涨460%的那天，五千只股票在流血

2026年8月19日，三件事同时发生。

第一件：宇树科技上市，人形机器人第一股，开盘1100元，涨629%，收盘845元，涨460.34%。市值3418亿，单日成交230亿。

第二件：A股崩了。沪指跌2.4%，深成指跌5.01%，创业板跌6.26%，科创50跌6.89%。超5000只个股下跌，MLCC、CPO、光刻机、存储芯片、先进封装，全线跌停。

第三件：美国联邦债务突破40万亿美元，十年翻倍，利息支出首次超过医保。现货黄金单日暴涨4.35%，冲破4500美元。中国国有大行的五年期大额存单，年化利率只剩1.6%，上线20分钟售罄。

大多数人的反应：宇树牛逼，股市惨，黄金疯，存款难抢。四件事，四个新闻。

我的反应：这是一件事。同一场资金迁徙，开了三个窗口。

钱正在离开普通人能买到的东西，涌向普通人买不起、买不到的东西。而你站在哪个窗口，决定你是乘客，还是行李。

一、宇树这场秀，谁在台上，谁在买票

先把数据摆平，再谈感情。

宇树的发行价150.8元，静态市盈率219倍。什么概念？通用设备制造业的平均市盈率，38.56倍。它贵了同行5.7倍。

贵这么多，图什么？图故事。人形机器人第一股、春晚跳舞、杭州六小龙、世界模型、具身智能——每一个词都是故事，每一个故事都在涨价。

故事确实卖爆了。网上申购中签率0.018%，万分之二。创了科创板历史新高。

然后开盘。1100元，涨629%。收盘845元，涨460%。一天的走势从最高点回落了23%。成交230亿——相当于全市场成交额的0.9%被这一只股票吸走。

注意这个数字。230亿，是钱，不是情怀。谁在卖？开盘就卖的人。谁在买？冲进去的人。

现在看基本面。2025年，宇树营收16.99亿，人形机器人出货5500台，全球第一，毛利率60%。这公司不差，是有真东西的。

但2026年一季度，营收增速从332.64%骤降到68.49%，扣非净利润同比腰斩52.55%。基本面在降温，股价在沸腾。

市值3418亿，营收16.99亿。市销率200倍。意思是，这家公司一年卖出去的所有东西，你要给它200年的销售额，才配得上今天的市值。

我不说泡沫，泡沫是情绪词。我说账：这个价，不是为今天的宇树付的，是为一个遥远的梦付的。梦成不成的风险，谁扛？最后接棒的人扛。

再问一句：涨的460%去了谁的口袋？

美团系持股9.65%，最大外部股东，账面回报40亿以上。红杉中国持股7.11%，2019年投了1500万，当时估值1.5亿，现在账面回报25亿以上。创始人王兴兴持股34.8%，投票权68.8%，身价直接千亿级。社保基金、中石油、南方电网、中国电信，战略配售，全在发行价上车。

他们是演员。你，是买票的观众。

二、同一天，五千只股票在流血

宇树在台上领奖的时候，台下五千个股票在挨打。

沪指-2.4%，深成指-5.01%，创业板-6.26%，科创50-6.89%。成交2.51万亿，放量1100亿——跌着放量，是钱在跑，不是钱在进。

跌得最狠的，恰恰是昨天还在讲故事的板块：MLCC跌停、CPO二十只跌停、光刻机跌、存储芯片跌、玻璃基板跌、先进封装跌。人形机器人板块自己都跌——上纬新材一天跌19%。

逆势涨的是什么？银行。煤炭。防御。

注意这个结构：散户最爱的科技题材血流成河，避险资金抱团银行煤炭。这不是风格轮动，这是同一批钱在决定去留。

钱没有离开市场。钱在挑地方。

三、钱在往哪跑

第三个窗口，拉开看。

美国联邦债务40万亿美元，十年翻倍。利息支出第一次超过医保预算——美国政府借新债还旧债的利息，已经贵过给全美老人看病的钱。这不是新闻，这是火车票：全球最大的债务人，用未来的钱，买今天的平安。

黄金单日+4.35%，收4523美元。上一次这么涨，是市场在害怕。这一次怕什么？怕纸。怕所有印出来的、发出去的、承诺兑付的东西，最后都打折扣。

再看中国。五年期大额存单，7月回归，年化1.6%。1.6%啊各位，六年前这批钱存进去的时候是3%以上，收益直接腰斩。就这，上线20分钟售罄，国有大行6天卖光。

为什么抢？因为2026年全年，一年期以上定期存款有50万亿到期，其中五年期5到6万亿。这批钱到期了，拿着3%的旧存单去续，只给1.6%。储户不想接受，又不敢去股市——4月5月，居民存款合计减少2.05万亿，近十年少见的连续回落。钱去哪了？理财、基金、保险，增了4.65万亿。

看懂了吗？普通人的钱在搬家：从存款搬到理财，从理财搬到黄金，从黄金搬到任何一个能锁住利率的地方。每一次搬家，都接受更低的收益。因为他们最怕的已经不是赚不到，是亏掉。

四、三个窗口，同一个逻辑

把三件事叠在一起看。

宇树窗口：钱在买故事，买未来的梦，219倍市盈率不嫌贵。但这个故事的门票，万分之二的中签率，二级市场开盘价845元——普通人在用最高价买最不确定的东西。

美债黄金窗口：钱在买确定性，买"不会消失"的东西。全球最聪明的钱，在用真金白银投票：不要故事，要安全。

大额存单窗口：普通人拼手速抢1.6%——他们在用最保守的方式，买最确定的收益。哪怕收益腰斩，只要确定。

一个市场在给不确定性定价，两个市场在给确定性定价。这不是矛盾，这是同一场游戏的上下半场。

越往金字塔尖，越买确定性。越往下，越被卖出不确定性。

一级市场的钱，1500万变25亿，持有十年，确定性拉满。二级市场的钱，845元接棒，持有十分钟，不确定性拉满。谁在中间过手？每一层都在把不确定性卖给下一层，顺便赚走差价。

宇树的460%，不是给散户的礼物，是给一级市场的颁奖礼。散户不是演员，是观众。但观众买了票。

五、follow the money

看不清的时候，看钱的流向。

这场迁徙里，收钱的人很清晰：美团，红杉，社保，战略配售机构，王兴兴。他们拿到的，是十年布局的兑付。

付钱的人也很清晰：230亿成交额里追高的散户，5000只下跌股里割肉的持仓者，3%利息转1.6%的储户。

钱从不撒谎。故事会换，热点会换，情绪会换。但利益流向永远不变：从信息少的人，流向信息多的人。从等不及的人，流向等得起的人。从买故事的人，流向卖故事的人。

收尾：先认清你的位置

我不劝你离开股市，也不劝你追宇树。我只劝你做一件事：下注之前，先想清楚自己在食物链的哪一环。

好公司和好股票是两回事。宇树可能是好公司，但219倍市盈率不是给你的优惠，是给你的考题。你买的从来不是公司，是情绪，以及排在情绪后面的那个人的获利。

给自己三个检查点：

这钱是谁赚的？如果答案是"一级市场、战略配售、创始人"——那你赚的是谁的钱？

我赚的是谁的钱？如果答不上来，你大概率就是那个被赚的。

如果我不买，我会错过什么？如果答不上来，那你不买，什么都不会错过。

涨460%的那天，有人拿到了十年一遇的奖赏，有人拿到了十年一遇的教训。区别不在眼光，在位置。

而位置，在你下注之前，就已经定好了。

On August 19, 2026, three things happened on the same day.

Unitree Robotics, "China's first humanoid stock," opened at 1,100 yuan, up 629%, and closed at 845 yuan, up 460.34%. Market cap: 341.8 billion yuan. Single-day turnover: 23 billion yuan.

The A-share market collapsed. Shanghai Composite -2.4%, Shenzhen Component -5.01%, ChiNext -6.26%, STAR 50 -6.89%. Over 5,000 stocks fell. MLCC, CPO, lithography, memory chips, advanced packaging — entire sectors hit limit-down.

And: U.S. federal debt crossed $40 trillion, doubling in a decade, with interest payments now exceeding Medicare spending. Spot gold jumped 4.35% in a day, breaking $4,500. In China, five-year certificates of deposit paying just 1.6% annualized sold out 20 minutes after launch.

Most people saw four separate news items. I saw one event with three windows: capital is leaving things ordinary people can buy, and rushing into things ordinary people can't afford or can't get. Which window you stand at decides whether you're the passenger or the luggage.

Part One: The Unitree Show — Who's on Stage, Who's Buying Tickets

The IPO priced at 150.8 yuan per share: a static P/E of 219 times, versus an industry average of 38.56 times. Five-point-seven times more expensive than its peers, paid entirely for a story — humanoid robots, the Spring Festival Gala, embodied intelligence, the world model. Every keyword is a story, and every story inflates the price.

The story sold. The retail subscription hit rate was 0.018% — two in ten thousand — a STAR Market record.

Then it opened at 1,100, up 629%, and faded to close at 845, up 460%. A 23% round trip from the highs in a single session. Twenty-three billion yuan of turnover — nearly one percent of the entire market's volume, sucked into one ticker.

Who sold? People who got in at the open. Who bought? People who rushed in.

Now the fundamentals. Unitree's 2025 revenue was 1.7 billion yuan; 5,500 humanoid units shipped, first globally; 60% gross margin. This is a real company with real products.

But in Q1 2026, revenue growth collapsed from 332.64% to 68.49%, and non-GAAP net profit was halved. The fundamentals are cooling while the stock is boiling.

Market cap of 341.8 billion on 1.7 billion of revenue: a price-to-sales ratio of about 200x. You're not paying for today's Unitree. You're paying for a distant dream, and the risk of that dream failing is carried by whoever takes the last handoff.

Where did the 460% actually go?

Meituan holds 9.65% — the largest external shareholder, book return over 4 billion yuan. Sequoia China holds 7.11%: 15 million yuan invested in 2019 at a 150-million valuation, now worth over 2.5 billion. Founder Wang Xingxing holds 34.8% with 68.8% voting rights — a billionaire on paper. The social security fund, CNPC, China Southern Power Grid, China Telecom — all in at the offering price via strategic placement.

They are the performers. You are the audience — but you paid for the ticket.

Part Two: The Same Day, Five Thousand Stocks Bleeding

While Unitree collected its prize on stage, five thousand stocks were taking a beating below.

The worst-hit sectors were exactly the ones that were telling stories yesterday: MLCC limit-down, twenty CPO stocks limit-down, lithography, memory, glass substrates, advanced packaging. Even the humanoid robotics sector fell — one name dropped 19% in a day.

What went up? Banks. Coal. Defensive names.

Read the structure: retail-favorite tech themes bleeding, risk-averse money huddling in banks and coal. This isn't style rotation. This is the same pool of money deciding where to stay. The money didn't leave the market. It's choosing its spots.

Part Three: Where the Money Is Going

The third window, zoomed out.

U.S. federal debt at $40 trillion, doubled in a decade, interest costs now exceeding Medicare. The world's largest debtor is borrowing new money to pay old interest, buying today's peace with tomorrow's bills.

Gold +4.35% in one day, closing above $4,500. The last time it moved like this, the market was afraid — afraid of paper. Afraid that everything printed, issued, and promised will eventually be discounted.

China: five-year CDs returned in July at 1.6%. Six years ago, that same money earned over 3%. Half the yield, and it still sells out in 20 minutes.

Why the scramble? Because in 2026, roughly 50 trillion yuan of term deposits mature — 5-6 trillion of them five-year paper. Rolling over at half the old rate is a pay cut savers don't want but can't avoid. In April and May, household deposits fell by 2.05 trillion yuan — a rare two-month decline. The money moved into wealth management, funds, and insurance: +4.65 trillion.

Ordinary money is migrating — from deposits to wealth products, from wealth products to gold, from gold to anything that locks in a yield. Every move accepts a lower return, because the thing they fear most is no longer earning less. It's losing what they have.

Part Four: Three Windows, One Logic

Stack the three events.

Unitree window: money buying stories at 219x earnings, with a 2-in-10,000 allocation rate and an 845-yuan opening price. Ordinary people paying the highest price for the most uncertain thing.

Treasury and gold window: money buying certainty — "things that won't disappear." The smartest money voting with real metal: no stories, just safety.

CD window: ordinary people sprinting for 1.6%. Buying the most conservative certainty available, even at half the old rate.

One market is pricing uncertainty; two markets are pricing certainty. Not a contradiction — the same game, two halves.

The higher up the pyramid, the more certainty they buy. The lower down, the more uncertainty gets sold to you.

Primary-market money: 15 million into 2.5 billion, held a decade, certainty maxed. Secondary-market money: 845-yuan handoff, held ten minutes, uncertainty maxed. Every layer in between sells uncertainty to the next layer down and skims the spread.

Unitree's 460% isn't a gift to retail. It's an awards ceremony for the primary market. Retail isn't a performer — but they bought the ticket.

Part Five: Follow the Money

When you can't see clearly, watch the money.

The recipients are clear: Meituan, Sequoia, the social security fund, strategic placement, the founder. They're cashing in a decade of positioning.

The payers are clear: retail chasing 23 billion of turnover, holders cutting losses across five thousand falling stocks, savers rolling 3% deposits into 1.6%.

Money never lies. Stories change, hot topics change, sentiment changes. But the flow of interest never changes: from the poorly informed to the well-informed. From the impatient to the patient. From those who buy stories to those who sell them.

Close: Know Your Position First

I'm not telling you to leave the market, and I'm not telling you to chase Unitree. I'm telling you one thing: before you bet, know which link of the food chain you're in.

A good company and a good stock are different things. Unitree may be a good company — but a 219x P/E isn't a discount, it's an exam. You're never buying the company. You're buying sentiment — and the profit of whoever sits behind that sentiment.

Ask yourself three questions:

Whose money is this? If the answer is "primary market, strategic placement, founder" — then whose money would you be taking?

Whose money would I be giving? If you can't answer, you're probably the one giving.

What do I lose if I don't buy? If you can't answer, then nothing. And nothing lost is a position too.

On the day something rose 460%, some people got a decade-defining reward, and some people got a decade-defining lesson. The difference wasn't eyesight. It was position.

And position — that was set before you placed your bet.

Contents