourish to start off with.
Number two, I think we also looked at,
you know, just the the that because
there was such a frankly poor um
protection of IP laws and everything. We
saw that you know a lot of the uh
softwares that came into China were
actually kind of operating in this gray
area of like pseudo legal space where
you can actually just buy pirated
Microsoft and like even mega legitimate
companies would just use pirated like
word do word and excel whatever all of
this laid the foundation to now where
then we look at the internet
conglomerates because there's no massive
enterprise companies right um even the
companies that have enterprise software
like Alibaba or bite dance those
products are their minor businesses. So
it's the whole internet economy is so
dependent and reliant on consumerf
facing applications and everyone knows
about the super apps right the massive
touch points and like the 1 billion
people DAU kind of um figures. So now
for China on top of all of that and then
Deep Sea completely broke the moat where
like said and told everyone like look
you don't actually need to pay for LM um
go and get your free LLMs and the real
innovation and use cases are an
application. So everyone in China right
now is like okay why should I pay for
this and um you know why why why don't I
just like try to innovate on top of it
essentially create like LM rappers right
um I think so that that's like one
aspect why we think China just went into
consumer straight head-on like you know
they never really tried to really go
enterprise direction except for Alibaba
they kind of they still are trying to
leverageing the cloud so can I can I
just follow up here so is is what you're
saying that we're seeing all these open-
source AI companies come or open source
AI models come out of China because
they're being developed by consumerf
facing companies and so the plan for
them is instead of let's sell our AI
models to other businesses the plan for
them is we're going to build consumer
apps uh that are based on top of the
large language models that we are
developing and if we open source them we
might get developers in there we might
speed this technology and then we'll
basically play in our area of expertise
and try to win that way. Is that why
we're seeing so much open source? Is
that what you're saying? 100%. I think I
think I think you put it in a very
succinct way. I like for me like I was
thinking about the macro and then the
micro, right? Like but like the way you
put it exactly what it is and I think
even in during Alibaba's recent earnings
call, Eddie Woo like the current CEO
even said something like look we're
going to be like the provider of
basically everything as in the
infrastructure layer. We're the we're
the grid, but in the future, we don't
know what monetization is. We're going
to completely We're going to continue to
do our cloud thing, like our cloud
business, which they're leading in China
and actually in Asia. We're going to
continue to enhance and um make sure our
model is frontier level like their Qin
and then we're going to push out all
these um AI empowered products um
applications like their Quakula and etc.
Right? So, so I think really the
monetization strategy is all an
application end and they're hoping that
one day whether it's you're plugging in
ads or whatever, you know, charging at
friction points, that's where it's going
to make the money. It is interesting
because I had a conversation with Kyle
Lee once uh for my book and he had I
think maybe in one of his books about AI
called Mark Zuckerberg the most Chinese
uh of all like American CI CEOs. And the
reason why he said that is because there
is an environment in China where people
see apps come up and then immediately
copy them. And it's accepted that
copycat apps are just the law of the
land there. And you better be ready to
move and adapt and copy quickly. And
that's clearly what Zuckerberg has done
with him copying stories with Snapchat
and copying reals copying Tik Tok with
reals. And it just totally makes sense
that he's also using this now Chinese
strategy of open source AI to help
develop consumer apps with Facebook
because that is their that is
effectively I think if I'm getting what
you're saying right in terms of why you
know Alibaba is uh developing AI open
source AI and Tencent is developing uh
open source AI Facebook is doing the
same thing where they're basically
saying we want to lead the direction of
where AI goes. We want everything to be
compatible with our systems and that
will in turn lead to growth in our
products and we're going to put it
everywhere uh including you know within
messenger and within WhatsApp and within
Instagram and that's how we're going to
grow and I I was just speaking with Yan
Lun the chief AI scientist at Meta and
he said there's 600 million people using
their their AI assistant within their
apps compared to 400 million uh of open
AI's chat GPT. now they use it less. Uh
but the parallels in that strategy is
just uncanny. No, 100%. I think you
completely nailed it on that like I
think Zuckerberg is the only one so far
out of the American big tech that
actually openly said I think in the
recent earnings call as well he said
something like we're going to have a 1
billion people serving AI agent right
like that is his like vision and that's
exactly what you said the Babas and the
bite dance are um imagining to do.
What's interesting is I I do want to
point out is Tencent actually has not
really been leading in their LLM
development at all. In fact, they're the
only and first one that have actually
integrated Deep Seek into their products
um into their WeChat, which which
obviously um all of you know about. It's
like the super app of China, right? They
got 1.3 for those who don't, right? The
leading messenger, you can do everything
there, payments. So, they they've put
DeepSeek within Yeah. And I think I
wrote a lot about this and then I do
want to highlight how significant this
is and why that's completely also
changed the mentality of Chinese AI
players. Before Tencent integrated Deep
Seek, which is I think they did about a
month ago. Um, everyone's still kind of
trying to like close up their doors like
like do of Bite Dance. They were still
being like, "Okay, we're going to do our
own apps. We're going to spend a ton of
marketing. We're going to try to get
people onto our app, new app, right?"
But the issue with that is you actually
need to spend a lot of time and effort
to convert new users. Exactly. To your
point where if you already have an
existing app like Facebook and in this
case WeChat for 10 cent, you have the
really really strong strength of
distribution and reach already. And I
think this is something they did really
well. So they basically said, you know
what, we have a pretty crap LLM. We're
not even going to try to compete on this
right now. Deepseek is um great. We're
going to just inter uh integrate it. So
they first did that and then I think
another thing that like you said Meta is
doing completely the same is they have
these walled garden or proprietary data
per se where if you think about it um
like you said we chat is a messaging app
right but beyond that you pay utilities
on it you call for your taxi on it you
actually write blogs on essentially
there's a built-in substack on it so all
of this information is your proprietary
information that no one else can scrape
cuz if you go on China's Google which is
bu and you want look up an article that
you know like Alex wrote right in
Chinese. You're not going to find it on
BU easily, but you got to search it
within WeChat. And then last but not
least, I think it's also the um the
point on functional adjacency because I
said WeChat has an built-in search kind
of engine or search function similarly
to Facebook, right? Like we go on
Facebook, we search for events, things
around us, uh people, we creep on
people, we creep on companies, whatnot,
right? There's a search engine built
into it. So there's a very natural way
for people to kind of say hey maybe I'll
just try the AI function right versus if
you think about Tik Tok which is bite
dance like main business I'm not going
to be like watching people dancing and
buying cosmetics all of a sudden be like
hey let me search you know something
using AI there's no functional adjacency
so even though bite dance shelled out so
much money on marketing and brought in I
think uh I would have to find a number
but like couple like like here 20
million DAU do um which was quite
impressive in the beginning. They were
the leading kind of um product for for
the GBT like chatbot in China called
Dopal. They they never really took off.
Once Tencent integrated um DeepSeek into
their WeChat product, it was like game
over for them.
So I I want to I don't want to let my
original question go without going a
little bit deeper into it asking whether
Deep Seek was an outlier, whether
there's more uh that China is doing. So,
okay, I'm trying to like think through
our conversation. You've definitely said
that China is quite good at productizing
AI because there's this culture of open
source and getting it into the product
and building with consumer. But then I'm
thinking through like all right, so we
have the Deep Seek
breakthrough. Uh we know that Alibaba
has a pretty solid LLM or LLM in it's
called Quen. Um, but where what gives
you the confidence to say that China is
positioned to be a global force in AI
beyond the Deep Seek breakthrough?
Because I'm trying to think through like
what I mean I'm looking at the US, we've
got Open AI, we have Anthropic, we have
Google working on this stuff, we have
Meta working on this stuff. uh the
innovations that have um sort of been
central to uh this entire generative AI
moment have been developed within these
research houses. There is a widely cited
uh AI paper that that did come out of
China um that Yan mentioned on the show
a couple weeks ago. Uh but yeah, I just
am curious to hear your perspective on
like what am I missing? like what else
is happening within China that that
should lead us to believe that it's
going to be this global competitive
force in AI.
I think on a high level first I just
want to address there's like this point
where people are like it's a binary
thing you got a US in win or China win
right and I think for where I sit in
Hong Kong where really best we get best
of both worlds we have access to both
the west and the talent and kind of
information in China beyond the great
firewall right is that there could be
innovation in in parallel in these
worlds um and they might not be actually
directly competing each other in some
ways Like you said, I think for sure in
terms of frontier models, the US
companies, like you mentioned, the open
eyes and the topics of the world have
definitely led China's um LLM
development by far for a long time until
Deep Seek made a breakthrough, right?
And and this is something I've not shied
away from writing about where everyone
just assumed China's LM was lagging. Now
that conversation now is kind of changed
and it goes to your point where like
we've now reached a point where if your
LM is just good enough per se, then we
can see people build on top of it and
now we're seeing a lot of the um I think
strategy by businesses pivot into um
products and the products can be two
kinds like you mentioned. one is what we
just discussed. A lot of is in consumer
application and leaning on um the reach,
the distribution, the you know just the
wild garden kind of proprietary
information etc. right that we just
discussed. And then the other part I
think China is really leading on and is
being underappreciated. I think people
are picking up now is the physical AI
aspect. And this actually again it goes
back to if your LM is good enough. It
the competition there is not really in
can you innovate better in in the LLM,
but it's actually in can you produce
products in a costefficient way and
bring together the software and
hardware. And now because given that
China we just talked about it as well.
This is robotics. Robotics physical
robotics embodied AI physical AI and
we're seeing this right where
um it's not even like it's it's it's
like given that actually people in
industry are knowing know that China's
definitely leading this just given that
you know we've had 30 years in China
here where all the manufacturing is done
here all like I'm talking about not just
like thinking about your clothes right
not your cotton t-shirts or whatnot it's
like all manufacturing this is like your
um your scanners your your batteries,
your TVs, whatnot, anything and
everything, right? All of these things,
the the the mechanics of it, the
engineering of it, the cheap labor,
frankly, the cheaper product, the supply
chain, it all kind of brings it
together. So, we're seeing companies
like Unitry taking up I think like 40%
global share in
quadruplet, I can never say this word,
right? quadruplet um robots and like
something like 30% in humanoid robots
and they're like 1/3 1/4 of the price of
Boston Dynamics. Um so if you're looking
at things like this, it's quite
interesting and a lot of people have
been asking me actually even on my own
newsletter or um while I'm getting
interviewed is what about embody AI?
This is something where we're seeing um
companies like Chinese leading EV
companies bring in the Deep Seek LLM and
again because it's good enough we don't
need it to like right now nothing is
significantly better right it's good
enough you're actually advancing your
intelligence
um empowered cockpits so if you have
again renewable energy available
abundance of energy frankly which the US
right now is facing a shortage you have
manufacturing pess you have cheap
products and you existing pretty strong
like these like uh EVs and you embody
them with AI, they're really really
strong. So like I said, the AI
competition per se is not really just
like in LLM. I think in a very business
and pragmatic way, it's actually in the
product layer and China has a lot of
strength in the product sense. Now if
say in another year like LLM's develop
into a complete different thing where
like we cannot even envision what it
might be able to do then maybe the US
again will have the advantage of having
the knowhow right having the most
advanced um models to kind of be the
foundation of these products and China
will just be at this level right but if
we reach a point where everything is
kind of just good enough then I think
commercialization is also going to be in
product layer yeah and I just want to
say something about the nature of the of
the questioning here. Look, uh I'm not
um coming out and like attacking China
or anything like that. I'm just trying
to represent the conventional wisdom in
the US and to have somebody who's an
expert like you be able to like give us
give us that perspective just for the
listeners also like you know nothing
against the Chinese AI industry. So I
just think it's good to get these
questions out there and like start to
address them. Um, couple other things.
First of all, we have a conversation on
this podcast uh a couple times a month,
if not more, about whether it's the
product or the models that matter. So,
it's very interesting to hear you say
that China's perspective is it's
basically the products and these
companies are putting the product uh the
model innovation into play in the
products uh in in various ways because
we've seen very little consumer AI in
the US. very little and so they're
actually putting it into play which is
interesting. Um and not only that they
also have now built cutting edge models.
So you have both. Now look there's
there's definitely competition between
the China and the US. Um I won't you
know say that there isn't but it is it
is really interesting to to hear more
about this strategy. So Grace I'm
actually curious if you could share Oh
go ahead. I just want to add one point
like actually to my point on the um that
China is really focused on product they
have a strength in product but this only
changed after deepseat pre deepseat
frankly there was like no really like
global frontier leading model right like
even the Q-in series from Baba were
doing okay but there were nothing comp
like compared to the benchmarks you know
against the benchmarks um with American
leading models like I don't think anyone
was like oh I'm having a better
experience using QIN or Kimmy um
compared to like you know open eyes chat
GBT but I do think deepseek was a going
back to your initial question deepsee
was a huge milestone turning point for
China and I will use one example to
prove this point is again going back to
bite dance they didn't have a very
strong LLM they put a lot of money into
converting people into their do consumer
app it didn't do well cuz the experience
was just average and the
hallucin came out and that completely
changed a foundation of it became it
setting the bar for everyone having
access to good enough model then I think
the product uh product advantage could
really come through I do think there is
that so the first two years of since
GBT's u release since November 2022 the
US for sure by I mean US companies 100%
had a lead and no matter how we compared
it but I do think once we start shifting
our focus into product maybe Chinese big
tech will have certain advantages.
So what are they building in China? Like
what are the products that are coming
out with AI baked in? Um like I
mentioned you know obviously there's a
chatb style chat bots we have Kimmy we
have do we have like you know we have uh
like obviously deepse directly etc. And
then we have the
um we have a lot of like video editing
tools. There's a lot of video editing
tools like um cap cut there's um even do
itself the interesting bit is
multimodules so you can actually like
chat within the app having multiple
conversations open and a lot of them
will be thematic so you can have like a
love guru you can ask them about like
your dating life you can have someone be
like your therapy and you can be like
there's like a fake Elon Musk built into
it you can like pretend you're having
conversation with Musk like the consumer
entertainment aspect is very different
versus like I think the as a chat
chatbot um products are very efficiency
productivity focused like you know you
help with your research again it goes
back to like are you knowledge based
worker or are you like a you know um
blueco collar worker who actually wants
to use this for entertainment purposes I
think the vision of it was quite
different in the beginning um so yeah
I'm pulling up I'm pulling up this rest
of world article and they have uh all
these great examples um so for
automotive they say 20 Chinese
Automobile brands have announced plans
to embed uh DeepSeek models. That
adoption appears to be focused on
improving AI features including voice
control, high precision mapping,
smartphone like access to music, web
search, and messaging. Uh it's also
going into smartphones. China's top five
smartphone sellers, I think you
mentioned this, have all adopted DeepS
in their system updates. Huawei used the
R1 model to upgrade its Siri like ex uh
assistant. I don't know, maybe Apple
could use that because Apple
intelligence has been a mess. It's also
in home appliances. So, this uh this the
article says China's biggest home
appliance company uh Mia has launched a
series of deepseek enhanced air
conditioners. The product is an
understanding friend who can catch your
thoughts accurately and it can respond
to your verbal expressions such as I am
feeling cold. It's also in healthcare.
Near nearly 100 hospitals around China
have announced they will adopt deepseek
in their operations. The applications
include uh supporting diagnostic and
treatment processes, analysis of medical
imaging, quality control for medical
records and research on new drug uses.
Now, US companies are doing some
variation of this uh but not as bold and
not definitely not as quick. And there's
there's a care I think or a a caution in
US companies that you wouldn't see this
stuff get adopted so quickly. So what is
it about Chinese consumer tech and I
guess Chinese companies in general that
would create the conditions that you
would see such mass adoption in such
short amount of time?
I think it's it's really twofold. On a
very high level, the government has
really been encouraging this. So
honestly in a very top-down driven
economy when the government is pushing
incentives and encouraging various
industries to really um embrace the
technology I think there is like more of
a preference for companies say hey like
we should be part of this macro
countries movement right like countries
trend whatever there's that but then on
the other end actually um in general I
don't want to like I don't want to say
China doesn't have strong data
protection but I'm just saying in
general China's data protection kind of
regulations and um awareness is
definitely not as sophisticated or or
alert people aren't as alert about it um
when compared to the western markets
obviously you most most cautious maybe
market for um data privacy issues and
the US may be second and then if you
compare it to China people are a lot
more open to just kind of like sharing
their data online and I think because of
there's less of a concern or backlash on
that front companies are more able to
just say I'll adopt these uh these
technology new technology and there
might not be a lot of push back by
consumers now the third aspect is like a
bit more nuance and cultural is I think
when you look at China and the adoption
of internet and smartphones um it is
wild like I'm talking about like
95year-old my 95-year-old grandma
onchath sending me stickers 24/7
like you know they're so sophisticated
and it's it's like a world it's like a
complete like countrywide kind of
sophistication just because people
didn't really have PCs so like a very
small elite circle had PCs basically and
then everyone just adopted cheap
smartphones and now obviously everyone
has iPhones and Androids and whatnot and
because of that internet savviness I
think trying out new technology just has
a less of a like barrier in that sense
So I remember even a year and a half ago
I was visiting Beijing for a work trip
and I just popped into TCM doctor's
office for like acupuncture just as I'm
like generic shoulder pains right and I
was like this wasn't a big deal and then
the little like the young doctor she was
maybe like you know in her late 20s she
was quite new and she started asking me
she's like I know that you like you're a
tech reporter before like tell do you
know anything about this thing called AI
she's like I now use it for my diagnosis
I was this is crazy cuz in my head I'm
like first of all This is like you know
like TCM is you know what we think of it
is pretty traditional. It's legally call
traditional Chinese medicine. It's not a
high-tech industry but even like
practitioners in the TCM industry are
adopting AI to enhance their diagnosis
and their their workflow. And I think
it's really interesting because they
almost use as like a fun entertainment
tool because they're like I'll chat with
them and they'll tell me maybe it's
right, maybe it's not right, then I'll
assess again. And and I think just that
that sense of adapt like um that sense
of acceptance new technology is
something quite different where I find
even my friends in the US right now who
actually even there's one who works in
self we were talking about it recently
and he was so reluctant to try out AI.
He's like I don't need someone to write
my emails for me. I can do that myself.
And I'm like dude you literally work in
tech. How are you not using this to
enhance your productivity? He's like no
no no no. Like are they just like
polishing my emails for me? like you're
not trying the deep research like
products. You're not trying a lot of the
products that can actually help you cut
like one-third of your time. Um and and
I think there's that definitely
mentality difference. I think you're
hitting on a good point here. uh which
is that I think the conventional wisdom
in the US has been like why can't China
compete before deepseek the mentality
was why can't China compete on AI and I
think that everybody in the US knew that
the data protection was not anywhere
close to um at what it is in the United
States I also think that there was an
awareness that Chinese companies were
getting around export controls from the
US uh but definitely not to this extent
that people in the US are aware
like clearly there's a lot of GPUs that
have gone through Singapore and then go
right to China. Um and that is that
might be tougher to get to. Uh so data
so they they they Chinese companies had
more data. They had the compute. Um but
I think the thing that people in the US
underrated was the entrepreneurialism of
China. I think that China was basically
looked at as like you know a more
collectivist society that has less
entrepreneurialism. Uh but clearly there
is a lot of entrepreneurialism and maybe
that was due to the fact that China
clamped down a little bit on on
entrepreneurialism for a while and now
it seems like it's it's back in style.
Um but clearly I think that's something
that was underrated uh by west by wester
people in the west. I think definitely I
think you know look um I actually was
working at Alibaba during the clampdown.
It was quite a like wild ride um in the
sense of you know like the stock market
just you know and um we saw a lot of US
capital pulling out of China and to your
point actually I think
entrepreneurialism or entrepreneurship
in China was always quite um vibrant but
it was seen in a different way because
the last generation of entrepreneurs
were very different and it goes back to
um the ones born in the 60s7s a lot of
them and I'm not saying it's bad it's
just that their business model was this
happened in the US like there's Amazon
let's create an Amazon China but make it
like more Chinese right there's a
whatever there's a there's a WhatsApp
let's make a WhatsApp in China but make
it better or right so the mentality was
like can we copy copy and then innovate
right and I think um the Uber founder
said this on Alden podcast even quite
recent like a couple months ago he was
like when he was trying to fight um DD
in China back in I think the early
2010s. He's like he was it was madness.
The copying culture was madness. He's
like we'll come out with a new product
and the Chinese uh competitor will just
come out in another week and they're
very proud. They're like we just did it
better. And then like they'll come out
with a new product and they spent all
their R&D on it and obviously US
engineers extremely expensive and
Chinese talent especially then was just
frankly cheaper and they come push out a
new one. And at one point he's like
actually we just couldn't keep up
because they were able to turn out a
copycat version within a week or two but
then it at one point just turns to
innovation. So that was the last
generation of entrepreneurship and we
actually if you look back at it whether
it's Baba or Tensson or any of these big
tech companies their initial big pot of
gold per se were all American or at
least western. Okay actually like US
investing in China is extremely
extremely super vibrant. All private
equities had presence there like the
Warberg, the Hikar or whatnot. Okay. And
then all hedge funds had public like
publicly listed companies of ADRs. They
had positions in them. It's just that
during 2020 to
2023, China's internet um industry had a
bit of a went a little detour. Okay. And
that was caused by domestic regulatory
issues where there was a lot of anti-
monopoly probes and kind of
investigation went into the doings of
Alibaba and tensen etc. And there was a
bit and and the rise of kind of I guess
geopolitical tension kind of came to the
forefront of even business narratives. I
think there was an understanding before
where politics was politics but business
went on as usual before that. But during
that period, China US geopolitical
tensions actually became the front and
center of even business relationships
and we saw a lot of US capital pull out
and that therefore we actually did see a
huge dip in just the confidence in the
Chinese economy and honestly the Chinese
internet space for a little bit. Now AI
has definitely definitely powered it
again and even like you mentioned when
we were chatting you know like um like
Baba stock prices went shot up again
like you know like now people are
discussing is bu a good buy. I think
they become kind of like the weather
wayne of Chinese economy and Chinese
tech because the assumption is these big
tech companies can actually leverage
whatever AI innovation that comes
through given certain restrictions. US
private equity firms and VC firms still
cannot invest in Chinese AI companies
directly. The best way to get exposure
to this innovation is through public
investment into these Chinese big tech
right now. Yeah, it has been interesting
to see the the culture shift and uh you
mentioned you were at Alibaba during the
clampdown. I mean Jack Ma basically as
far as we're we know disappeared for a
while but he just made his way back into
Chinese public life. So that was a good
sign I think for everyone. like seems
like that's I don't have him on my speed
dial. He's not my best friend or
anything. But can you call bring Jack
into the podcast? You got to get him in
to tell us what happened. Uh but okay.
But it is interesting because you know
it it is it does seem to me and I think
this is important to talk about that
there has been
a after deepseek like you know people
saw deepseek and then we started talking
about a US context within China there
has been a a big change in terms of the
way that the government thinks about
tech entrepreneurialism and artificial
intelligence and it was really
interesting to read some of the coverage
when um China has this annual meeting,
the National People's Congress, and it
was all about tech uh and AI this year.
This is from Bloomberg. AI fever sweeps
China's political huddle, fueling tech
optimism. For some years now, China's
annual gathering of its national
legislature has been an increasingly
disciplined and cor choreographed
affair. Its muted vibes particularly an
echo of deepening concern about domestic
stagnation. Not this time. The National
People's Congress 7-day gathering came
on the heels of a breakthrough in
artificial intelligence by China's
homegrown startup Deepseek that's fired
up investors, politicians, and even
regulators. It also followed President
Xiinping's high-profile meeting with
business chiefs, including Jack Ma. In
this meeting, Communist part Communist
Party cadres from different regions
competed to market their loces as
China's next AI hub. Lawmakers made a
raft of proposals from promoting AI
related education and scaling up the
technologies application to boosting
research and regulating the social
impact. I mean I think people outside of
China need to pay attention to this.
That is a major major shift for the
government.
Yeah, definitely. I think there was um
well there's been a lot of like
municipal level or provincial level
policies throughout the country that's
really been boosting AI per se and like
I said physical AI especially in areas
like Judyang uh province or Guangdong
province where they're strong on
manufacturing they're traditionally
manufacturing hubs and they're actually
home to the overlooked kind of um the
tech companies like Jang is home to
Alibaba J EV well leading EV owner as as
well and GuangDong where is like DJI uh
sense time a lot of them are all in this
area the GBA area greater Bay Area what
they call um so there was that but this
time with with the AI policy push at the
two sessions it's something I think is a
really positive signal to everyone
because it just means that actually um
on a macro level the government is very
supportive of AI entrepreneurship like
you said and it wasn't ever like
explicit said that the government did
not support entrepreneurship. It's just
that the last couple years coupled with
co frankly it was quite tough on the
domestic economy and I think you know
when the world was so used to double
digit growth for like so long all of a
sudden when it came down to like 5% GDP
growth was there was a bit of a
readjustment of expectations whether
it's domestic or internationally. So I
think this just boosts that confidence a
little bit and I think again coupled
with what I said earlier that this
generation entrepreneurs have are a lot
more missiondriven than money driven I
find when when I read about them it it's
it's it's quite different like it the
whole atmosphere is quite quite positive
and it's really amazing to um speaking
of the entrepreneurs and sort of the
celebration of them it's kind of amazing
to see what's happening uh with the Deep
Seek founder uh Lang Languin Fang. Am I
pronouncing it? Yeah.
Yeah. I mean, you have to like we have
to talk about what's going on in his
hometown, which you mentioned, this
small village. Um this is from the
times. Uh the village where his parents
moved was uh was small and now sees his
fame as their opportunity. The village
has become a shrine to the man they
called top scholar. Uh he was a star
pupil who was the top of his class and
finished the high school math
curriculums three years early. Um the
street leading to his house has been
renamed Top Scholar Road. Um to serve
the tour parties that visit, they are
now selling Top Scholar bis biscuits,
Top Scholar hot dogs, and Deepseek
baseball hats. It's pretty amazing.
It it's it's really funny from a western
perspective, but I think if you if you
actually look back at Chinese like more
like recent history in the last even
couple hundred years, I mean except the
culture revolution like scholars were
actually a really celebrated and
recognized way to pick even like
politicians and like like leaders in in
the government way back, you know. Um
and it's like the Chinese society really
really values and they call this the dr.
So you can be the drum which means you
are the number one scored person of your
city or your province or whatnot and
these people will go on to go to the
like I said the bata which is a picking
you tinga university
etc. And even to this day, you know,
like it's still very celebrated if you
were like the number one of your city or
your province. And you definitely wear
with a badge of honor wherever you go.
um it's not that different from like you
know scoring number one in your high
school in the US like have like you know
we all had a validator and you know in
that sense but I think there's more
societal recognition and praise and
because of his frugal beginning because
like you said it's literally like a
village I think it just contrasts like
the kind of success he had it's it's
it's a bit more um it's widely
celebrated because China also celebrates
this idea of like upward mobility
because you know there's still such a
huge um actual um income disparity
within China. Like if you think about
China of the modern China, you think of
Beijing, Shanghai, the quality of life
is no different from New York or LA,
right? But if you think about actually
the vast majority of China out in the
third, fourth year cities, you're you're
actually looking at you know like like
still a developing nation, you know. Um
so so he came from that kind of
background and now to be one of the most
like celebrated ent entrepreneur in the
world even right like for this for this
innovation it's a huge pride for the for
the for the family and the folks back
home yeah you and I spoke before uh
before this recording and I mentioned
that I spent a day in China uh on a
stopover but did leave the airport and
that does not qualify me to speak at all
on China although it was a lovely visit
but as we left Beijing and went to the
Great Wall uh it I saw some I think I
saw some of those villages and it was
pretty pretty astounding to see sort of
Yeah. It's it's undeveloped and it's
it's very different from you know the
big cities where that that's a lot of
people's picture of China today. Yeah,
for sure. I think so. I spent um the
last nine years across Asia. So now I'm
in Hong Kong, but I've spent time in
Singapore, Beijing, and Shanghai as
well. And when I So for context, my
parents are from Beijing. So I did grow
up visiting Beijing here and there but
even then I never actually lived in
Beijing as an adult or really understood
this country and when I was a journalist
traveling around the country I was
astounded by the income disparity the
the awareness of the outside world. If
you talk to people on the streets in
Beijing Shanghai especially like college
educated kids um people are quite fluent
in English you know you can talk about
Obama you can talk about whoever and
they know about it really well. Like if
anything, Chinese uh school system
actually really focus on what's
happening in the world, outside the
world as well. Like that's part of their
education. But then you go to like a
third tier city and because my Chinese
wasn't that good, especially when I
first moved to China. Um and they're
just like people just come up to me cuz
I had a slight accent and they're like,
"Are you Korean? Are you Mongolian?" And
I've had these crazy questions like,
"Are you Arabic?" And I was like that
I'm definitely not any of those. But
like it's but it's just like this um
this very like siloed world of these
like little cities. They don't actually
even travel to other cities within the
country and they've definitely not seen
foreigners um and they have a very small
like lack of awareness what's happening
out in the world. So when we think of
China sometimes you need to think
there's the Beijing, Shanghai and
Shenzhins and then there's really the
rest of the country. Definitely. And I
guess from what I saw, if you drive
outside of Beijing, uh maybe not even
that far, you can be in that
countryside. Definitely. I want to take
a break because um well, we do need to
take a break, unfortunately. But I I
want to come back and talk about three
topics and then uh Grace, we can let you
get on with the rest of your day, but I
want to talk about robotics quickly. I
want to talk about development uh some
other developments including this one
area that has what you call these six
dragons and then if we can get it in a
minute about manis and agents. So we're
going to take a quick break and come
back right after this. And we're back
here on Big Technology Podcast with
Grace Shia. She is a Hong Kong based
writer and analyst. You can get her
newsletter AI prom on Substack. It's a p
r
oem.substack.com. And Grace, it's great
having you here. Thank you again for
fielding all these questions. You've
mentioned robotics a couple times. So,
let's just go there uh and spend a
little bit of more time. The videos I
see on Twitter of Chinese robotics are
insane. Uh they have a level of
adeptness, flexibility, awareness
um that is above and beyond I think
anything I've seen from coming from
Western. uh companies. Of course, you've
had Elon Musk with a person in a robot
suit. Okay. Maybe they've had some other
uh people piloting the Optimus, but um
the stuff that we're seeing from China
has been pretty pretty impressive. So,
give us the state of uh humanoid
robotics in China in particular. And um
and where where can we expect this to
go?
Yeah, I think um to even start with I
think in the start of the year at CES,
you know, there were people were saying
there were like more than half the
exhibitors were Chinese and majority of
them were actually robotic companies and
you know like you said there was
humanoid robots like the the one that
kind of went viral which is like the dog
the robotic dogs from Uni and then you
know you have um obviously the other var
variations of embodied AI products. I
think in China industrial robots not
been new and that demand for industrial
robots not been new right like we've uh
you know like everyone talking about you
know the aging population the
urbanization the reluctance to work in
uh manufacturing labors etc and even you
know the cost increase of human labor
that's been happening in China all of
these have driven a lot of companies
find industrial robot solutions and
that's been around for at least the last
5 10 years right but Now I think human
or robots in China is really interesting
right now is because um it it's very
rare for I think a place where you can
have the engineers to actually have
understand the software and then the
hardware and that's where China's kind
of I think strength in the last couple
decades of being a manufacturing hub
really plays into it now. um because
it's been the manufacturing hub for
basically all appliances we can think of
under the sun like any home appliance we
can think of right there's a lot of
knowhow of hardware um building is in
layman terms right in that sense and I
think beyond that now we can embed this
kind of um embed AI into it software
into it what was the word for it mechan
let me find it
uh mechatronics. So the the the
understanding of meatronics is something
that you know really China's had a huge
advantage in and even Jensen Hang
himself he was at HKD Hong Kong
University Science Technology in
November last year for a fireside chat
and he was saying look this GBA which is
a greater Bay Area has all the talent in
the world to lead in this technology
because you got the um headquarters of
DJI, you got the headquarter of Sensime
and this those companies kind of like
the last generation AI companies, they
really trained up a whole generation of
um engineers and mechanical engineers
who are able to really embed software
into hardware. Um so there's definitely
that and then on top of that because
it's been a manufacturing hub, it's just
frankly cheaper. Like the supply chain
is just here locally. You can source
everything here. So, we're seeing that
um you know, Uni Tree Robotics currently
account for about 70% of global sales of
humanoid robots, 40% of market share of
quadruplet um robots, and that's just
like a solid leader in the lead
industry. And I think their cheapest
model goes for something like
$16,000. So, it's crazy that you can
just have that. We're seeing other robot
companies like Galbot, etc. Like all of
these coming out as well. They're
selling them as like nanny robots. you
know, they can help you, you know, clean
your dishes and throw your garbage out,
whatnot. Um, it's really crazy how
advanced these technologies are in terms
of their abilities of movement. My
concern is right now actually the lack
of regulation or safety um kind of
standard for them. Um, as a mother of a
young kid, I would never ever bring on a
robot to my household because you just
don't know if it can go rogue, right?
And I think that's kind of the current
discussion around this um globally as
well as in China really why it's not
really gone to mass consumer market yet.
That is fascinating. I think just the
advantage that you're talking about is
you know it's you've elucidated so well.
It's just the fact that China has the
hardware and the software and being able
to be so intimately familiar with both
just gives you a massive leg up when it
comes to humanoid robotics. That's
that's fascinating. I want to go now to
a we talked a little bit we had people
who are interested in like where is uh
some of the other innovation within
China coming from outside of um outside
of deepseek and we've talked a little
bit about the productization we've
talked a little bit about what Alibaba
is doing and I think it would be
instructive to go through this example
of the six dragons which are all coming
out of this area called Guangjo and
Grace if you could tell us a little bit
about what's happening there and what
these six dragons are. That would be
great. Yeah. I think to first to
understand is Ho is considered a second
tier city for again China standards but
the population is like 9 million people
or something like that. It's nowhere
like the biggest city in the US just FYI
folks. It's it's not a small city you
know. So in that sense it's not it it's
quite well known in China and I think mo
oh sorry it has actually nearly 12
million people including its suburban
area. So so yeah it's not 150% the
biggest city in the US. Okay no big
deal. Um so in that sense Ho is it's
quite fascinating. So I want to start in
a cultural context. Ho is a really
really really pretty city actually. It's
um named the paradise paradise on earth
um city of silk. It has a west lake. So
in historical times, Hjo, the whole area
is kind of where the scholars used to
kind of come together. You know, they
would like admire the beauty of like
nature and write poetry. And you know,
it's also known as where like the
prettiest girls came from. It's it's
it's really kind of well um documented
in Chinese traditional history. Um
modern days, why is it relevant? Well,
it actually is the home to Alibaba as
you mentioned, home to Zeli, um, you
know, the conglomerate of car that also
owns a lot of the EV brands like uh like
Ziker. Um, it is also home to Net Ease,
the massive gaming platform. So, itself
is like actually like a home to like
some of the biggest publicist companies
in China. Okay. And then on top of that,
it's been traditionally uh a
manufacturing hub because I think the um
the proximity to Alibaba and obviously
then the like the rampant and vibrant
e-commerce space. We have a lot of home
businesses around the whole area where
they're the manufacturers of like the
clothing, the textile, the toys and
whatnot that you know of. Okay. Um even
near like the fake bag places and the
the toy factories and all those things.
So, so it's a really vibrant economy.
Now, couple now on top of that is I
think a really special thing. I was kind
of thinking about it that day is it's
about a two-hour drive from Shanghai.
So, meaning it has access to
international talent capital um the you
know the most vibrant culture and like
art art scene in in China. And then it's
kind of far from Beijing. It's like a
three-hour flight, two three-hour
flight. So you're kind of away from the
regulations, you know, like Beijing is
and Shanghai are very different vibe.
Like think about New York and DC, right?
You go to DC, people dressed in suits
and and and it's different vibe, right?
People are more serious. You go in New
York streets, everyone's wild, right?
They are in their own world. And kind of
like Shanghai and Beijing is a bit like
that, okay? Right. Like I I remember
when I first moved from New York to
Singapore, I was like, "Oh my god, where
are the crazies? I I don't see anyone
crazy anymore." And I miss that energy.
But my point is um I think if you think
about that HJO is so strategically
located to actually help entrepreneurs
flourish especially smaller businesses.
So it's not as expensive Shanghai
Beijing. It's not as regulated and it
still has all the talent because of the
all the big companies that's been there.
It's brought in this whole ecosystem of
people that they can people lean on. So
now because of that the local government
is also very very pro businesses um and
they've really pushed out a lot of
policies to incentivize AI development.
you've really seen a lot of um AI
companies really flourish there to start
off with deepseek himself right like
Deepseek itself is actually from HJO and
Unitry what I just mentioned the the
leading robotics company also based in
HRO and it's really interesting actually
out of the six small dragons three of
the CEOs were educated at Jodang
University which again I just mentioned
is a very very strong has a very very
strong STEM education um so this whole
ecosystem system has really served it
well. Um, and now beyond Unutri Deep
Sea, there's a company called Game
Science, Many uh Many Core, which just
went public in Hong Kong, and uh, Brain
Co, and Deep Robotics. So, they they
each kind of work in different realms of
AI, but supporting the whole layer like
the whole full stack and infrastructure
and the ecosystem of of the AI
development. And I don't think we need
to go into details of each of them, but
I just think it's really interesting
that people are kind of overlooking this
whole area of development. Um, and I
think 10 years ago, if you ask where is
China's startup hub, it would have been
Shenzhen, which I mentioned was like the
hub of the headquarter of Tensson, DJI,
Sensime. Now, it's actually up there in
Hung. And it's really interesting
because that hustle culture,
entrepreneur culture also really feeds
into each other and I think it's making
it more and more vibrant. Yeah, I looked
at a photo of it as you were talking and
we'll confirm to our audience. It is
beautiful. So, it's a very pretty city.
Yeah. my next trip to China. I think we
have to add it for sure. All right,
we're running out of time. Oh, go ahead.
Oh, it's actually also the famous tea
plantation. So, if you ever see China's
like tea plantation pictures, it's out
in Hungro suburbs. Okay. All right. So,
definitely adding it to the list. Uh, we
are running out of time, but I don't
want to leave here without getting your
perspective on Manis, which is this AI
agent that has sort of taken I don't
know, taken the world by storm. Maybe
that's a little bit too strong. uh but
it is another agent outside of China uh
or coming out of China that people are
paying attention to. Some people said is
this the next deepseek? Uh do you think
is is China all all about the ai agents
now speaking of putting the stuff into
products and what did you think about
manis?
Um so full disclosure I haven't actually
tried it so it's all secondhand
information. It's like the most
exclusive product. There's like, you
know, 1% of people being led out and I
think invites to Manis are going for
$1,000 on the black market. It's crazy.
Okay. We were paying like 10,000 USD for
this right now apparently. All right.
Even my like a lot of my like media
friends and tech friends out like based
in mainland, they're not being they're
not able to access. So, it's a bit
secretive. So I don't have a really
strong take on it because I don't think
it's a deepseek moment for China. Um
because I think it did not fundamentally
change how we believe AI is being
developed right like in the sense that
deepse completely challenged the scaling
law and what we believed in more capital
use better training better model equals
better AI like this kind of mentality.
Um, I don't think Mattis has disrupted
anything we already believe in, which is
AI will one day become like an assistant
or a extension of our our capabilities.
And I think if you've used like OpenAI's
operator, etc., like it's not that
different. And frankly, what I've heard
from people that have tried it, it's
still glitchier than a demo. I think
it's an interesting thing because it
does show that means if like you know
inference becomes cheaper and usage
becomes cheap usage of AI becomes
cheaper more and more smaller businesses
like people with less money can actually
jump into this game and compete and
innovate on top like we talked about
innovate in the product and application
layer but does it really change
fundamentally how we believe AI is
developing? I think it just actually um
reinforces what we've been talking about
the mainstream narrative which is EI is
going to take over junior staff jobs um
essentially that way right or a lot of
administrative work um and that's really
scary um and so how do we train our next
generation of young talent to be AI
native right that's something where like
the conversation might lead to um but
yeah I don't think it's deepseek moment
but again I've never tried it myself so
maybe I'm wrong maybe this thing is just
like crazy okay um last thing that's
been coming up as we've been talking
people we haven't really talked about
the export controls and the shortages of
GPUs within China and you know we could
get into the advisability of the policy
but I think that'll be for a different
show uh to me the question is something
like manis right we just talked about it
1% off the weight list it's probably
compute heavy uh something like deepseek
right they had to cut off access I think
for to either access to the API or
access to the product um because there
was such high demand uh so I'm curious
curious from your perspective, can
Chinese companies keep like really
become global forces if they're going to
be limited uh by the effectively I guess
the chip shortages? Uh we have everybody
has the chip shortages, but I think they
have them more acutely and that might
limit people's ability to you know
effectively use the products. So what do
you think about that? It's definitely
not going to be easy and with Deep Seat
rolled out, everyone's using it. I don't
know if you you tried it, but the server
would always say like it was like busy
or like it couldn't essentially the
server couldn't actually serve the
amount of interest coming in. Um so I
think it's not easy and deep sea trained
on the H800's right and so that that was
designed to circumvent the the chip
bands in 2022 I believe. So, so all of
these things basically like we can say
like it for SteepC to innovate beyond
using the best top tier kind of newest
chips but right now it's definitely also
hindering Chinese like you know AI
development for sure that's probably it
biggest bottleneck because we talked
about it it has innovation it has data
it has energy it doesn't have chips so
unless domestically you know we see like
Huawei or Baba or any of the other chips
um players really make a breakthrough.
This is going to be their biggest
hindrance, I think, for sure. All right,
Grace, look, it's late my time. It's
early your time. I was like, uh, am I
going to be able to stay up for this
interview, folks? We're pushing uh past
10:00 p.m. Now, I I have not felt tired
for a moment. This has been fascinating.
Thank you so much for coming on and
sharing all this fascinating perspective
on AI in China. Can't thank you enough,
Grace. For sure. Thank you so much,
Alex. And thank you for staying up late
for this. Yeah. Thank you for uh
starting your day with this as well. I
know. Yeah. Beginning of the day with
the podcast is not always the easiest.
So, thank you for that. The uh Substack,
we'd be remiss if we didn't call it out
one more time. It's a prom. Uh on
Substack, a P
r.substack.com. Many of the stories that
Grace talked about today are on her
Substack. So, we do encourage you to go
check it out. Thank you, Grace.