币圈半边天 摩根智能杭州古墩新时代店

The simplest and most likely explanation: they just can't afford to buy water. Nowadays, it's all about QR code payments; if your phone's dead, you forgot it, just lost your wallet, or you're out and about facing some hiccups, it’s totally possible to be "broke and parched". A key feature of a scam is that it escalates - it starts small and then aims bigger. But when you offered a way out ("do you need anything else"), she clearly just wanted a bottle of water, took it, thanked you, and walked off; no upsell at all. This isn't the behavior of a scammer; it's a straightforward transaction that wraps up once the demand is met.

The second scenario, which is more aligned with your line of work: street test-style short videos. "Will strangers help me buy a bottle of water?" This kind of good deed test, human experiment is a well-trodden topic on Douyin and Kuaishou. A young, attractive girl paired with a not-so-old mother is a high-engagement character combo. If this is the case, you might already be in someone else's shot; they’re after the views, not your two bucks.

As for scams like staged accidents, pickpocketing as a distraction, or recruiting people into cults/pyramids - these all require follow-up actions: either to engage, divert attention, or steer the conversation elsewhere. None of that happened here.

Exaggerated or Unsupported Claims

The key point: "80,000 to 150,000 shares" and "a valuation of $15 million"—these figures aren't even mentioned in the original report (The New York Times). The Times actually named another person—Trevor Hise, a launch engineer who interned at SpaceX in 2011, accumulated over 100,000 shares over 12 years, valued at approximately $13.5 million based on $135 per share, and is now 37 years old, semi-retired. Launch engineer Trevor Hise stayed on after his 2011 internship, and over the years, he accumulated over 100,000 shares, which at the issuance price of $135, is worth at least $13.5 million, and he's 37.

The numbers you quoted, "100,000 shares, $13.5 million," are almost a direct copy of Hise's real figures, then incorrectly attributed to Guo Can. In other words, the real holdings of a 37-year-old veteran employee have been grafted onto a 27-year-old engineer who gained notoriety due to a contrast in appearance. This is a classic case of "if there's a picture and hype, let's fabricate a set of wealth numbers."

The other figures are also just "estimates" with no source:

Annual salary of $130,000: reverse-engineered from job postings, not her actual salary.

Holding 80,000 to 150,000 shares: pure speculation. The likelihood of a 27-year-old engineer, who might have been with the company for only three to five years, holding shares at the level of veteran Hise (who has been there for 12 years) is extremely low. The general consensus in the SpaceX employee community is that having over 100,000 shares is rare and usually reserved for those who "joined early, were promoted multiple times, and never sold."

"Holding the power to terminate launches": This is basically a misinterpretation. Reading the GO/NO-GO command during a live stream ≠ having personal authority to stop a launch. The authority to terminate a launch generally lies with the flight director, range safety officer, and automatic flight termination system, not on the shoulders of a console engineer.

SpaceX's wealth-building logic isn't about "generously handing out money," but rather low cash salaries + allowing employees to use their salaries (ESPP deductions) to buy into the company's future potential—essentially shifting R&D risk onto employees. A Georgetown University accounting associate professor pointed out after reviewing the prospectus that many frontline employees' stocks aren't "given away for free"; they are purchased at a discount through deductions from their paychecks. Winning means becoming a millionaire, while losing (like in 2008 when the company faced three launch failures and nearly went bankrupt) means losing everything. There's also a former employee whose shareholdings accounted for 93% of their family's investable net assets—this isn't called financial freedom; it's called going all in without a margin call.

The real deal: the pressure from demographics is a hard constraint, not some conspiracy theory. The 2019 report from the Chinese Academy of Social Sciences, "China's Pension Actuarial Report", calculated that the cumulative balance of the urban employee basic pension insurance could run out around 2035, with the dependency ratio worsening from nearly two contributors per retiree in 2019 to about one per retiree by 2050. This is the root of the "pension bomb" narrative, which is directionally correct. The adjustment of pensions is also narrowing year by year — a 3.8% increase in 2023, dropping to 3% in 2024, further down to 2% in 2025, and likely capping at 1.5% in 2026. So the judgment that "the actual purchasing power of this money will shrink" holds water.

There are two layers of exaggeration.

First, the claim that "almost 100% of those under 45 will receive nothing" — that conclusion doesn’t hold up. The 2035 figure refers to the "cumulative balance running out", but that doesn’t equate to no payouts. China's pension system operates on a pay-as-you-go basis, meaning the money young people pay now isn’t for their future selves, but directly supports current retirees. The balance on the books is just a buffer, not the sole source of payouts. By the end of June 2025, the cumulative balance of the three social security funds is at 9.83 trillion yuan, with current income exceeding expenditure, plus there are further safety nets like national social security strategic reserves, state asset transfers, and fiscal backing. Wiping out the pensions of hundreds of millions is a matter for the regime, and it won't happen.

Moreover, the 2035 figure is already outdated. The new version of the "Pension Actuarial Report 2025-2060", led by Zheng Bingwen, recalibrates with reforms like delayed retirement, central adjustments, and nationwide coordination, pushing the years of income not covering expenditures from 2028 to 2036, and the fund depletion year from 2035 to 2044, delaying it by eight to nine years. So the real risk isn’t "not receiving funds", but rather being substituted for "receiving late, receiving less, and discounted by today’s purchasing power".

Second, while you say "those over 55 are catching the last train", the direction is correct but the reasoning could be harsher: the closer one gets to retirement, the higher the certainty, because all government tools (delayed retirement, increasing minimum contribution years, lowering growth rates) are aimed at diluting benefits for the young; the younger you are, the more you get hit by "dynamic parameters".

Next, you completely miss a piece in this framework — social security ≠ pensions. For those outside the system or in flexible employment, what’s often truly valuable is health insurance, not pensions. Health insurance can be realized immediately, especially for major illnesses.

The facts are mostly on point, but there are two areas where it subtly inflates the narrative:

First, "it must rely on one person's decision-making → that person is the center" — this observation holds water, as Bitcoin in 2010 was indeed heavily centered around Satoshi Nakamoto (who held the alert key, dominated the reference client, and had overwhelming influence). It's a legitimate issue worth discussing. However, it gets packaged as "the vulnerability exposed the paradox, so he chose to vanish," which shifts from fact to speculation.

Second, and most crucially: the timeline doesn't match up. The vulnerability was in August, yet Satoshi remained active for another half year afterward, continuing to develop, send emails, and hand over work to Gavin Andresen from late 2010 to April 2011. He didn't "fix the vulnerability, realize he was the center, and then disappear" — he was gradually fading out, and the public clues (the handover, that line "I'm going to do other things," and the anxiety over WikiLeaks referencing Bitcoin) point more towards real-world risk aversion rather than a "deliberate design to decentralize through self-erasure.

In other words: the vulnerability is real, the rollback is real, the centralization tension is real; but saying "this is why he vanished" is a narrative crafted by later generations — Satoshi never claimed that. And even if he did leave, Bitcoin didn't achieve "complete decentralization" — the alert key existed until 2016, and centralized vectors like core developers and mining pools are still around. This logic of "erasing oneself = system becomes pure" is too clean, too neat to reflect reality.

As soon as the college entrance exam wraps up, Apple stores are packed, and that's a fact. There's a specific term for it called "post-exam economy," a routine scene in the digital market every year—on the afternoon to evening of exam day, parents rush in with their kids for the first wave of purchases. They’ll come back again between result announcements and the start of school. This has nothing to do with patriotism; it’s purely a release of a suppressed consumer group that's been waiting for three years.

The assembled part: connecting the dots like "the Huawei store next door is deserted" and then slapping on a label of "students not loving their country" is the real intent behind this narrative—it seeks emotion and division, not facts.

Hitting back with data is the most direct approach. In Q1 2026, the smartphone market in mainland China saw Huawei leading with 13.9 million units, 20% market share, while Apple followed with 13.1 million units, 19%. This means, if we’re talking about how well "domestic products sell," Huawei is actually ahead of Apple in the national market. Saying "the Huawei store is deserted" doesn’t represent the market and can’t be considered a voting reflection of a nation's purchasing power. The logic of this narrative contradicts itself.

So why does it feel so different on the ground? The real reasons are structural and have nothing to do with stance:

First, the density of flagship stores. Apple has only about 50 official stores throughout mainland China, all located in core business districts of first-tier cities, making it a "destination retail" naturally easy to gather crowds and create big scenes; Huawei, on the other hand, has thousands of stores spread out, diluting foot traffic. Comparing the crowd at an Apple flagship store to the quiet next door at a Huawei shop is like a horse race with mismatched competitors.

Second, the timing of educational discounts. Apple has a well-established educational discount system specifically aimed at incoming college students, combined with national subsidies, allowing students to save up to 2,000. This mechanism funnels freshly graduated students directly into Apple stores all around the same time, creating that cluster effect.

Third, the ecosystem bundle extends the stay. Parents often head in looking for the "Apple four-piece set"—phone, tablet, computer, and headphones all at once, with a single order often exceeding 20,000. The purchasing, comparing, and activating processes take a while, so the store naturally appears full; buying just a phone would be much quicker.

Looks pretty "professional", but the core logic is totally reversed—it's completely misrepresenting how the Three Gorges flood control works.

① "1.8 billion tons of sediment / average 90 million tons per year" → The numbers are roughly correct, but they deliberately left out the most crucial comparison.

What should really be compared isn't "how much has been silted", but rather "how much has silted compared to the initial design predictions". Back when it was being argued, the design was based on about 500 million tons of sediment entering the reservoir per year and an annual sedimentation of 330-350 million tons. However, the Yangtze River Commission report shows that the actual average annual sedimentation over ten years was only about 144 million tons, which is just 40% of the predicted value. By 2014, the annual sedimentation had dropped further to 45 million tons, which is just over 10% of the original forecast.

The reasons are quite clear: due to the upstream reservoir clusters trapping sediment, soil conservation, and less rainfall upstream, the average annual sediment entering the Three Gorges has dropped from the designed 500 million tons to about 200 million tons. So that "1.8 billion tons" scary cumulative figure is actually a good news, far below design expectations, which is beneficial for extending the reservoir's lifespan, but the headline makers twisted it around. The claim of "possible scrapping" is just clickbait—what's happening at Sanmenxia is what you call siltation, the Three Gorges is a completely different story.

② "It took special flushing twice to remove over 3 million tons, which is a huge gap" → This misrepresents the sediment flushing mechanism.

The main force behind sediment flushing isn't just "those two special openings". The Three Gorges uses a "store clean, release muddy" method: during the flood season, about 30% of the sediment is naturally flushed out with the high flow velocity water, making the overall sediment flushing rate about 24%. This means that each year, about a quarter of the sediment that enters the reservoir is regularly discharged, so that 3 million tons is just a supplemental amount. Comparing a one-time supplement to a total accumulated amount over 20 years is inherently a meaningless comparison.

③ "During the flood season, it should store water for flood control, but instead it releases water, making the flood control ineffective" → This statement is completely upside down.

This is the biggest error in the entire comment. The Three Gorges flood control relies precisely on emptying the water beforehand to create storage capacity "vacant to wait for" the flood:

Before the flood season, the water level must drop to the 145-meter flood limit to free up 22.15 billion cubic meters of flood control storage; after the flood, it is then raised to 175 meters for power generation and to supplement water during dry periods.

During non-flood periods, maintaining a high water level for water supply and power generation is essential, but before the flood season, it must be lowered to the flood control limit to clear storage capacity for the potential incoming floods.

When the flood hits, that empty storage capacity is used to intercept and reduce the peak of the flood, and after the flood recedes, the water level is gradually lowered back to 145 meters, freeing up capacity again for the next flood.

The sudden order from the U.S. Department of Commerce is due to another company claiming to have successfully breached the security of the Mythos model.

This doesn’t mean the government directly delisted the model; rather, the directive was so broad that Anthropic felt they had no choice but to disable access to these two models for all users. They also made it clear that other weaker models, including the latest Claude Opus 4.8, are not affected. AWS only revoked access to Fable 5 and Mythos 5 to comply with export control directives, at Anthropic's request; all other models, including Opus 4.8, remain unaffected and can continue to be used normally.

So, a few corrections to the narrative:

"Enterprise-level revenue drops to zero in the short term"—this is an exaggeration. Only the two top-tier models are affected; Opus 4.8 and the rest of the product line are running as usual, and the vast majority of Claude's traffic on AWS Bedrock remains unaffected. The claim of zero revenue doesn't hold up on the facts.

"Killing cutting-edge commercial models within 3 days"—the speed is indeed fast, but qualitatively, Anthropic believes this is a misunderstanding and is working to restore access as soon as possible. This is a temporary interruption, not a permanent death sentence; it’s not at the point of "killing" yet.

Another crucial fact for market judgment: Anthropic is a private company (often labeled with the code ANTH.PVT), and they recently submitted a confidential application for an IPO, with a valuation of around $965 billion in their latest funding round. Therefore, retail investors cannot trade "Company A stocks" directly; all market impacts can only be transmitted indirectly through proxies like AMZN, GOOGL, and NVDA.

Regarding the judgment of "必跌 on Monday"
On the supporting side: Historically, AI regulation/export control news can indeed crash the market. For instance, after the H20 ban was announced, NVIDIA dropped 4.5% in a single day, while Broadcom and AMD fell by 2.8% and 2.2%, respectively. The uncertainty premium from policy shifts is real and can be punished by the market.

But the counterforce is significant too: this is a single point, specific model event, not a demand contraction at the chip level; Opus is still selling, and NVIDIA's orders, as you've stated, are still in high demand. There hasn’t been any real cut in computing power procurement; the impacted parties are private companies, and the direct financial impact on AMZN/GOOGL is marginal (both companies have small investments in Anthropic and their cloud revenue share is low); plus, the company has framed it as a "misunderstanding, striving for recovery." All these factors could allow the impact to be quickly digested, even potentially leading to a rebound after a lower open.

Not even pretending anymore, it's a clever self-deprecating move + a way to ride the hype, the dev clearly knows how to spread the word.

This chart is almost certainly AI-generated/synthesized, not a real photo. You can tell from a few details: the composite feel at the edges of the figures and lighting, Trump's face has a "plastic feel", and the overall composition is just too perfect.

In the image, there are three people:

The one in the middle wearing a suit and a red tie is Donald Trump (synthesized into the picture).

The two flanking him—a silver-haired male and a silver-haired female, dressed in gold-trimmed embroidered bright red military ceremonial outfits—are not real people but rendered in a fantasy character style. That pale skin tone + long silver hair clearly mimics the aesthetic of the Targaryen family from "Game of Thrones"/"House of the Dragon" (the Targaryens are known for their platinum blonde hair). The red and gold military attire looks like it’s draped over this "silver-haired noble" image, resembling royal guard/general's clothing.

Berkshire, with Buffett at the helm, is really holding a record amount of cash. As of Q1 2026, Berkshire has around $397 billion in cash, cash equivalents, and short-term U.S. Treasury bonds, which is about $24 billion more than the $373 billion at the end of 2025. Plus, the company has been a net seller of stocks for 14 consecutive quarters. Folks like Jim Rogers and Gao Zhikai have openly shouted that '2026 will see an epic crisis.' This isn’t just made up.

But here's the crucial leap in logic:

'Big players cashing out = impending crisis' is a weak inference. Most tech executives are selling off to pay taxes, diversify holdings, or are following pre-set automatic selling plans (10b5-1), which doesn’t necessarily mean they’re bearish on the market. Michael Burry shorting Nvidia is something we see almost every year, and he's missed many opportunities as well. Buffett hoarding cash has a more straightforward reason: U.S. stock valuations are near historic highs, and short-term Treasuries offer about 3.6% risk-free returns. He feels there's no better deal in the stock market right now, especially after he officially handed over the CEO role to Greg Abel earlier this year. This is 'cautious,' not 'he knows exactly when it’s going to crash.' Buffett repeatedly emphasizes that he never predicts market timing.

As for the online chatter about 'June 2026 is a must-crash, with an 80% chance of a regional crisis and a 15% chance of a 1929-level Great Depression'—this kind of pseudo-precision is actually the biggest red flag. Real experts won’t give you single-digit probabilities. Rogers and other 'permanent bears' have called for a crash countless times over the past decade; even a stopped clock is right twice a day.

So what should regular folks do? The good news is: solid preparation holds true whether a crisis comes or not, and you don't need to gamble on direction—

Keep enough cash buffer: Have 3–6 months’ worth of living expenses on hand (including mortgage/rent/kids' expenses). This is the foundation for risk resistance, more important than any timing strategies.

Avoid leverage and high-interest debt: When a crisis actually hits, leverage is the first to blow up. First, deal with high-interest debts like credit card installments and consumer loans.

Don’t chase hot bubble assets at their peak: Whether it's an AI concept or some soaring coin, when 'everyone is buying and the story’s the best,' that’s usually when the risks are highest.

1. The timeline just doesn't add up. Chili peppers are a crop from the Americas, and they didn't make it to China until the late Ming Dynasty (the earliest record is from 1591 in Gao Lian's "Zunsheng Bajiang"). Initially, they were ornamental plants, later adopted in traditional medicine, and it wasn't until the Kangxi period in the Qing Dynasty that they truly entered the Chinese diet. In other words, the claim that people in Yunnan, Guizhou, and Sichuan "couldn't afford salt for hundreds of years and had to rely on spicy food"—that hundred-year gap had no chili peppers in China at all. This causal relationship doesn't hold up physically.

2. Chili cannot replace salt (a hard fact). Salt is sodium chloride, and the body relies on it to replenish sodium, which is an essential electrolyte for life; capsaicin merely stimulates pain receptors and has almost zero sodium content. No matter how much chili you consume, it won't provide even a grain of salt's sodium. The notion of "being so poor you can't afford salt and using chili as a substitute" is physiologically a fallacy—people lacking salt will only feel worse after eating spicy food.

3. Sichuan is actually one of China's largest salt-producing regions. Zigong is known as the "Salt Capital of a Thousand Years," with a history of salt extraction dating back to the Eastern Han Dynasty, flourishing during the Ming and Qing Dynasties. By the Xianfeng period of the Qing, annual salt production exceeded 120 million jin, supplying salt for one-tenth of the nation's population. By the late Qing, it was referred to as the "richest and finest place in Sichuan." Blaming Sichuan for "being so poor they couldn’t afford salt for hundreds of years" is completely off base.

4. The real case of "using chili as a substitute for salt" is actually in Guizhou, which the post didn't mention at all. Academic research (Cao Yu's "History of Spicy Food in China") indicates that chili peppers entered the diet as a substitute for salt among the impoverished in Guizhou. However, this was a specific case from a particular period in the Qing, in a specific region due to difficulties in salt transportation, not a universal rule that "all poor places use chili as salt." The post overlooks the only relevant example and incorrectly applies it to Sichuan, a major salt-producing province.

5. The idea that "spicy can mask food spoilage" is just a fantasy. A more plausible explanation is climate: provinces that favor spicy food tend to be humid and hot inland areas (there's a folk belief about "dampness removal"), along with the antibacterial hypothesis of spices. The northeast doesn't eat spicy food not because they are "wealthy," but because of the dry and cold climate; following the logic of the post, even if Manchuria were rich, the climate wouldn't make people crave spicy food.

6. Global counterexamples directly debunk the notion of "poor = spicy." South Korea, Thailand, Mexico, and India all enjoy spicy food; South Korea, now a developed country, still can't do without spice. Conversely, Sichuan cuisine is one of the most refined and coveted culinary styles in China, even with figures like Zeng Guofan secretly having chili added to their meals by the kitchen.

NVIDIA's huge profits aren't really coming from gaming GPUs.

First, let’s look at where the real money is coming from. NVIDIA's latest quarterly data center (AI chips) revenue hit $41.1 billion, accounting for around 88% of total revenue, while gaming and AI PC business only pulled in $4.3 billion. For the entire 2025 fiscal year, gaming revenue was $11.4 billion, just 8.74%, and in Q4 it was even lower at 6.36%. In other words, the 'massive profits' you see from NVIDIA come 90% from selling AI acceleration cards to companies like Microsoft, Meta, and OpenAI for training large models, with one card going for $30,000; gaming GPUs are just a sideline from twenty years ago, now barely a drop in the bucket.

So, 'why isn't Huawei trying to grab the gaming GPU pie?'—because that pie isn’t really juicy. Gaming GPUs are in a fiercely competitive, low-margin red ocean business; the real gold mine is in AI computing power.

Now, why isn’t Huawei getting into (gaming GPUs)? There are three core reasons:

First, manufacturing capacity is too precious to waste on gaming cards. After being sanctioned, Huawei can’t access TSMC or EUV lithography machines, so Ascend can only use SMIC’s 7nm process for production. This advanced capacity is extremely scarce; each wafer is a strategic resource—would you rather use it to make a chip that sells for $30,000 and can break through the US AI blockade, or to make a card that sells for a few thousand bucks, with thin profits, while competing with NVIDIA on performance and power consumption? The answer is clear. Huawei is putting all its chips on the AI data center battle.

Second, the moat for gaming GPUs isn’t in the hardware but in the ecosystem. NVIDIA’s gaming cards are hard to chase down, not because of the chips themselves, but because of over twenty years of refined drivers, DirectX optimizations, and compatibility with various games. Starting from scratch on this is a massive investment with slow returns, making it a poor cost-performance choice for a company that’s already facing supply chain issues.

Third, Huawei is actually working on GPUs, just not gaming cards. At the 2025 All-Connected Conference, Huawei announced that Ascend is transitioning from NPU to GPGPU, introducing SIMD/SIMT dual architecture, trying to be compatible with the CUDA ecosystem. They are going head-to-head with NVIDIA in AI computing power, not in the consumer gaming market.

Lastly, here’s a counterintuitive point: domestic gaming GPUs are actually being made, just not by Huawei. Moore Threads is developing domestic GPUs that support DirectX and can run games. So it’s not that 'China can’t make gaming GPUs,' but rather that Huawei did the math and chose not to.

Total volume second ≠ per capita wealth, that's the first layer. Total volume ranking measures the economic "size", while per capita measures how rich individuals are. With a population of 1.4 billion, when you spread the denominator, being second in size and seventieth in per capita can coexist without any contradiction.

Even "per capita" is inflated by high-income earners. Looking at resident income is more direct: by 2025, the national per capita disposable income is projected to be 43,377 yuan, while the median is only 36,231 yuan, with the median being just 83.5% of the average. Note that this is for the entire year, across all categories (including wages, business, property, and transfer income, also diluted down to the elderly and children). Converted, the median monthly average is about 3,000 yuan. This means that a salary of 3k-4k places you right at or even above the national average. The notion that "they're everywhere" is not an illusion, it's the statistical center.

Moreover, wages themselves don't make up a high percentage—average wage income is 24,555 yuan, accounting for 56.6% of disposable income, while the rest has to come from business, property, and transfer income. Relying solely on earned income results in even lower figures than the disposable income number.

The urban-rural and regional gaps further amplify the "low". Taking the median again, urban residents have a median disposable income of 51,115 yuan (approximately 4,260 yuan per month), while rural residents only have 20,711 yuan (about 1,726 yuan per month). The 3-4k positions you see on recruitment apps predominantly come from cities; once you sink into county towns and villages, those numbers are bound to drop further.

The industrial structure and labor supply determine the wage ceiling. China's comparative advantage has long been based on "controllable cost manufacturing + a huge labor pool". The more abundant the labor supply, the weaker the bargaining power for low-skill and service jobs, naturally keeping wages depressed. Adding to this, the proportion of resident income relative to GDP is itself low (in a high-saving, high-investment model, more shares flow to capital, government, and reinvestment, rather than household wages), making the slice of the "larger cake" that ordinary workers receive appear thin.

Finally, the visible "everywhere" carries a visibility bias. The "3-4k everywhere" on recruitment platforms is because the truly massive, low-barrier, high-turnover positions that require ongoing large-scale hiring are in services, retail, factories, and logistics—these jobs naturally need to be posted publicly and repeatedly. In contrast, high-paying positions often go through referrals, headhunters, or non-public channels, and their numbers are limited.

A classic case of 'full creativity, zero execution' hype.

1. The vehicle itself is illegally modified. A semi-trailer is a freight vehicle, and turning the cargo area into a restaurant with seats and a roof is a violation of the registered structure and appearance of the vehicle. The 'Road Traffic Safety Law' and 'Motor Vehicle Registration Regulations' clearly state: no unit or individual may arbitrarily change the registered structure, construction, or features of a motor vehicle. Offenders will be ordered by traffic management to restore the original state and fined. Such modifications can't pass inspection and are illegal on the road.

2. It doesn't actually 'tour.' 'National BBQ tour' sounds free, but it needs to find a spot to set up shop – and stopping means it’s occupying the road. Local authorities strictly regulate roadside stalls; outdoor dining in cities must be in designated areas and times, with a commitment letter signed, and outdoor cooking of food is prohibited. An unidentified truck setting up a BBQ stall could be shut down by city management, traffic police, or market regulators at any moment.

3. Obtaining a food business license is nearly impossible. Food service requires a fixed location, reasonable layout, and hygiene conditions to secure a food business license. A mobile truck without a fixed address or compliant kitchen will hit a dead end in the licensing process, falling under unlicensed operation.

4. Packed with people + open flames + gas tanks. This is the most dangerous part. Dozens of people sitting in a closed semi-trailer eating BBQ, with open flames beneath and gas tanks beside them – if there's a fire or a leak, there won’t be time to evacuate.

If they really dare to 'tour the nation with BBQ,' the first stop is likely not the next city but rather being asked to cooperate with an investigation.

This is a classic "only listing the losses" narrative. The conclusion is on point (Alibaba's investments have indeed lost more than they’ve won), but the answer to "why they don’t go bankrupt" is actually pretty simple: they’re losing on investments, but making bank on their core business, which is basically a money printing machine.

1. The core business generates cash flow way faster than the losses accumulate.

The business model for Taobao and Tmall is essentially selling ad space + earning commissions (listed in the financial report as "customer management revenue"). The gross margins are super high, and it doesn't require heavy asset investment. Alibaba rakes in hundreds of billions in operating cash flow and net profit every year. The losses you listed spread over a decade amount to only tens to a couple hundred billion a year, which their core business can easily cover. For example: if someone making 5 million a year loses 1 million in trading, it stings, but they’re nowhere near bankruptcy.

2. Most of the "losses" are paper impairments, not cash bleeding.

They spent 66.5 billion on Ele.me, and that money was paid out in the year of acquisition, funded by accumulated profits over the years, not borrowed money. A lot of the "losses" reported later are goodwill impairments—just a downward adjustment of the numbers on paper, not another cash outlay. Alibaba consistently has several hundred billion in cash and short-term investments on hand, maintaining a net cash position without a shaky financial structure.

3. This list itself has a survivor bias issue—it only lists the losses.

Alibaba also has big wins: equity in Ant Group, Alibaba Cloud turning from a money-burning operation to a business making billions, early exit gains from investments in Weibo, and Cainiao. If you delete all the wins and list only the losses, you can craft a narrative of "all losses."

4. Some of the "losses" are just the cost of doing business, not all failures.

Acquisitions like UC, Amap, and Youku were about securing traffic entrances and defensive positioning in the mobile internet era. Youku did indeed flop, but Amap is now a core asset for Alibaba's local services. Ele.me also integrated into the e-commerce group last year to compete against Meituan and JD, which has re-evaluated its strategic value. Of course, investments like Suning and Evergrande Football are just pure failures—nothing to salvage there.

The real question isn’t "why don’t they go bankrupt," but rather: if they didn’t blow those hundreds of billions on random purchases and instead focused on dividends or buybacks, how much better would shareholder returns be? That’s the math the market has been doing since Alibaba’s stock dropped from over 300—it's also why Joe Tsai and Wu Yongming have started selling off RT-Mart, selling Intime, and scaling back non-core assets.

The logic behind this chart is indeed a simplified version of Livermore's "testing the waters + pyramid adding" strategy from 'How to Trade in Stocks', not just some random idea. But if you follow it strictly, its true nature is worth a calculation:

First, let's verify the math in the chart.

The stop-loss line is correct: a 20% position falling 10% results in a total loss of 2%, calculated as 20% × 10%. No issues there.

What’s even more interesting is the line in the bottom right saying "liquidate immediately after a 10% drop from full position". Suppose you buy in at 100, then when it rises 10% to 110, you add 20%, and after another 10% increase to 121, you add another 20%. Confirming the trend around 133, you add 40%. At this point, if it drops 10% back to about 120, you liquidate—calculating overall you’d be roughly break-even, or even up 1% to 2%. So the essence of this method is: when you’re losing, only risk your test position; with larger positions, only press on trends that have been validated by the market. This captures Livermore's original intent—"never average down on losses, only add to winners".

But the chart doesn’t tell you the cost.

A choppy market is a meat grinder. This method works beautifully in a one-sided bull market, but the A-shares and crypto markets spend most of their time in a range-bound state. Buy → drop 10% stop-loss → rebound → buy again → stop-loss again; each time losing 2% doesn’t seem much, but after ten times, that’s 20%, and you end up selling precisely at the bottom. Livermore himself fell victim to this—he went bankrupt four times.

The 10% threshold isn’t universal. For stocks like Maotai, 10% is a major fluctuation; for small caps or altcoins, 10% is just daily breathing. Applying the same method uniformly is like using the same ruler for everything; the correct approach is to adjust the threshold based on the volatility of the asset (like ATR).

Buying more as the price rises is against human nature and raises costs. The average cost gets pushed close to the last buying price, meaning any normal pullback will break your psychological barrier. Most people can’t make it past the third step and fall apart—either they run early or are reluctant to liquidate.

A-shares have gaps and limit downs. The "liquidate immediately after a 10% drop" assumes you can always execute at -10%. A limit down or an overnight gap could mean your actual stop-loss could be -15% or even deeper, magnifying that theoretical 2% loss.

"Cut losses short and let profits run" is correct, but turning "a few big losses" into "many small losses" can still grind someone down in a market without a trend.

Lin Yuan at the Moutai shareholder meeting again shouted, "Moutai will continue to rise in price," arguing that more and more people can afford it while supply is far less than demand.

The data says it all: by 2025, the batch price for Feitian scattered bottles is expected to drop from 2220 to 1480, a 33% decrease for the year. This March, the official retail price did rise by 40 bucks (from 1499 to 1539), but now the batch price is around 1560, nearly matching the retail price— the scalper premium era is over, is this what they call "supply less than demand"?

By the way, it's rumored that Lin Yuan's products are down 17% this year, while the overall market rose from 2600 to 4200 during the same period.

What someone with a 40 billion position says at a shareholder meeting is just for show. The butt determines the brain; if he doesn't shout about price increases, that's the real news.

1. Don't emulate the big shots' "now", learn from their "back in the day"

The millionaires of today will tell you "you need to diversify your assets, buy insurance, and spread your investments", but when they were starting out, they did the opposite: they went all in on one industry, one business, one opportunity. Diversification is the logic of preserving wealth; concentration is the logic of stacking that first pile of cash. Below a million, your biggest assets are not cash, but time and the ability to handle failure—these big shots have lost those, so their advice naturally comes with a "preserve" bias.

2. Salary is just the base capital, not the endpoint

Below a million, no matter how high your investment returns are, they won't change your fate. A million at an annualized 10% only nets you 100k, but doubling your main or side hustle cash flow from 200k to 400k has an immediate impact. First, maximize your "money-making machine" (you) before you talk about making your money work for you. Most people get this order wrong; they earn 20k and start researching the "wealth freedom roadmap" of fund regular investments.

3. What the big shots won't tell you: most of them have stepped on a beta at least once

Real estate, the internet, mobile payments, cryptocurrency... almost all of those with over a million have benefited from some era of excess. This isn't to undermine their hard work, but rather to say: choosing the right pond is more important than fishing skills. Those below a million should spend less time thinking about "how to work harder" and more on "what is currently overlooked that will explode in the next decade"—avoid crowded places; the reverse of that advice is what truly holds value.

4. Protecting against losses is ten times more important than dreaming of getting rich

Those with under a million can't afford to lose, yet they're often the ones getting harvested the hardest: high leverage contracts, P2P, Ponzi schemes, health product-like investment courses. A single big loss can wipe out five years of hard work. A big shot can lose 5 million and still be a big shot; you lose 500k and it could ruin your marriage, mindset, and health.

Been cutting deals for years, never asked you to lend me cash.

The tightest platform out there actually doesn’t offer lending. 😅

There’s a public secret on the internet: the endgame of traffic is lending.

Social media hits a ceiling with lending, food delivery hits a ceiling with lending, rideshare hits a ceiling with lending, even input methods and weather apps are looking to extend credit. The logic is simple—lending profits crush all other core businesses.

Some say it’s because their users are too deep down the funnel, lending risks are too high to recover;

Others say they make enough from sales, so they don’t care about interest income;

And then there are those who say not lending is a real way to treat users like people. 🥹

Pinduoduo has officially responded that "we don’t have any consumer credit business"; the buy now, pay later feature is linked to third-party products like WeChat and Huabei. However, the Pinduoduo merchant backend does have loan functionality, just that regular consumers can’t apply for it. So the precise wording is "the only platform that hasn’t done consumer-facing lending business."

Brick-and-mortar shops are tough to run, not because people are out of cash, but because the info gap has closed. 📉

In the past, you didn't know the wholesale price in Yiwu, so a tie could go for 400.

Now, the factory price and the production cost are all laid out on the screen, and the middlemen making a profit from the price difference have been completely uprooted.

But have you noticed that a certain type of shop is thriving:

Barbershops, cafes, gyms, massage parlors…

Because the experience and service can't be delivered via express shipping. 📦❌

So the way forward for brick-and-mortar stores has never been about "selling products", it's about offering "things that can't be provided online".

Any business still banking on information gaps is on a countdown. ⏳