Japan’s politics

People noticed the market’s volatility on Aug 5 with the yen carry trade, but few discussed the notable changes among the next Japan PM candidates & the pressure on current PM Fumio Kishida.

In early Aug, Kishida seemed failed to secure his party’s support. The meetings on Aug 2 (Friday) between party “leaders” seemed to be a really important time. Other news pointed to the same direction around those days.

The LDP’s internal election dynamics, especially factional operations, determine the outcome. And Taro Aso seems to be a very powerful figure. He is the “King Maker“.

Currently, there are 9 candidates within LDP. We shall know the vote by end of Sep. If no one gets over half of the vote, which is almost certain, then the two with the most votes will enter the runoff. In this runoff, the LDP lawmakers will have a much greater voice so they are determining in the end.

The winner shall be the next Japan PM.

 

New era for Chinese liquor

1/ Post pandemic, the growth rate of various food consumption products did not meet the expected rapid increase. Additionally, government consumption is strictly limited, reducing the scenarios that support alcohol consumption, leading to insufficient effective market demand.

2/ Under the guidance of leading brands like Moutai directly increasing sales volumes and strictly controlling retail price hikes, the price increase room for high-end to sub-high-end products has been compressed, and the main business revenue and profitability of well-known products have declined.

3/ Key enterprises and traditional key production areas continue to expand brewing facility construction, and the production capacity of solid-state raw liquor is gradually being released. Inventories of raw liquor and finished liquor continue to increase.

4/ Well-known brands are accelerating their expansion to broader sales regions and nationwide layouts, penetrating into third- and fourth-tier markets and through omni-channel, both online and offline. The regional brands’ baijiu market has visibly shrunk.

5/ After the revised standards for beverage alcohol classification and baijiu industrial terminology were issued, they affected the raw materials of liquid baijiu. The labeling of food tags has increased the cost of some low-end baijiu.

6/ The adjustment of the national industrial structure catalog lifted restrictions on baijiu production lines. Various regions, especially traditional baijiu-producing areas, have successively approved the establishment of a number of new baijiu enterprises, leading to a slight increase in the total number of baijiu enterprises. However, due to the current macroeconomic environment, the number of large-scale baijiu enterprises has not increased but decreased. We have now entered an era where there is neither a shortage of alcohol nor a shortage of good alcohol.


see the other post for Chinese Baijiu consumption – already quite high volume, hard to grow, likely decline IMO.

Nvidia revenue problem

While investors mostly focus on below whispered q3 guidance, or that revenue beat is smaller than recent quarters, a less discussed problem seems to be declining contribution from the US.

Singapore revenue contribution to Nvidia is rising qoq to 19%, while the US declined to 43% from 52%.

In absolute number, US revenue declined from $13.5bn to $13bn and Singapore rose from $4bn to $5.6bn.

Sure Nvidia says it’s not an indication for the end location. But Malaysia/Indonesia is likely, where end users include ByteDance or US CSP services (whose end user is ByteDance).

See the other post for ByteDance capex – it’s probably $10bn+ per year on data centers, or 1GW of IT load assuming $1mn per MW.

US CEOs need be careful about having a popular book in China

This is interesting: when I think of recent popular books in China written by or about US CEOs, it doesn’t seem to be good!

1/ Bob Iger’s book: The Ride of a Lifetime (一生的旅程 in mainland China)

  • published in 2020 (same year in China)
  • Disney stock now is only half of end of 2020 price

2/ Books about Elon Musk

  • one by Ashlee Vance in 2015 (硅谷钢铁侠 published in China in 2016) & one by Isaacson Walter in 2023 (埃隆·马斯克传 published in China in 2023)
  • in China, interest about Musk leveled up in 2021 (above average), per Baidu Index

  • If you compare current Tesla stock to end of 2020/2021, it’s down ~15%/~40%

3/ Ray Dalio’s book: Principles (原则 in mainland China)

  • published in 2017 (2018 in China)
  • Returns from Bridgewater for the past few years are not amazing (read this note)

Personnel changes among China’s top financial regulators

1/ A new anti-graft sub-committee was created for the financial sector within Communist Party’s Central Commission for Discipline Inspection.

Wang Weidong heads this committee/working group.

2/ CSRC vice chairman (similar to SEC in China)

Li Ming replaced Fang Xinghai.

3/ National Financial Regulatory Administration’s (NFRA) Shanghai branch head

Qi Xiang replaced Wang Junshou.

4/ National Financial Regulatory Administration’s (NFRA) Beijing branch head

Pei Guang replaced Yang Dongning.


On a separate note, in recent days it was announced that

Dong Guoqun, an executive vice president of Shanghai Stock Exchange was charged / investigated.

Chen Xiaopeng, CSRC Shenzhen branch former head was charged / investigated.

How much should ByteDance spend on data centers?

ByteDance has deep wallet: in 2023, it generated $40bn+ EBITDA.

To compare, Meta had over $60bn in EBITDA during the same period.

Meta will grow to $80bn ebitda in 2024, while capex guidance is as much as $40bnhalf of its EBITDA.

Assuming half of Meta’s capex goes to data centers -> $20bn. And half of that goes to GPUs > $10bn; building out those data centers etc. will be $5-10bn for Meta.

ByteDance is also growing, say at a similar rate as Meta, so $50bn+ EBITDA in 2024. And assume ByteDance invests at a similar ratio as Meta, then it needs to spend ~$25bn in capex, or ~$12bn (or 60% of $20bn) for data centers$6bn  for GPUs, and $3-6bn to build out data centers etc.

USDJPY

Just using interest rate parity.

Before the interest hike cycle, in early 2022, 1 dollar = 115 yen.

To make it simple, assuming USD interest rate is 5% and JPY is 0%.

If you exchanged 115 Yen [$1 worth of JPY] for USD and held USD for for 3 years, you should get $1.05^3.

If you hold Yen for 3 years, you shall have 115 Yen. So 115 Yen = 1.05^3 USD

1 USD = 99 Yen…


Real life is not like that.

In the real world, people exchange JPY for USD to earn higher yield, which created demand in the short-term.

That portion of USD can be deployed in the equity market, buying S&P 500 for example.

In this case, if you borrow Yen, you pay very little cost of capital (near zero), and you earned S&P 500 return.

Meanwhile, as everyone wants to do this kind of trade, and due to this imbalanced demand in the short term, USD will appreciate against JPY << this means you earn extra return when exchanging USD back to JPY (1USD = ~160JPY a few weeks ago, vs the 115 exchange rate 3 years ago).

In sum, you get S&P 500 return, you get USD appreciation return (~40% in 3 years), and you pay only JPY interest rate.

This is crazy.

Why would this happen? against the interest rate parity?

The other side of the trade seems to be those who are bearing the low rate of return on JPY… The savers in Japan I shall say.


this is also happening again in China?

The victims are obvious.. savers who are only paid 2-3% in China.

An underrated growth driver for Spotify

Not sure if it’s been widely seen as a growth driver, but Spotify is actually not blocked in China!

Yes, you can use Spotify w/o VPN in China, including podcasts.

If you are on a short trip to China, you can use Spotify for 14 days (no payment needed, just regular account). This corresponds to the visa-free 15 day stay in China for several countries – that list is expanding.

If you are staying in mainland China longer, you need a premium account. With Spotify Premium in mainland China, you don’t need VPN.

This is rare – for any “foreign” internet service provider to provide service in its original format.