Glenn Profile picture
Co-founder/BoD @HealthCareInc | Previously @Catalyte_io | VC/PE @Investcorp Technology Partners — Tech | Econ Development | Investing | China/APAC
14 subscribers
Apr 1 7 tweets 2 min read
What has struck me about today’s youth is the change in aspirational ceiling.

20-30 years ago Chinese youth dreamed of one day maybe reaching “First World” level quality of life.

Today, its youth aspire to something beyond. You are starting to see this in the next generation of entrepreneurs, the problems they are trying to solve and the way they talk about the future. Contrast this with the last generation of entrepreneurs, like Ren Zhengfei.

He has always looked up to the West for inspiration: you can even see this in the architectural design of the Huawei Dongguan campus which was inspired by European architecture.

euronews.com/culture/2024/0…
Mar 23 19 tweets 4 min read
Vietnam is now running an electronics net export balance of >$100B per year.

China’s comparable figure is ~$650B.

That means on a per capita basis Vietnam is now exporting electronics at over double the intensity as China. When people who ask “who is the next China?” The context is usually figuring out where *labor-intensive manufacturing is going.

The answer was “inland China” for a few years after the GFC but today the answer is a medley of countries that are replacing China in the aggregate.
Mar 21 31 tweets 10 min read
A key difference between China and India as a destination for manufacturing FDI is that China made it attractive place for foreign MNCs to make tons of money.

India has not: returns on FDI generally for foreign MNCs has been mixed at best in India. Companies like Apple, VW, BMW, Nike, Tesla, GM, Samsung etc. have made *trillions in FDI income in China cumulatively over the years, mainly selling to Chinese households.

HK and Taiwan-funded exporters have also made *trillions through the years by setting up export processing factories in China.
Mar 17 7 tweets 2 min read
Embedded in this post is an implicit assumption that there is a single engineering development path to achieve an end goal, and/or that the end goal is to specifically replicate what ASML has achieved. The underlying engineering objective here is NOT to recreate EUV “the way ASML did it” (although that is certainly one approach).

It is to print increasingly finer patterns on a semiconductor substrate in a process optimized for key vectors like cost, scalability, defect rates.
Mar 11 13 tweets 4 min read
China is flipping traditional macroeconomics on its head because its policy focuses more on elasticity of supply than elasticity of demand.

e.g. macro folks equate low inflation / deflation to lack of demand growth, whereas in China it is really more driven by supply factors. Where economists have a tendency to focus more on how price signals (driven by demand) ➡️ changes in supply, much of China's rapid catch-up development has been from making sure supply/capacity runs ahead of demand (i.e. does not become a developmental bottleneck). Image
Mar 6 23 tweets 6 min read
As I wrote here about the palpable shift to a "looser" fiscal environment, China is making a pivot into a "kinder, gentler" phase, at least economically speaking.

But there are other key signals as well, such as calls by Beijing to address excessive neijuan 内卷 ("involution"). Li Qiang recently discussed the importance of addressing "excessive, self-defeating competition" at the Two Sessions.

This is significant because it is the first time it has been officially discussed at such a high level.

business-standard.com/world-news/exp…
Mar 5 5 tweets 2 min read
It’s under-appreciated just how much cash/deposits Chinese households have been building up / sitting on.

It dwarfs holdings of *speculative real estate by 3x and stocks/bonds by 4-5x.

There is tremendous liquidity sitting on the sidelines just waiting to be activated … The question is … when risk appetite / confidence shifts … will they spend it or invest it? (Hint: both)

And the spending portion … will it be on upgrading their apartment? Cars? Clothes? Travel? Education? Luxury? Healthcare?

When is it going to shift? Has it already started shifting?

What does the HF data tell us? What do investment flows (e.g. southbound connect) tell us.
Mar 4 16 tweets 4 min read
It’s striking how differently China is approaching the “compute” segment of the value chain.

Unlike privately owned and funded model that dominates outside China, the state is investing heavily in computing clusters as a *public resource. I suspect there are several structural / philosophical reasons for this.

First, nearly all of the power infrastructure is also managed as a public resource (via SOEs).

Electricity / grid access is one of the three critical success factors for these clusters.
Mar 3 17 tweets 4 min read
Been saying this for a while now: China dgaf about U.S. tariffs this time around. One of the ironies about the last round of tariffs is that they forced China to accelerate something they should have been doing anyway for the long-term good of its economic development: reducing the labor-intensive export processing sector.
Mar 1 20 tweets 6 min read
What’s more likely?

The company that pushes out its methods onto open source is lying about its compute needs

Or research analyst guessed wrong on the number of chips they inputed into an excel table Six consecutive days of open source code releases validate the OoM efficiency and cost improvements …
Feb 28 12 tweets 2 min read
When outside pundits opine that China’s savings rate is “too high”, people should realize they are effectively complaining that China is building too many houses and too much infrastructure for its own people.

Why do they care so much how China chooses how to develop itself? Savings ~= Investment (gross capital formation, GCF)

Housing and public infrastructure account for 50-60% of total GCF and accounted for more than 100% of the growth of GCF since the GFC.

(it’s only been the last 4 years — since 3RL — where housing has declined) Image
Feb 27 26 tweets 6 min read
Was having a DM conversation on "breakthrough invention" with Noah.

I had not originally counted "the drone" as a Chinese invention but ultimately it depends on how you define "invention":

Don't need to be 1st, but need to be early & achieve commercial success. The conversation delved into the deeper topic of incremental improvements vs. true "breakthrough innovation" and where China fit on the curve.

@ruima discusses the topic of incremental innovation here:

Feb 26 19 tweets 7 min read
1. It is Argentine shortfin squid (illex argentinus) season. We know these boats are looking for shrimp from the lights (squid are attracted to the lights).

2. Argentine shortfin squid habitat includes deepwater (i.e. outside EEZs).

3. Spain and Italy are the largest importers of squid. The U.S. is also a squid importer.

4. Deepwater fishing is hard work, dangerous and requires a certain boat specification.

5. China has leading share of both (i) these boats, (ii) and work crews willing to take on this hard and dangerous work.

6. Chinese work crews and/or work crews working on Chinese ships capture most of the world's shortfin squid to satisfy demand from the net importers.

7. People from squid-importing countries suggest
sinking these boats and killing these work crews. 1. It is Argentine shortfin squid season

2. Argentine shortfin squid aggregate on the seafloor to depths of 800 meters https://www.seafoodnews.com/Story/1292255/Illex-Squid-Fishing-Season-to-Open-Early-in-January-2025
https://en.wikipedia.org/wiki/Illex_argentinus
Feb 25 12 tweets 3 min read
The challenge with re-industrialization is reality that the U.S. will need to re-learn how to walk before it can run again.

Competitive mfg is incremental and cumulative; leapfrogging is rare.

People sometimes think China leapfrogged. No, it had to learn step-by-step. Take my favorite topic HSR as an example.

China did not leapfrog into HSR.

1st it learned how to build slow trains + basic tracks. Then it learned building bridges and tunnels. Then it learned how to speed them up. Finally, it started building HSR.

Feb 23 20 tweets 5 min read
It's really quite simple:

We can trust Brad's alternative estimates when he can he can definitely point out evidence where trillions of dollars of hidden capital outflows have accumulated in global asset markets, thus far undetected. This is the undisputed implication of claiming SAFE has been understating its official CA surplus by $500-600 billion per year since 2021.

I raised this in October ...

Feb 19 18 tweets 5 min read
Remember all the people who said China wouldn’t be able to make the shift away from the real estate-driven economy? This transition was possible in part because of the continuously underrated migrant “floating” population.
Feb 14 16 tweets 3 min read
There's of course a certain level of irony with rumors of 50/50 JV with TSMC being considered with "tech transfer" when juxtaposed next to complaints about China's mandated use of 50/50 JVs and tech transfer, e.g. in the automobile sector in the 90s and 2000s. Instead of correctly viewing it dispassionately as a rather effective structure to transfer tacit knowledge in a "learn by doing" activity like manufacturing, the 50/50 JV structure itself was demonized and associated with allegations of IP theft and unfair trade practices.
Feb 7 20 tweets 5 min read
Rising U.S. goods trade deficit is (in part) a mirror of the rising power and profitability of U.S. MNCs that generate large profits offshore that are accumulated in tax havens.

You can see this by walking through the balance of payments. Everyone knows that the U.S. goods deficit has been growing.

The CSIS (Pettis/Setser) explanation is that this reflects imbalances in the global economy, e.g. China “overcapacity”.

The problem as I have been saying is that the goods trade is only a subset of trade.
Jan 31 32 tweets 8 min read
This key chart from @SemiAnalysis_ appears to have been the key source for claims of "50,000 Hoppers" and more detailed disclosure on their CapEx buildup analysis ("$1.3B").

But the table has errors/inconsistencies. More significantly, key assumptions don't pass sanity checks. https://semianalysis.com/2025/01/31/deepseek-debates/ 1⃣ First thing you might notice is that it says 60,000 in the Total column.

But the A100s aren't "Hoppers". So the 50,000 is just the last three columns.

Ok, so far so good. Image
Jan 29 22 tweets 5 min read
One focus area for DeepSeek was going "lower level" with PTX.

Here's a nice explainer on CUDA/PTX 👇.

While this is an admittedly arcane/technical subject, it actually provides critical insights into:

1⃣ the development cost question
2⃣ export controls
3⃣ future trajectories First some quick background: PTX is a low-level coding language analogous to Assembly.

This is "deep" software engineering: you are now talking in the native language of the machines, with the thinnest of abstraction layers.

I discussed more here, too:

Jan 28 32 tweets 8 min read
If people freaked out about DeepSeek wait until they hear about how it is scaling inference. “Third of the cost” is

1) Likely conservative, and

2) Didn’t require EUV