Blog

Solar, Baijiu, and Banks

From 2021 to today,

Solar in aggregate went from 3 trillion rmb to 2 trillion rmb;

Baijiu in aggregate went from 5.4 trillion rmb to 2.4 trillion rmb.

These two sectors along lost 4 trillion rmb in market cap.

Guess who gained?

Banks in aggregate went from 10 trillion rmb to 14 trillion rmb in market cap, gaining 4 trillion rmb in about 4.5 years.

Solar and Baijiu are more private companies.

Banks are more SOEs.

Solar is a representation of how involution destroyed value;

Baijiu is a representation of how declined consumption/macro pressured stocks.

2026 vs 2021 stock market bull run

2021 hot stocks = everything tied to cheap money + retail speculation:

1. Meme stocks: GameStop, AMC, BlackBerry, Bed Bath & Beyond. GameStop became the symbol after the Reddit short squeeze in Jan 2021.

2. EV / clean energy: Tesla, Rivian, Lucid, Fisker, Nikola, QuantumScape, ChargePoint, Plug Power. Many EV SPACs later collapsed 70–90% from peaks.

3. SPACs / IPOs: 2021 was peak IPO/SPAC mania. Big names included Rivian, Coupang, Roblox, Coinbase, UiPath, Affirm, Robinhood. 2021 had about 980 IPOs globally, roughly double 2020.

4. Cathie Wood / ARK-style growth: Zoom, Teladoc, Roku, Square/Block, Coinbase, Palantir, Unity, Twilio, Shopify. ARKK was the “disruptive innovation” proxy, but by late 2021 it was already down sharply as rates rose.

5. Crypto / fintech: Coinbase, Silvergate, Signature Bank, Marathon Digital, Riot Blockchain, MicroStrategy, Robinhood, Block.

6. SaaS / cloud: Snowflake, Datadog, Cloudflare, Asana, Monday.com, Atlassian, CrowdStrike, Zscaler.

7. Stay-at-home / reopening extremes: Zoom, Peloton, DocuSign first; then airlines, cruises, casinos, hotels during reopening rotation.

Meme, crypto and EV have no real earnings.

Even SasS/Cloud depends on future projection of rapid revenue growth, not near-term earnings.

The 2021 bull run is more like internet bubble in its essence. 

And when interest rose, bubble burst.

2026 is different – it’s more about tech advancement.

In the current bull run, most companies have real near-term earnings and interest rate is 3%+.

Current interest rate is more in-line with inflation, while 2021 interest rate is obviously lower than inflation.

However, space and quantum can still make me feel more bubble-like.

US treasury vs US market cap

From 2016 to 2026,

US federal gross debt grew from ~$19 trillion to ~$38 trillion.

US total market cap grew from $25 trillion to ~$75 trillion.

Debt doubled and market cap tripled.

That is probably higher valuation but also can mean better structure of the economy, better competitiveness, better margins.

A rough decomposition of the gap might be:

  • ~50–60% from real earnings growth and improved business quality.
  • ~40–50% from valuation expansion and expectations for future AI-driven productivity.

Value trap and being contrarian

Consensus can be right for a long time — ride it while reflexivity is reinforcing it, but be ready to turn contrarian near the inflection point.

It is not “always go against consensus.”

George Soros says the crowd is right 80% of the time.

“Most of the time I am a trend follower … Only at an inflection point are we rewarded.”

His key idea is reflexivity: market prices are shaped by investor bias, and those prices can then affect fundamentals, creating boom-bust loops. So the contrarian opportunity is when the market’s “prevailing bias” has gone too far and starts to reverse.

Being contrarian for most of the time is dangerous.

Being contrarian in investing for too long can be dangerous.

On the upward trend, if you are against the trend and short the stocks, Tiger Mgmt is a case in point – Tiger Management, run by Julian Robertson, effectively shut down in March 2000, right as the dot-com bubble was peaking.

Robertson was acting like a classic contrarian:

  • “These internet stocks are absurdly overvalued.”
  • “Eventually fundamentals will matter.”

Both statements were true.

But Soros would likely argue that Robertson fought the trend too early. A reflexive bubble can become much larger than valuation alone would justify.

On the downward trend, if you are against the trend and long the stocks, there is a term called “value trap”.

Sometimes the crowd is right: the business is structurally deteriorating, ROIC is falling, or the terminal value is lower than historical multiples suggest.

For value traps, the key question is not “is it cheap?” but:

What changes the prevailing trend?

In the case of investing and cigar butt investing like what Warren Buffett did in his early days, he bought near/below liquidation value.

In addition, he had an exit path.

Cigar-butt investing often worked through liquidation, tender offers, buybacks, asset sales, control pressure, or mean reversion. It was not “this bad business will become great.”

So Buffett avoided value trap with asset/liquidation value protection + catalyst/control.

That is even harder in China, as there is little space for an activist in China investing.

Top buyers of gold

Central banks buying gold is a theme for bull investors in gold.

But that doesn’t automatically become a reason to buy. Several things to know before buying.

1/ Who exactly are buying?

I looked it up.

In 2025,

Biggest central bank buyer Poland: 102 ton

2nd place Kazakhstan: 57 ton

3rd place Azerbaijan: 53 ton

Down the list: Brazil, Turkey, China, India, etc.

Surprisingly, most developed countries are not on the list.

European central banks are net sellers.

2/ Net-net, are they buying more

In 2025, they actually bought less than 2024, about 20% lower in tons.

3/ Why do they buy?

The interesting surge happened in 2022.

This coincides with US kicked Russia out of SWIFT etc.

This explains the difference – if you are a retail investor living in OCED countries, you probably don’t need to buy gold vs. some central banks must buy gold to de-risk.

Munger said before, gold is “a great thing to sow into your garments if you’re a Jewish family in Vienna in 1939.”

If you are not trying to relocate out of a turbulent country, there is probably not much reason to buy gold.

置身钉内？

一个方向看似正确的 AI 办公产品，如何在组织目标、商业压力、产品基因和真实用户需求的拉扯中走偏？

ONE 想做“让事找人”的 AI 工作入口，但最终被做成了“让组织更容易看见、追踪、管控人”的工具；它失败的关键不在 AI 技术本身，而在产品定位、组织机制和用户立场错了。

核心观点：
// 产品发心过载
ONE 同时想服务用户、证明钉钉 AI 化、承载无招回归后的战略战役、消化商业化和集团指标。目标太多，导致产品无法保持一个清晰主线。

// To B 的根本矛盾没有解开
钉钉的付费方是老板/管理者，真实高频使用者却是员工。老板要“可控、可视、确定性”，员工要“边界、缓冲、自由度”。ONE 试图同时满足两边，最后实际更偏向管理者。

// AI 放大了钉钉原有的强管控基因
钉钉早期成功依赖“已读未读、强触达、组织确定性”。ONE 把这些能力 AI 化后，用户感受到的不是效率提升，而是更细颗粒度的监控。

// 用户真实反馈被误判
内测和大客户灰度阶段，管理层反馈很好，但基层员工强烈抵触。团队一度把这种抵触理解成“不适应新工具”，而不是产品方向本身有问题，这是文章认为最致命的误判。

// “信息收敛”变成了“信息轰炸”
ONE 原本想用卡片减少消息混乱，但各业务线都往 ONE 接入内容，最后变成“原有钉钉消息 + ONE 卡片”的双重轰炸。

// 发布会倒排工期摧毁了敏捷迭代
项目不是从 MVP、小范围验证、真实数据纠偏开始，而是围绕固定发布会日期高举高打。负面数据出现后，也没有真正修正顶层定位。

// 组织结构让问题被持续放大
产品、设计、研发、商业化、各业务线之间权责不清；业务线为了 AI 指标纷纷接入；一线用户反馈向上传递时被淡化，最终形成“上层看到正反馈，一线承受真实反噬”的断层。

// 它也并非毫无价值
文章承认 ONE 的方向踩中了办公 AI 的趋势：主动式信息流、AI 进入工作流、基于组织上下文做服务，这些都可能是未来方向。失败的是落地方式和用户立场。

无独有偶，硅谷也在反复证明同一个结论：当企业试图用数据和 AI 把人的工作过程彻底量化，最后得到的往往不是效率，而是反感、抵触和讽刺性的“表演生产力”。

Microsoft 的 Productivity Score 是一个很典型的例子。它原本想帮企业理解员工如何使用 Microsoft 365、Teams、Outlook 等工具，提高组织协作效率。但因为它能暴露个人层面的使用数据，很快被外界批评为“员工监控工具”。最后微软不得不调整产品，弱化个人级别可识别数据，改成更聚合的组织洞察。这个案例说明了一件事：员工并不是天然反对效率工具，但他们不接受被这样管理。

Amazon 内部统计 AI 使用量的做法也很讽刺。公司想推动员工使用 AI，于是开始看谁用得多、谁 token 消耗高、谁更积极拥抱新工具。但指标一旦被看见，就会被表演。有人开始为了显得自己更 AI-native 而刷 token、制造调用量。最后据报道，Amazon 撤掉了内部 AI 使用排行榜。企业以为自己在衡量生产力，实际衡量到的是员工表演生产力的能力。

Meta 的鼠标和键盘追踪则更直接。据报道，Meta 的内部项目会在员工工作电脑上记录鼠标移动、点击、键盘操作，甚至偶尔截图，用来训练 AI agent 理解真实办公行为。员工反应很强烈：内部帖子的高赞评论是“这让我非常不舒服，怎么退出？”；公告下最常见反应是愤怒表情。Meta 最初明确表示工作电脑无法 opt out，后来在压力下只开放了非常有限的临时暂停机制。它没有真正取消，但已经暴露出一个事实：当公司把员工的工作过程本身变成训练数据，员工首先感受到的不是效率，而是被采集、被观察、被占有。

这些案例和《置身钉内》里的钉钉 ONE 放在一起看，重点就很清楚了：这套逻辑正在失败。

钉钉 ONE 看见了未读、待办、遗漏和事项流转，却没看见员工需要缓冲和边界。

Microsoft 看见了工具使用行为，却撞上了隐私反弹。

Amazon 看见了 AI 使用量，却催生了刷指标。

Meta 想看见鼠标和键盘背后的工作过程，却激起了员工对数据采集的愤怒。

它们共同说明：工作不是越透明越高效，人也不是越可测量越好管理。一旦企业把“过程数据”误当成“真实贡献”，员工就会开始保护自己：要么关闭入口，要么消极抵抗，要么表演指标，要么彻底不信任系统。

所以，钉钉 ONE 的教训并不孤立。它只是更完整地呈现了一个趋势的终局：当 AI 办公产品从“帮人工作”滑向“帮组织看人”，失败几乎是迟早的事。

隐私不是效率的敌人，自由也不是管理的漏洞。很多时候，它们恰恰是一个工具能否被长期信任、长期使用的前提。

$SOXX rebalance

$SOXX will rebalance quarterly.

Its top 5 positions have capped weight at 8% and the remaining will be capped at 4%.

During the quarter, if some stocks run too hot, it could have a larger weight than the cap.

This will create excessive selling as companies like Marvell has heavy weight in $SOXX now but needs to be at 4%.

The rebalance will take effect on Jun 18, 2026.

Other companies like $MU will also need to be sold to reduce weight.

The unusual inflow

In recent 2 weeks, US stocks especially tech/QQQ went straight up, before the AVGO disappointing earnings.

There is an unusual capital inflow that accelerated.

It’s not just FOMO after fabulous run in April and May.

China’s recent crackdown also played an role – as Chinese investors investing overseas fear that they can’t add/buy for the next two years, they rushed to buy a lot!

And they are more likely to chase high beta stocks, leveraged ETFs.

That’s a problem in the short run.

And a correction should normalize this.

These investors can still sell according to rules; they can also buy/add when being abroad.

The April US inflation print

In April, CPI hit 3.8% annually and 0.6% MoM. Excluding food and energy, the core CPI increased 2.8% and 0.4%.

The 3.8% annual figure is the highest since May 2023.

However, as shelter is a big part of CPI, and Oct 2025 data is missing, there is a one-off bump here (0.6% MoM).

  • Normally, each housing unit in the CPI rent sample is surveyed once every six months. BLS converts the observed six-month rent change into a monthly rate.
  • Because October 2025 rent data were missing, the October index recorded zero rent change for that panel. When the same units were finally surveyed again in April 2026, BLS measured the accumulated rent change from April 2025 to April 2026 and converted it into a monthly estimate. BLS says the April estimate is therefore effectively based on the sixth root of a 12-month change, instead of the usual sixth root of a six-month change

Oil price is obviously rising a lot, in March/April/May.

but that has eased in Jun

Other interesting items like tomatoes.. which rose 15% MoM in April, is impacted by bad weather and US gov July 2025 17.09% antidumping duty on Mexico tomatoes.