周周聊投资

Solana’s event in Shanghai was abruptly called off before it started today ⚠️⚠️ Mainly because this event had gotten way too much attention. Maybe it was because several major people in charge had flown in specifically, including Toly, so the Asia-Pacific team wanted to make a good impression. But then...

Looks like CZ understands the local situation better after all haha #solana

Haha, I didn’t expect CZ @CZ to have seen Solana’s ad yesterday too. Talk about throwing shade at yourself: “It’s amazing that Solana can set up an office in Shanghai and advertise there! BNB Chain still can’t do that 😂😂”

#投资观察实录 The AI short-drama industry is working itself to death. TikTok CEO Shou Zi Chew recently said at an event that AI is dramatically reducing the cost of producing video ads.

In the past, if a brand wanted a promotional video for a product, it had to hire a production team, arrange a shoot, and handle post-production. Even the cheapest option would cost $100 to $1,000. Now, just hand product photos and a reference video to an AI tool like Seedance, and you can get the job done easily for just a few dollars. Costs have been slashed by several orders of magnitude.

So what are all those unemployed production teams doing now? Making AI short dramas... Forced to join another, even more cutthroat race.

Where have you brought me? Is this even still China?

Setting up an office is one thing, but Solana actually put ads on giant outdoor screens in Shanghai’s Lujiazui?? Who arranged this? And who approved it? 😡😡 This needs a thorough investigation!!!!#狗头

I finally got the whole story 🍉. Here’s roughly what happened:

A well-known livestreamer celebrated their birthday in Chengdu a couple of days ago and invited some major traders from the plaza. At the dinner, they put out four bottles of Romanée-Conti from the same vintage—all at once. At market prices, each bottle costs over 200,000 yuan, putting the total for all four close to a million yuan. And they’re incredibly hard to get.

The plan was to make a big splash, but as soon as the photos were posted, plenty of wine-savvy netizens questioned whether they were fake, pointing to details on the labels and bottles. Cue a wave of mockery. Oh well, real or fake～fake or real～ #罗曼尼康帝

Fortune China recently released its 2026 list of China’s 40 most influential business leaders under 40 👍👍 The top six are:
-Wang Xingxing, 36, founder of Unitree Robotics
-Yang Zhilin, 33, founder of Kimi by Moonshot AI
-Zhang Xue, 39, founder of Zhang Xue Motorcycles
-Yao Shunyu, 28, Tencent’s chief AI scientist
-Feng Ting, 38, president of REPT BATTERO
-Yan Junjie, 37, founder of MiniMax

What happened???? Why is everyone on the feed talking about Romanée-Conti? 😳😳😳

#字节股神 leto bao revealed why he hasn’t been talking much about stocks lately, and shared his AI trading workflow. He said trading is really a process of turning ideas into mechanical execution: he’s mainly responsible for the ideas, while AI handles the execution.

His main strategy now is yield enhancement, with a few smaller agents exploring other possibilities.

Lily Liu really knows her stuff—she actually set up a Solana office in Shanghai 👍👍 #solana

Is it true that tradexyz was acquired??? Will this affect the price of the hype coin? 🙉🙉🙉 #hype

People have been talking these past couple of days about Netflix turning FTX’s collapse into a TV series. A former FTX employee has spoken out: “Watching the trailer made me so sick I wanted to throw up” 🤮🤮

1. Even at its peak, FTX’s core team had fewer than 200 people. Many worked 16 to 18 hours a day and barely had time to sleep.

2. No matter the time of day or night, whenever employees walked into the office, Sam was always at his desk.

3. The real-life Caroline was nothing like the aggressive person portrayed in the trailer. She was quiet and gentle, and often lost in her own thoughts.

4. Most FTX employees didn’t know she and Sam were dating until they read about it after the company went bankrupt.

5. FTX’s collapse ruined many people’s lives. It was far more than just a love story for Netflix to turn into a show. #Netflix计划推出FTX题材剧集利他主义者

#黄仁勋女婿 Jensen Huang’s daughter got married, and the groom really hit the jackpot—reborn as the son-in-law of the CEO of a company worth $6 trillion 🤭🤭

The lucky guy is named Nico Caprez. In his younger days, he was a professional athlete on the Swiss national alpine skiing team. After retiring, he went into consulting at BCG. Later, while pursuing his MBA at London Business School, he met Jensen Huang’s daughter, Huang Minshan.

In early 2024, Nico joined NVIDIA. Apparently, Huang Minshan personally recruited him. His biggest achievement at NVIDIA has been driving the expansion of GPU-backed financing and partnerships with AI cloud service providers.

And Jensen Huang seems to think highly of his son-in-law, too. This January, Nico was promoted ahead of schedule to vice president of AI infrastructure growth, overseeing billions of dollars in partnerships between NVIDIA and a whole host of cloud service providers. $NVDAB

A few days ago, the top female influencer said on a livestream: “The era of nitpicking over words for memes is over.” With the market pulling back over the past few days, Chinese-language memes on BSC have also fallen quite a bit. Hope everyone still holding Niulai and “I’m F***ing Here” is okay 😭😭😭

Going forward, don’t just keep your eyes on posts from the top female influencer or CZ—that’s basically handing money straight to token-launching groups. Let’s keep our attention on quality projects instead. $牛来 $我踏马来了

#美联储9月会议纪要 The Fed’s September meeting minutes are here 👇👇

① The September 25-basis-point rate hike was approved unanimously, bringing the rate to 3.75%–4.00%
② Most officials believe another rate hike may be needed this year (though there’s no clear urgency to hike in October)
③ Inflation risks remain tilted to the upside; energy, tariffs, and the AI investment boom could all push prices higher
④ The labor market remains stable, with unemployment at around 4.1%. For now, there’s nothing forcing the Fed to pivot to easing
⑤ Bank reserves remain ample, but that doesn’t mean the Fed is turning the “money taps” back on

The short-term outlook is clear: high interest rates weigh on risk appetite, but the absence of a clear economic downturn also lowers the likelihood of a liquidity crisis.

What will really drive the market next is inflation data and the late-October policy meeting.

Rate hikes are weighing on gold prices, but in the context of the cycle it may not necessarily be a bad thing.
The trigger was the Federal Reserve’s 25-basis-point hike on September 16. The market now assigns roughly a 70% probability of another hike in October. The 10-year TIPS real yield rose to 2.79%, the highest in 18 years, and gold fell 24% from its January peak. The old rule of thumb is that for every 50-basis-point rise in real yields, gold tends to pull back by roughly 10%.
The real variable, however, is on the debt side. Someone has estimated the interest pressure for five countries through 2030: in the U.S., net interest in fiscal year 2030 could reach 34.8% of fiscal revenue; in Japan, debt is equivalent to 206% of GDP, and the average coupon on outstanding government bonds is only 0.98%. With each step of the rate-hike transmission, interest gets re-priced step by step. Even France’s finance minister has said that interest payments may exceed education and defense by 2027.
History offers a reference point: after 1945, the U.S. and the U.K. did not rely on austerity to repay debt; instead, they brought the debt ratio down from around 120% to below 50% through many years of negative real interest rates. Today, debt is even higher and the path is narrower. This round of hikes has not been able to disprove gold; instead, it is adding to the script of diluting debt via inflation.
Gold’s major swings have, time and again, appeared during phases where high debt coincides with real yields being suppressed. Rather than staring at the gold price every day, it’s better to watch for turning points in real yields. Take it slower—place your judgment on the side of the cycle.

Technology stocks are not a zero-sum game. Only by understanding this can you profit from the era

Many people treat tech stocks as a "you win, I lose" inventory game—a mindset that is fundamentally wrong from the start.

The essence of technology stocks is growth. It’s not a fixed-size pie that gets sliced and diced. Instead, the pie keeps getting bigger—and it even keeps baking entirely new pies, because every round of technological iteration creates demand that previously didn’t exist.

In the Stone Age, there were no means of transportation, so people naturally had no concept of "commuting." Without phones or groups, the very idea of "socializing" didn’t make sense either. From stones to iron, from the Industrial Revolution to automation, from the internet to AI—each leap in efficiency comes with the birth of a new set of demands. Demand isn’t something that exists naturally; technology is what makes it grow.

So what is the real zero-sum game?

In mature inventory markets, rivals carve up a cake that doesn’t grow—perhaps even shrinks. A classic example is baijiu (white liquor). Small investors in premium baijiu love telling a story: in the future, people will drink less but drink better, so they hope to take market share from lower-end segments. But when the low-end continues to shrink and weaker players get eliminated, premium brands end up fighting each other for market share—this is exactly the situation today.

Financial stocks are a different case. Licenses are limited and tied to financial security and social stability. With benchmark interest rates and LPR as guidance, it’s less likely to fall into a vicious zero-sum spiral.

Back to why we’re discussing this today.

Not long ago, Tencent fell to the 400 range, and then voices emerged saying, "Tencent will be eroded by ByteDance." How long has ByteDance existed? In the internet landscape, multiple players have coexisted for a long time—no one can crush the others. The AI track has only just started. Who can capture the largest share is still far from certain.

My personal judgment is that China’s AI market will ultimately be multi-player and parallel, not a single winner taking all. Based on the current layout, Alibaba has the highest probability of winning, followed by ByteDance and Tencent. The global picture is similar.

It’s important to emphasize that we’re nowhere near the stage of "value-for-money" competition. When the marginal advantages of model performance are no longer obvious, the industry will move into a mature phase where it competes on price and efficiency. Right now, it’s still a phase of rapid expansion—racing to grab territory and build fortifications.

AI ultimately needs to be deployed. There are roughly three deployment paths: consumer (To C) social AI, AI for the Internet of Things, and B2B (To B) cloud-service AI. Beyond the hard skills—large models, computing power, and chips—another key to winning is scenarios and key deployment points.

Tencent: the king of social platforms + mini-programs (connecting merchants, enterprises, and public services) + WeChat Pay + Moments + short videos. The ecosystem and barriers are unique—call it a moat if you like.

Alibaba: one of the world’s leading cloud service providers. It builds its own chip and computing-power centers, combined with a massive internet commercial operations platform and Alipay. The richness of its deployed scenarios is also among the best.

Xiaomi: a leading AIoT player across the entire person-car-home ecosystem. It’s hard to find a second one globally. Recently, vivo has connected to Xiaomi’s ecosystem, allowing the voice assistant to directly control Mi Home devices. As partners and devices keep joining, its IoT network becomes increasingly dense—traffic and profitability naturally follow.

So, when investing in tech stocks, please drop the zero-sum way of thinking. This is a high-investment, high-output game—and of course, a highly volatile one. If you understand the track and the scenario-based competitive barriers, the rest is simply to make friends with time.

U.S. Treasury yields hit a 18-year high—crisis or opportunity? What ordinary people must understand

Recently, the “U.S. Treasury” has been all over the news—an outstanding size of $40 trillion, yields climbing above 5.3%, and headlines full of talk about a “debt crisis” and “financial storm.” But many people finish reading even more confused: what exactly is a yield? Why has it risen to record highs, yet the market seems reluctant to buy? Why is gold rising and stocks falling? Today, I’ll walk you through the numbers clearly.

I. $40 trillion in U.S. Treasuries—who is lending money to the United States?

At its core, U.S. Treasuries are the U.S. government’s “IOUs” to the whole world. Recently, the total scale has surpassed $40 trillion. Pay attention to one detail: the borrowing pace is accelerating. The first $1 trillion took more than 200 years, but the additional $1 trillion from $30 trillion to $40 trillion took only 4.5 years.

Who are the lenders? $7.7 trillion is “the government owing itself” (retirement funds were moved early). Among foreign debt holders, Japan is the largest sovereign creditor (about $1 trillion). China ranks third—and has already halved its holdings from the 2013 peak of $1.3 trillion.

II. Interest rate vs. yield—what’s the difference?

This is the easiest part to get tangled up in. Every bond has a fixed coupon rate printed on its face, but the yield fluctuates—because Treasuries can be bought and sold like stocks.

Remember this counterintuitive rule: yield and bond price move in opposite directions.

Example: A $100 bond with a 5% coupon pays $5 in interest each year. If investors rush to buy it, the price rises to $105. The interest is still $5, but you paid $5 more in cost—so your actual yield drops to about 4.76%. Conversely, if the price falls to $95, the yield rises to about 5.26%.

So this time, the 30-year yield surged from 4.91% at the end of June to 5.3%, and the 10-year yield jumped to 4.75%. Both set highs since 2007. In plain language: the market is selling U.S. Treasuries and no longer wants to “hold the bag.”

III. Rollovers—how long can this game continue?

U.S. Treasuries have a particular feature: they pay interest but do not repay principal until maturity. When old bonds mature, new bonds are issued so the holders of the new bonds can take over—this is “rollover.” In theory, as long as someone is willing to buy, the game can keep going.

In U.S. history, there have been two “near-default” episodes. In 1933 during the Great Depression, Roosevelt refused to redeem with gold and instead repaid using devalued paper currency. In 1979, due to a computer malfunction, payment of $120 million was delayed; the market treated it as a technical default, yields spiked, and the government ended up paying tens of billions more in interest each year.

Look at how other countries fared and the picture is even more alarming: Argentina defaulted 9 times; Greece cut pensions meant for debt repayment by 40%, youth unemployment surged to 64.9%, and ATMs could only dispense €60 per day—finally dragging the entire eurozone down with it.

IV. Is the problem really that serious? Look at two indicators

You can’t only look at how fast the IOUs are piling up—you also need to examine earning capacity.

Debt-to-GDP ratio: the United States is above 120%, ranking 9th globally. Japan is highest at 204%, but 90% is domestic debt, so there’s no default risk.

Interest as a share of fiscal revenue: it jumped quickly from 17.2% last November to 23.7% this May, then eased slightly to 22.8% in July—meaning roughly for every $5 the government collects in tax revenue, $1 goes directly to pay interest.

Conclusion: the fundamentals haven’t deteriorated to an extreme level, but the signals are indeed not encouraging.

V. The real reason yields have surged

It’s not a supply issue—the core is “uncertainty.”

2020 was the key turning point. During the pandemic, 30-year yields once fell to a historic low of 0.7%. Then, the massive $2.2 trillion CARES Act and an explosive rise in deficits flooded the system and triggered a long-cycle reversal that lasted around 40 years. This year, the conflict between the U.S. and Iran in February pushed oil prices higher, renewed concerns about inflation, and kept yields elevated. Even if oil prices fall again, the central trend can’t return to the old level—because markets are no longer focused only on oil prices; they’re watching whether the Federal Reserve can control inflation.

VI. Impact on ordinary people’s wallets

U.S. Treasury yields aren’t just numbers on Wall Street—they filter directly into your bills:

30-year mortgages rise to around 6.7%, the highest level in nearly 20 years; auto loans and credit card interest rates move up across the board.
Gold surged to 4,600 within less than a month, up nearly 15%; Bitcoin broke above $80,000.
The Nasdaq fell 1.3%, and Philadelphia Semiconductor fell 5%—high interest rates suppress tech stock valuations.

VII. Which way will it go next? Watch two sides “tug of war”

The Federal Reserve: if it can’t curb inflation, it keeps high rates. Then banks compete for money with Treasuries, and Treasuries must raise yields to attract investors;

The U.S. Treasury Department: how much debt to issue and what maturities to issue directly determines the pressure on take-up.

The most conflicting part is this—while the Federal Reserve wants to raise rates to control inflation, the Treasury Department wants to cut rates to reduce financing costs. The two sides are pulling in opposite directions.

My view
In the short term, as long as there isn’t clear evidence that inflation is easing and the Treasury continues issuing large amounts of debt, U.S. Treasury yields will most likely remain in a high-range, volatile pattern. For ordinary people, rather than obsessing over whether a crisis will break out, it’s better to focus on asset allocation: inflation-hedging and sovereign-risk-hedging assets like gold and Bitcoin are worth paying attention to; tech stocks may face pressure in the short term, but the AI story still points to longer-term logic.

Every so often, a wave of "U.S. Treasuries are about to collapse" starts making the rounds. This time is no different: lots of bloggers are talking, and lots of people are believing it. But let’s calm down and think it through.

First, the conclusion: U.S. Treasuries are not going to collapse in the short term. Rising interest rates are real, but collapse is another matter.

What does collapse mean? For government bonds, it means the government can no longer pay the interest. Rising rates do not equal collapse; they just mean debt-servicing pressure is increasing. We are still far from the point of not being able to pay interest.

With U.S. Treasuries totaling $40 trillion, if there were truly a massive selloff, then what would people buy instead? A huge share of global capital is stored in bond form—if not U.S. Treasuries, then which country’s bonds? Behind them stands the backing of a nation’s overall strength, and for now there is no substitute. Capital may reduce exposure and demand higher yields, but it will not flee en masse.

What is more realistic is that no one wants U.S. Treasuries to collapse. A huge amount of global wealth exists in the form of U.S. Treasuries; if they really collapsed, everyone’s money would go to zero together, including ours. So U.S. yields can rise gradually, but a sudden collapse benefits no one.

From a cyclical perspective, the U.S. does indeed have many problems, but the decline of an empire is a long-cycle process. The Ming and Qing dynasties held on for many years; Japan has lived through more than two decades of stagnation. The U.S. has not actually weakened relative to other countries; rather, China’s rise has reduced its share of global GDP. Right now the U.S. is betting on AI precisely to widen the gap—if that bet pays off, Treasuries can still hold up.

How do we judge debt-servicing capacity? Look at net interest expense as a share of nominal GDP. Over the past two years, nominal GDP growth has been around 6%, while net debt service has been around 4%, not exceeding the overall growth rate, so the game can continue. In plain terms, this is about diluting debt through inflation, and the massive capital expenditures in AI are also supporting GDP.

But we cannot ignore the pressure that rising rates place on other assets. A sustained rise in the risk-free rate will suppress valuations of risky assets, and for non-U.S. countries it may trigger a dollar tide. As for how the U.S. will deal with high interest costs, it has many options: buy back long-term bonds and issue short-term bonds, engineer a recession to push rates down, QE—each comes with costs, and it depends on which one it is willing to bear. Most likely, it will delay action until the situation becomes urgent.

For investors, don’t be frightened by collapse narratives, but don’t ignore rising rates either. A slow rise in rates is relatively manageable; the key is to focus on your own holdings—corporate earnings quality and cash flow matter far more than worrying about whether Treasuries will collapse.

Who is the real king of cycles? Eight assets, eight businesses

Many people lump copper, gold, oil, coal, electrolytic aluminum, chemicals, phosphate ore, and shipping together as "cyclical assets." But the cycles behind them are not the same at all.

First, understand "profit elasticity": if one ton of copper costs 50,000 and sells for 80,000, the profit per ton is 30,000. If copper rises to 100,000, revenue rises only 25%; if costs rise to 53,000, profit per ton becomes 47,000 — profit rises by about 57%. When costs are relatively rigid, most of the price increase falls straight through to profit.

The four questions that really determine the "king of cycles": How much profit elasticity is there? How long does supply take to catch up? How long can high profits last? What determines the price?

1. Copper: the king of long cycles
Supply adjusts especially slowly. A large copper mine can take years from exploration to production. Demand picks up → supply fails to keep up → copper prices rise → profits increase → capital spending expands → years later new supply is released. Once a gap forms, it can last a very long time.

2. Gold: the king of monetary cycles
If gold rises from 2,000 to 3,000 and costs rise to 1,700, unit profit jumps from 500 to 1,300. But gold is first and foremost a financial asset and a non-credit currency. What matters are real interest rates, the U.S. dollar, central bank gold buying, ETF flows, and geopolitical risk. Both gold and Bitcoin are making money from "fiat currency credit dilution."

3. Oil: the king of macro cycles
For an oilfield with a cost of $50, if oil rises from $60 to $80, profit goes from $10 to $30 — up 33%, or triple the original profit. But oil is a commodity + strategic resource + political commodity. OPEC+, shale oil, the Middle East situation, sanctions, inventories — any variable can reverse expectations. The more complex the variables, the lower the predictability.

4. Coal: the king of cash flow
If coal costs 400 per ton and the coal price rises from 600 to 800, profit per ton doubles from 200 to 400. In recent years, safety and environmental approvals have constrained new supply. If supply is restricted and demand does not fall quickly, coal may shift from a strong cyclical trade to a high-cash-flow asset with cyclical characteristics. The key question is: under a neutral coal price, how much free cash flow can low-cost companies generate?

5. Electrolytic aluminum: the king of supply constraints
China’s electrolytic aluminum industry is constrained by capacity replacement, total output control, energy consumption, and environmental requirements. But supply constraints do not mean aluminum prices must rise — overseas new capacity, restarts, and imports can all change the balance. Constraints increase the probability of sustained high profits; they are not a guarantee of higher prices.

6. Chemicals: the king of capacity cycles
High profits → expansion → oversupply → profit decline → outdated capacity exits → a turnaround in the cycle. Mature chemical products can replicate capacity far faster than mines, so the most profitable period often sows the seeds of the next oversupply. "Chemicals" covers dozens of industries, so you must drill down to each sub-industry and examine supply and demand.

7. Phosphate ore: the king of resource barriers
What is truly scarce is high-grade, low-impurity, low-cost, easy-to-develop quality phosphate ore. The baseline demand comes from agriculture and phosphate fertilizers, which is resilient but does not mean prices will always rise. New energy brings incremental demand, but that alone does not make it a growth stock. The money here comes from rising scarcity of high-quality resources.

8. Shipping: the king of asymmetric payoff
If the daily cash cost is $20,000, a freight rate of $22,000 contributes $2,000 per day; if the freight rate rises to $40,000, it contributes $20,000 per day — the freight rate has not doubled, but cash contribution has increased tenfold. The core of shipping prices is effective capacity, not cargo demand. Port congestion, rerouting, and war can all cause capacity to collapse and freight rates to soar. High odds, explosive upside, but high volatility and poor sustainability.

9. So who is really the king of cycles?
There is no standard answer — the eight assets win on different dimensions. Copper makes money on time, gold on money and credit, oil on macro and geopolitics, coal on cash flow, aluminum on supply constraints, chemicals on capacity clearing, phosphate ore on resource barriers, and shipping on extreme payoff asymmetry.

On the surface they all look like cyclical assets, but behind them are eight very different businesses. The hardest thing in investing is not recognizing "this is a cyclical stock," but judging: in this round, what cycle is it really?

Valuation can’t be measured clearly with just one ruler.

What’s the difference between the “price-to-earnings yield” and “earnings-adjusted price-to-book value”? In essence, both methods do the same thing: pull profits into the valuation to correct for any distortion from looking only at a low P/E. A company that’s “cheap” merely because its earnings are low—while its profitability is collapsing—can look deceptively inexpensive. But the divergence begins right after profits are brought into the picture.

The price-to-earnings yield is a tool for horizontal comparisons. Take the P/E ratio and divide it by ROE: the smaller the result, the better the deal. For example, 20x PE with a 15% ROE yields 1.33; 15x PE with the same ROE yields 1.0. In a row, you can immediately see who’s expensive and who’s cheap. Pair this formula with Buffett’s “buy Coca-Cola at 40 cents on the dollar,” and it turns out retail investors can use it more conveniently than memorizing a pile of valuation models.

However, it has three boundary cases it doesn’t properly handle. First, Buffett’s “buy $1 worth of assets for 40 cents” refers to a 40% discount to net asset value. The “40-cent discount” implied in the price-to-earnings yield often isn’t a 40% discount to net assets—and it can even be a premium. If you copy it literally, you may copy it wrong. Second, cyclicals are the most dangerous: at the peak of the cycle, ROE is highest and the price-to-earnings yield lowest, making them look the cheapest—the very time you should sell. Third, without adjusting for the market’s “water level,” in bull markets everything appears artificially cheap and in bear markets everything appears artificially expensive. This method only tells you “who is cheaper than whom,” but can’t tell you “who is cheaper relative to its own past.”

Most worth praising is its dividend adjustment. “How much you earn” and “how much you distribute” are two different things. Some companies show high ROE on paper because they retain most profits and don’t distribute them. This kind of “earning” never really lands in shareholders’ pockets. The originator uses a 50% dividend payout ratio as a benchmark and introduces a coefficient N: companies that distribute more aren’t penalized, while those that distribute less are adjusted toward being “more expensive.” This dimension is precisely something the other method doesn’t explicitly address.

The earnings-adjusted price-to-book value takes a different direction: instead of comparing with others, it compares with the company’s own past. It adjusts long-term price-to-book using ROE. Any portion that can’t be explained by retained earnings is judged as high or low relative to its own historical levels. What it asks is “Is this company expensive relative to its own past?” not “Is this company cheaper than that one?” Stocks with declining ROE can still be read by it—as long as the main theme of “earn more at a higher price, earn less at a lower price” hasn’t broken.

After seeing a few bull and bear cycles, my takeaway is: the two rulers aren’t meant to replace each other. One handles the initial screening and the other handles confirmation—use them together for the most grounded approach. A bit slower, but it’s less likely to lead you into pitfalls.