Slow down.

Not to find the stock that moved the most. Not to chase the loudest AI headline. Not to pretend I can understand an entire sector from one revenue-growth percentage and a celebratory chart.

I slow down to separate fact from feeling.

This half-year review is focused on ten Hong Kong-listed AI names across computing infrastructure, semiconductors, memory, optical communications, PCBs, electronic materials, and foundation models.

These are some of the market’s hottest themes — which is exactly why they deserve colder thinking.

The central question is not simply:

Which company grew the fastest?

It is:

Which company is already earning real money from AI — and which is still selling the market a future that has not yet arrived?

As a Hong Kong Professional Investor, I have learned that an exciting story is not automatically a bad investment.

But an exciting story without cash flow, margins, or a credible return on capital can become an extremely expensive form of entertainment.

Two Different AI Trades

The clearest takeaway from this earnings season is that “AI” is not one investment category.

There are at least two very different businesses hiding under the same label.

The first is the AI application and model layer: companies building large models, APIs, enterprise services, and consumer AI products.

The second is the AI infrastructure supply chain: optical modules, PCBs, server components, electronic materials, foundry capacity, memory interfaces, and fibre networks.

Both can grow quickly.

But they carry very different economics.

For model companies, revenue may be rising at extraordinary rates while R&D, inference cost, sales expenditure, and computing expense remain even larger. The product may be real, demand may be real, and the loss may also be very real.

For infrastructure companies, the picture is more immediate. AI data-centre spending has already flowed into orders, pricing, capacity utilisation, gross margins, operating profits, and sometimes cash flow.

That does not automatically make hardware safer.

It just means the revenue has shown up before the dream.

MiniMax: Growth With A Bill Attached

MiniMax reported first-half revenue of approximately USD 117 million, up 283.1% year on year. Gross profit rose faster, by 464.8%, and gross margin improved from 12.1% to 17.9%.

That is meaningful progress.

But then I look at the bill.

The statutory loss was roughly USD 358 million. Adjusted net loss was around USD 293 million. R&D expenditure rose to approximately USD 297 million.

This is the central question I would keep asking myself: for every additional dollar of revenue, how much computing, research, and customer-acquisition cost is required?

High revenue growth is encouraging.

But high growth does not mean a company has escaped the “spend heavily first, explain profitability later” phase.

For MiniMax, I would watch gross-margin expansion, inference cost, customer retention, and whether losses narrow as revenue scales.

The business appears to be moving.

It is just not yet clear how much cash it will require to keep moving.

Z.ai: The API Engine Is Starting

Z first post-listing interim results were more encouraging at the commercial level.

First-half revenue reached RMB 954 million, up 399.7% year on year. Its open-platform and API business generated RMB 825 million, up 2,735.7%, representing 86.5% of total revenue. Token usage rose by more than 40 times, while average API pricing increased by 101%.

That is not merely a slide deck.

That is usage.

More importantly, the API segment’s gross margin moved from negative 0.4% to positive 24.6%. It may only be a first step, but moving from “every transaction loses money” to positive gross profit is a meaningful one.

Still, the consolidated gross margin fell from 50.0% to 26.4%, while R&D expense reached RMB 2.13 billion. The statutory net loss remained substantial at RMB 2.07 billion.

So I would not call the business proven yet.

I would call it commercially alive.

The next test is whether API scale leads to better overall margins, lower R&D intensity relative to revenue, and a path toward self-funded growth.

Revenue can grow at 400%.

The question is whether the economics eventually grow up with it.

The Hardware Reality

While AI-model companies are still proving monetisation, several hardware names have delivered something the market understands immediately:

Revenue.

Profit.

Cash flow.

And, in some cases, dividends.

建滔積層板（01888）reported first-half revenue of HKD 14.9 billion, up 55%, with net profit rising 209% to HKD 2.89 billion. AI servers, high-speed networking, electronic fibreglass, copper foil, and copper-clad laminates all benefited from stronger demand and pricing.

The company’s vertically integrated structure mattered. When competitors faced material shortages and delivery constraints, Kingboard could continue producing at near-full utilisation.

That is what an operating moat looks like in an industrial cycle: not a glossy presentation, but the ability to deliver when everyone else is scrambling.

長飛光纖光纜（06869） also posted striking numbers: first-half revenue rose 53.6%, while attributable profit jumped 889%. Gross margin reached 53.4%.

That sort of profit growth tells me product mix, pricing, and capacity utilisation have moved together.

But it also tells me to become more cautious.

When profits rise nearly ninefold, I do not only ask, “How good is this?”

I ask, “How much of this is repeatable?”

High-cycle earnings often look permanent at precisely the moment they are most vulnerable to competition, supply expansion, and price normalisation.

Cash Flow Is Harder To Fake

勝宏科技（02476） stood out because its earnings were supported by cash generation.

Revenue rose 28.8% to RMB 11.63 billion. Net profit rose 33.3% to RMB 2.86 billion. Operating cash flow was approximately RMB 2.91 billion.

For a manufacturing company expanding capacity rapidly, operating cash flow keeping pace with earnings matters.

It reduces the risk that growth is merely being booked on paper while working capital quietly becomes the real business model.

The AI server, high-density interconnect, and high-layer-count PCB opportunity appears genuine.

But I would still watch customer concentration, capacity additions, depreciation, and pricing carefully. A strong order book is useful. A strong order book at an inadequate return on capital is less useful.

The market tends to remember the first one and forget the second.

Semiconductors: Better, But Still Cyclical

華虹宏力 (01347) Hua Hong Semiconductor delivered a recovery that was hard to ignore.

First-half revenue reached USD 1.378 billion, up 24.5%. Gross profit rose 83.2%, while gross margin improved from 10.1% to 14.8%. Profit attributable to owners of the parent rose more than fourfold.

The drivers were clear: expanding 12-inch capacity, improved product mix, rising wafer shipments, and higher average selling prices.

That looks more substantial than a simple accounting rebound.

But foundry economics remain cyclical. New capacity can arrive. Customers can adjust inventory. Price competition can return. Margins are improving, but they are not yet close to historic peaks.

A good recovery is not the same as a permanently better industry.

I keep reminding myself of that because the market usually stops reminding me precisely when I need it most.

Optical Modules: The Market’s Favourite Exam

中際旭創（03308） may have produced one of the most powerful reports in the AI optical-communications supply chain.

Revenue rose 182.5% to RMB 41.78 billion. Net profit rose 241.7% to RMB 13.65 billion, driven by demand for 800G and 1.6T optical modules and higher cloud-capex spending from overseas customers.

Those are exceptional operating results.

The question is not whether the business is performing.

It is whether the valuation already assumes that it will perform exceptionally forever.

That is a much harder standard.

For companies like Zhongji, I would watch technological migration from 800G to 1.6T and eventually 3.2T, customer concentration, product leadership, gross margins, and the pace of hyperscaler capex.

The business can remain excellent while the stock still suffers if the market’s expectations were even more excellent.

That is one of investing’s less charming features.

Memory: Great Results Can Still Lose Money

兆易創新（03986）reported a spectacular first half: revenue up 178.7%, net profit up 1,091.5%, adjusted net profit up 796.9%, and operating cash flow up 531.5%.

Then the share price reportedly fell sharply after the result.

That is not irrational.

It is the market asking whether the memory-price cycle has already been priced in.

In memory, earnings can look strongest when supply discipline, pricing, and inventory dynamics all align. But the same forces can reverse when capacity expands or demand slows.

I would focus less on the headline profit percentage and more on adjusted profit quality, memory-price sustainability, inventory, and the durability of domestic-substitution demand.

A 1,091% profit increase is spectacular.

It is also a reminder that percentages can become theatrical when the starting base was small.

The Quietly Important Name

瀾起科技（06809） numbers were less explosive, but still strong.

First-half revenue grew 26.7%, gross profit grew 36.9%, and attributable profit rose 72.3%. Its interconnect-chip and server-platform businesses continued to benefit from AI-computing demand.

This is the sort of report I find useful because it shows operating leverage: profit growth outpacing revenue growth.

But the market’s reaction was cautious after the results, partly because the share price had already performed strongly and because investors were scrutinising dividend momentum and adjusted-profit trends.

That is the uncomfortable truth of a hot sector:

A company can report good news.

A stock can still fall.

The market does not only price what happened. It prices the gap between what happened and what had already been expected.

What This Earnings Season Says

Three themes stand out to me.

First, AI infrastructure is already producing visible earnings across the supply chain.

Optical modules, fibre networks, PCBs, electronic materials, memory interfaces, and semiconductor capacity are seeing direct demand from data centres and accelerated-computing investment.

Second, AI-model companies and AI-hardware companies are different investment questions.

Model companies such as MiniMax and Zhipu may be building important platforms, but they remain in a period of high R&D, high computing costs, and incomplete monetisation.

By contrast, many hardware suppliers are already reporting real revenue, real profit, and real cash flow.

Third, cyclical profits are not permanent profits.

When I see record margins in electronic materials, optical communications, foundry services, or memory, I do not dismiss the numbers.

I simply ask:

If industry demand falls by 30%, does this company still earn an acceptable return on capital?

That question is much less exciting than saying “AI is changing everything.”

It is also more likely to protect me when everything starts changing again.

Becoming Worthy Of Good Stocks

Charlie Munger once expressed an idea I keep returning to: good partners are rare, so the best way to find one is to become someone worthy of a good partner.

I apply a version of that to investing.

The best way for me to find a good stock is to become an investor worthy of holding one.

That does not mean predicting every quarterly result.

It means doing the quiet work:

• Understanding what the company actually sells

• Distinguishing revenue from cash flow

• Looking at margins rather than headlines

• Understanding the supply-and-demand cycle

• Checking customer concentration

• Tracking capital expenditure and depreciation

• Asking whether growth increases per-share value

• Remaining willing to revise my view when the facts change

Good companies are rare.

Good companies at sensible prices are rarer.

And investors who can hold them calmly through noise may be rarer still.

I do not need to have an answer to every market headline immediately.

Some questions require the next earnings report. Some valuations need sentiment to cool. Some investment theses need time to prove themselves.

For now, I will keep doing the work: reading, writing, recording, and updating my view.

Not because I expect to be right every time.

But because I would rather make a decision based on logic than allow the market’s mood to make it for me.

Views are personal and for discussion only. They do not constitute investment advice.

Alibaba’s HKD 80 Billion Placement: The Market Isn’t Selling the AI Story. It’s Pricing Trust in Capital Allocation.

Alibaba is placing 710 million new shares at HKD 112.70 each, raising roughly HKD 80 billion.

The stated destination: full-stack AI — chips, infrastructure, and model capabilities.

The transaction is expected to complete on 26 August.

That sounds like a straightforward AI headline. It isn’t.

The market is not necessarily saying, “AI is a bad idea.”

I certainly am not.

For a company with Alibaba’s cloud platform, merchant ecosystem, enterprise relationships, engineering depth, and balance sheet, not investing seriously in AI could be the more dangerous decision.

But “AI is necessary” and “any way of financing AI is justified” are two very different sentences.

That distinction is where the real investment work begins.

A placement is not free money

Equity financing has an obvious attraction: no coupon, no maturity date, no lender calling at an inconvenient moment.

But it is not free.

A placement permanently expands the share count. Future earnings, dividends, free cash flow, and the effect of buybacks must now be divided among more owners.

For an existing Alibaba shareholder, the question is not simply:

How much dilution is there?

The better question is:

Will the HKD 80 billion create enough additional per-share cash flow to justify the ownership given away today?

If it does, this placement may eventually look like a cheap ticket into a valuable new profit pool.

If it does not, then management has transferred part of its execution risk to existing shareholders — politely, efficiently, and at a discount.

Capital markets have many elegant ways to ask shareholders for patience.

Alibaba is not short of cash

Alibaba is not raising money because it is standing at the edge of a cliff.

But cash on the balance sheet and sustainably generated free cash flow are not the same thing.

In the latest quarter, operating cash flow was approximately RMB 22.9 billion. Capital expenditure rose to roughly RMB 67.7 billion, up 75% year on year. That left free cash flow at an outflow of about RMB 44.7 billion.

In plain English: the core business still generates cash, but the company is spending far more on GPUs, servers, and data centres than it is bringing in during the quarter.

That is not automatically bad.

If I am building a bridge that will collect tolls for decades, heavy upfront investment can be entirely rational.

If I am filling a hole that becomes larger each time I shovel money into it, then effort is not the same as progress. Neither is capex.

AI infrastructure is not a software business in the pure sense. A software product can often serve another customer at very low marginal cost. Compute is closer to a power plant or hotel: equipment must be bought, data centres built, electricity paid for, systems maintained — and the equipment starts depreciating before the excitement around the launch event has even faded.

Meanwhile, competitors can buy GPUs too.

The sequence I am watching

When I look at Alibaba’s AI spending, I do not want to stop at “AI revenue is growing.”

I want to follow the full chain:

Capital expenditure
→ Cloud revenue
→ Gross margin
→ Operating cash flow
→ Free cash flow per share

The last two words matter most: per share.

A company can become larger while its shareholders become poorer.

If revenue rises, but capex rises faster — and new shares rise faster still — then the business may look more impressive from a distance while creating very little additional value for the people already on the register.

Alibaba Cloud’s external commercial revenue reportedly grew 45% year on year. AI-related products reached an annualised revenue run-rate of more than RMB 49.5 billion, representing around 35% of Alibaba Cloud’s external commercial revenue.

Those are meaningful figures. AI is no longer merely a slide deck with heroic gradients.

Customers are paying.

But I do not see those numbers and immediately declare victory. I ask myself:

• How much of that ARR reflects durable demand rather than promotional pricing?

• What do retention and renewal rates look like?

• Are margins improving alongside revenue?

• Is utilisation of the new compute capacity increasing?

• How quickly can this investment become cash, rather than the justification for the next round of capex?

ARR is a run-rate. Capex is cash already spent.

They are not directly comparable, but the contrast matters: Alibaba remains in the phase of putting large amounts of money into the ground, rather than harvesting large amounts of cash from it.

The real examination begins when equipment depreciates, cloud competitors cut prices, and customers decide whether they still want to pay.

And then what?

This is the question I keep returning to.

Not: “Is the placement bullish or bearish?”

Not: “Is AI good?”

But:

And then what?

Financing is a tool. It is not a virtue.

Debt, equity, retained cash — all can be sensible. All can be destructive. Getting a bigger credit-card limit does not prove that I know how to allocate capital. It merely gives me more ways to disappoint myself.

Capital markets will fund a company when they believe one dollar invested today can eventually create more than one dollar of value.

So, for this HKD 80 billion, I will be watching:

• How much additional free cash flow per share appears within three years

• Whether compute utilisation rises with capacity

• Whether cloud economics survive a price war

• Whether hardware becomes obsolete before it earns an adequate return

• Whether management slows spending if returns disappoint

• Or whether another placement becomes the default answer

That final point matters.

One dilution event can be strategic.

Repeated dilution can become a governance signal.

The old bill: instant retail

The market’s reaction is not only about AI. It is also about memory.

Investors have seen this film before:

First comes the story — GMV, users, market share, expansion.

Then comes the spending — subsidies, infrastructure, incentives.

Then free cash flow begins to weaken.

Then the company returns to the market for more ammunition.

And only at the end does everyone ask the awkward question: what did the shareholder actually receive?

HSBC has estimated that Alibaba’s instant-retail competition may have generated cumulative losses of around RMB 87 billion from the second quarter of 2025 through the first quarter of 2026. That is an external estimate, not Alibaba’s own single reported figure, but it captures what the market fears: a subsidy battle consuming capital while AI infrastructure demands even more.

This is why shareholders may connect the food-delivery war and the placement, even if management treats them as separate strategic decisions.

The concern is not irrational:

Did Alibaba dig one hole through low-return competition, then ask shareholders to finance a different, more attractive hole labelled “AI”?

That may sound unkind. It is also a capital-allocation question worth asking.

Instant retail may have strategic value. It can defend consumer entry points, improve engagement, strengthen fulfilment, and prevent competitors from encroaching on Alibaba’s core commerce ecosystem.

But every strategy has a price.

Every dollar spent on subsidy is a dollar not used for repurchases, dividends, high-return projects, or simply retained until a real opportunity appears.

If subsidies create a durable moat, improve customer lifetime value, and lead to better unit economics, they may be an investment.

If they merely buy short-lived orders — while competitors match the discounts and customers leave when promotions disappear — then they are not an investment.

They are rented excitement.

What I will measure

Over the next few quarters, I will pay less attention to daily price movements and more attention to five things:

1. Cloud revenue growth versus capex growth — Revenue growth is helpful; capex growing much faster may mean scale without value.

2. AI gross margins — Is growth purchased through lower pricing, or is the economics improving?

3. Compute utilisation — Data centres are like hotels: a beautiful building full of empty rooms remains an expensive building.

4. Instant-retail unit economics — Are orders, customers, and cities moving toward profitability?

5. Free cash flow per share and share count — Are buybacks truly shrinking the share base, or being offset by placements and stock compensation?

Alibaba has suggested that payback periods on AI infrastructure may decline from roughly three years to two and a half years or less.

I will treat that as an operating hypothesis, not a fact.

Payback depends on utilisation, pricing, margins, depreciation, competition, and discipline. A spreadsheet can make a payback period look very attractive. Reality, regrettably, does not always read the spreadsheet.

Ticket or burden?

I do not think Alibaba investing seriously in AI is absurd.

I think failing to invest could be reckless.

But the necessity of AI investment does not automatically make this placement value-accretive.

If, over the next two or three years, Alibaba can demonstrate sustained AI-cloud growth, rising margins and utilisation, capex converting into operating cash flow, recovery in free cash flow per share, better instant-retail economics, and a share count that begins shrinking again — then today’s dilution may eventually look like a well-priced entry ticket.

If capex keeps rising, hardware becomes obsolete before earning its cost of capital, cloud services enter a price war, instant retail keeps burning cash, and Alibaba returns to the equity market again, then HKD 80 billion may not be a ticket.

It may be a burden.

The Munger-style error I try hardest to avoid is confusing spending money with creating value, or confusing a bigger company with a more valuable share.

When I see the word “fundraising,” I try not to rush toward “bullish” or “bearish.”

I ask:

And then what?

After the money is invested, who receives the cash return?

If the answer is not the existing shareholder, then even the grandest AI story may simply be someone telling a compelling story with my money.

Views are personal and for discussion only. They do not constitute investment advice.

Many will see this as another Wall Street story about tighter employer control.

From first principles, it is really about where the most valuable asset in investing is moving.

The question is not simply whether Citadel wants to retain employees. It is: what does a modern investment firm actually rely on to generate returns?

In multi-strategy, quantitative, and data-driven firms, the value chain is moving upstream.

The scarce capability is increasingly not knowing how to place a trade. It is understanding why the trade should exist.

Who can extract a repeatable, scalable, risk-controlled signal from datasets, industry developments, company information, and market noise? Who can distinguish useful evidence from an attractive but meaningless correlation? Who knows when a strategy is likely to fail?

Very often, that knowledge sits with the analyst or research team.

Today’s analysts do far more than update models, summarise earnings calls, or write reports. They work across data sources, research frameworks, industry knowledge, and strategy iteration.

More importantly, they possess one of the hardest assets to protect: tacit knowledge.

Code can be locked on a server. Data can be restricted. Documents can be classified. But an employee’s understanding of market structure, intuition about a signal, and memory of prior failures do not disappear upon resignation.

That is why Citadel’s concern is not simply that an analyst might leave. A competitor may spend a few million dollars on compensation and gain research capability that took years of talent development, data investment, computing power, and costly trial and error to build.

Now invert the problem: how does an investment firm lose its edge?

Not necessarily through one bad trade. It loses its edge when strategies are replicated, alpha is arbitraged away, key people leave, and competitors capture the benefits of research without bearing the original cost of discovery.

A lengthy non-compete is therefore not necessarily designed to stop someone from ever joining a competitor. Its purpose is to buy time.

Alpha has a half-life. If a departing employee is out of the market for 12 to 24 months, a strategy may already have weakened through crowding, capital inflows, and changes in market structure.

Citadel is not merely locking up an employee. It is allowing knowledge to expire.

The broader question: if a leading quant fund now treats analysts as core assets requiring stronger protection, will Wall Street’s most valuable people be those who can distinguish signal from noise under uncertainty?

And then what?

What looks like an internal employment policy could reshape talent dynamics across the hedge-fund industry.

The most likely first consequence: other large funds may follow.

Competitive logic may push them there. Every fund wants the freedom to hire talented researchers from rivals. Yet no fund wants its own researchers moving across the street with accumulated knowledge, experience, and relationships.

If a powerful firm such as Citadel raises its defensive barriers, firms that do not may become more exposed to talent leakage.

This is a classic prisoner’s dilemma. The industry may benefit from talent mobility, but for an individual firm, locking the door first can appear safer.

The second consequence is that career mobility will be repriced.

A leading fund role has traditionally offered more than high compensation. It offered a career option: perform well and move to a larger platform, become a portfolio manager, join a family office, enter PE, or eventually manage capital independently.

With non-compete periods, that option is no longer free.

Compensation may remain attractive, but part of it effectively becomes payment for reduced mobility. That cost is rarely obvious when signing an offer letter. It becomes real when someone most wants the freedom to leave.

The rational question is not only, “What happens if everything goes well?” It is also: if the strategy changes, the team restructures, or the role proves unsuitable, is there still a reasonable exit path?

That is the real significance of a non-compete.

Third, AI may make exceptional analysts more valuable, not less.

AI can process documents, summarise earnings calls, collect data, assist with modelling, and test hypotheses. But investing is not about producing more answers. It is about recognising which answers should not be trusted.

When every fund can access similar models, datasets, and computing power, the scarce capability becomes the ability to ask better questions, identify spurious correlations, understand causality, define risk boundaries, and say “I do not know” when the market is most confident.

Tools will become widespread. Judgment will not.

For investors, they also raise a question: how much of that edge resides in a small number of individuals?

The strongest institutions should not rely on contracts alone. They build systems that generate ideas continuously: sound incentives, disciplined risk management, robust research infrastructure, and an environment where talented people choose to stay.

Citadel’s move may be only the first domino.

In a world where data, models, and AI are increasingly accessible, the most valuable asset is still not the machine. It is the person who can make fewer avoidable mistakes under uncertainty.

As AI reshapes investment research, the key question for a single-family office is no longer simply:

Which CIO has the strongest track record?

It is:

Who can make sound long-term capital-allocation decisions when information is abundant, markets move faster, and AI-generated “analysis” becomes cheap?

A strong résumé shows past achievement. But a family investing across generations needs more: someone who combines technology, independent judgment, and experience across market cycles.

That combination is becoming rarer.

AI Will Polarise Investment Talent

AI will not raise the value of every analyst equally. It will likely polarise the profession.

For investors with strong fundamentals, deep industry knowledge, and the ability to distinguish causation from correlation, AI is a powerful lever. It can accelerate research, process data, monitor portfolios, generate hypotheses, and improve scenario analysis.

But AI does not create judgment.

It can produce a polished investment thesis without knowing whether the original question was sensible. It can detect historical patterns without recognizing data mining, temporary market structure, or weak economics. It can summarize management commentary without fully understanding incentives, culture, or what remains unsaid.

The future value of an investment professional will not lie in producing more content. It will lie in asking better questions, independently validating evidence, identifying weak assumptions, and knowing when not to act.

An effective CIO should combine three capabilities.

First: AI fluency with human accountability. The right candidate should use AI to scan information, monitor risks, map industries, and accelerate research—but never outsource critical judgment to a machine.

Second: a multidisciplinary mental framework. Investing is not merely valuation. An apparently inexpensive company may still face poor incentives, regulatory disruption, weakening competitive advantages, refinancing risk, or political uncertainty.

Third: experience across changing conditions. Many strategies look effective when liquidity is abundant and markets rise. Experience matters when assumptions fail, liquidity contracts, correlations converge, and a sound thesis meets an unexpected reality.

A CIO is not merely an external investment professional. For a single-family office, they are a long-term capital-allocation partner. They must understand the family’s return objectives, liquidity needs, governance, risk tolerance, and intergenerational priorities.

AI will make research faster and investment commentary more plentiful. That is exactly why judgment, multidisciplinary thinking, and experience will become more valuable.

The scarce asset will not be the person who produces the most analysis. It will be the person who asks better questions, makes fewer avoidable mistakes, and protects the family’s long-term option value.

In June, the government proposed tax amendments aimed at expanding the preferential regime for funds, family offices and carried interest. The direction is clear: Hong Kong wants to strengthen its appeal not only to private-equity managers, but across the wider alternative-investment ecosystem.

That could include hedge funds, venture capital, private credit, family offices—and, potentially in time, proprietary trading firms and their top traders.

The key is carried interest.

Under Hong Kong’s current framework, eligible carried interest can receive highly favourable tax treatment, including full exclusion from salaries-tax assessment for qualifying employees. But the conditions matter. This is not a mechanism for simply renaming ordinary salary or bonus as “carried interest”; eligibility, qualifying transactions and substantial-activity requirements remain central.

The strategic question is bigger than tax.

Hong Kong is competing with Singapore and other financial centres for people who can allocate capital, build investment platforms and generate returns across cycles. A favourable carried-interest regime may attract managers. Extending incentives toward proprietary trading could attract liquidity, quantitative talent and more risk-taking capital.

But it also raises a harder question.

If Hong Kong becomes increasingly attractive as a trading, quant and short-term-performance hub, what becomes of long-term value investing?

For family offices, the scarce talent may not be the person who can trade most frequently, react fastest, or monetise every market dislocation.

It may be the person who can do the opposite:

• Define a genuine circle of competence

• Reject opportunities outside it

• Understand a business, industry and incentive structure in depth

• Remain patient when the market rewards activity over judgment

• Protect capital from permanent loss

• Execute a long-term thesis without confusing volatility with failure

Trading talent and long-term capital allocation are both valuable. They are simply different crafts, with different time horizons, incentive systems and failure modes.

Hong Kong should absolutely compete for financial talent. But the city should also ask what kind of capital culture it wants to build.

A market full of fast money can be vibrant.

A market capable of patient capital can build institutions, companies and enduring value.

The strongest family offices know the difference.

And then what? Assume Hong Kong succeeds.

Citadel, Jane Street and a new generation of global quant funds, proprietary trading firms and AI-driven investment platforms begin expanding aggressively in the city. Capital arrives. Talent arrives. Compensation rises. The language of Central shifts further toward systematic trading, market-making, data, latency, models and execution.

Then what?

For Hong Kong’s bankers, analysts and investment professionals, this could be a genuine opportunity. More firms mean more mandates, more liquidity, more clients, more careers and more capital moving through the city.

But it could also mean something else: a far more brutal form of competition.

Because the standard changes, it is no longer enough to be intelligent, hardworking or well-connected. Those qualities may become merely the entry ticket.

Charlie Munger’s most useful lesson was never that everyone should become a great investor. It was that everyone must understand their own talents, their own limits, and the field in which they can genuinely become exceptional.

Trying to win in a field where you have no natural advantage, no accumulated knowledge and no genuine interest can turn a career into a mess. No amount of fashionable language—AI, quant, systematic, alternative assets—changes that.

A discretionary investor should not pretend to be an AI engineer, a quant trader, or a trading talent. And a long-term investor should not abandon a well-researched thesis simply because somebody else is making money faster.

There may be more money for everyone. But there will almost certainly be more pressure, more comparison and more temptation to become somebody else.

That is where the real test begins.

Hong Kong may be entering an era it has not experienced before: one shaped simultaneously by global quant firms, AI, alternative capital and increasingly sophisticated financial infrastructure.

Many of us are experiencing it for the first time.

So perhaps the honest response is not false certainty. It is this:

“I am doing my best. But I have never lived through a Hong Kong transformed by quant trading and the AI revolution. This is my first time too. I do not yet know whether I will navigate it well.”

Intellectual honesty is where any durable edge begins: knowing what I do not know, deciding what game is actually mine to play, and having the discipline to stay with it long enough for judgment to compound.

For family offices, this may become the decisive distinction.

Do they want the person who can trade every dislocation? Or the person who knows when not to trade at all?

Both may be valuable. But they solve different problems.

One extracts value from movement. The other protects capital across time.

Hong Kong will need both. But in a city increasingly rewarded for speed, the ability to wait may become rarer—and more valuable—than ever.

Today, it has crossed RMB 10 million at the box office. Some forecasts now point to RMB 100 million, after it became a viral phenomenon across Chinese social media.

No serious budget.
No institutional backing.
No professional marketing machine.

Its title—sounding like “the bull market is coming”—collided perfectly with market sentiment. Investors shared it. Memes spread. The audience arrived.

If those numbers hold, its MOIC may be among the most extraordinary cultural-investment outcomes anyone has seen.

Overnight, it looks like a miracle.

But it was not made overnight.

That is what investors tend to forget.

We love the moment a company breaks out: the funding round, the IPO, the viral chart, the ten-bagger. Once the outcome is visible, everyone calls it vision.

Before that moment, it may look like very little:

A founder with an unfinished product.
A team solving a problem few consider urgent.
A business with weak early metrics but unusually loyal users.
A technical insight with no obvious market yet.

For VCs, angel investors, and family-office CIOs, the challenge is not finding companies that already look like gold.

It is deciding when something might become gold—while accepting that most things will not.

That is why the work is not simply about finding exciting stories. It is about staying close to the right ecosystems, meeting founders before the deck is polished, following an industry long enough to recognize what is genuinely different, and separating a real problem from a fashionable narrative.

The mathematics matter:

• MOIC: How many times did the capital become?
  

• IRR: How quickly did it compound?
  

• ROIC: Can the underlying business reinvest capital at attractive returns?
  

But no spreadsheet removes uncertainty.

A brilliant founder can still fail.
An excellent product can arrive too early.
A huge market can remain inaccessible.
A good business can still be a bad investment at the wrong price.

For a family office, I would reduce the entire discussion to two uncomfortable questions.

If you were the CIO, would you have had the conviction to bring this opportunity to your UHNW principal before the public discovered it?

And if you were the principal:

Would you have written the cheque?

Not after RMB 10 million of box office.
Not after the memes.
Not after the consensus had formed.

Before any of that.

This is where risk management becomes real.

Too much caution, and the family never participates in exceptional compounding.

Too much conviction, and one seductive early-stage investment becomes an expensive lesson in concentration, illiquidity, and ego.

A capable CIO must be able to say:

“This is worth watching.”
“This is worth a small first cheque.”
“This is a good story, but not our investment.”

The real test is not whether a CIO identifies every future winner.

It is whether they can preserve the family’s capital—and its ability to keep looking—after several stones prove to be stones.

Gold may eventually shine.

Before it does, someone must be close enough to notice it, skeptical enough not to romanticize it, and disciplined enough to survive being wrong.

AND THEN WHAT?

For a family office, an angel investor, VC, PE, IPO, secondary, or anyone deploying their own long-term capital, finding an interesting opportunity is only the beginning.

The harder question is:

How do you participate without letting one beautiful story damage the entire portfolio?

Patient capital is often described as a family office’s advantage.

It can be.

There is no quarterly earnings call.
No pressure to show an exit next year.
No need to follow every public-market fashion.

But patient capital without discipline can become merely slow capital.

Too willing to fund a beautiful story.
Too reluctant to admit a mistake.
Too distracted by paper marks to notice weakening liquidity.
Too eager to call every early bet “strategic.”

The best CIOs do not promise to identify every future winner.

They build a process that allows the family to:

• See opportunities before they become consensus
  

• Say “no” to most of them without regret
  

• Size early bets so mistakes remain survivable
  

• Reserve capital for the rare cases where evidence strengthens
  

• Hold on when an exceptional business begins to compound
  

That is why I still return to a simple principle:

Even angel investing is value investing.

Not because every early-stage investor should obsess over next year’s earnings.

But because the central question remains unchanged:

What is this business truly worth?
What must happen for that value to emerge?
And what am I paying for the right to participate?

At an early stage, value may sit in a founder’s judgment, a product wedge, customer obsession, a technical moat, or a market that has not yet become obvious.

It is still value.

The investor’s task is not to worship the story.

It is to understand the economics beneath it—and to pay a price that leaves room for reality to be less perfect than the pitch.

That is one reason Niu Lai feels personal to me.

It was made over five years by a mother and son, working by hand, without a professional production team, outside capital, or a marketing machine.

After spending a decade deep in the medical-device sector, I am now building Lattice & Moat in much the same spirit: patiently, independently, and without assuming that attention will arrive on schedule.

A blockbuster? A flop? A GOLD? A lesson? I do not yet know what it can become within Hong Kong’s Chapter 18A market.

But meaningful work usually begins before there is an audience, a consensus, or a valuation attached to it.

Keep doing the work while it is still invisible.

Then let time decide what is truly worth compounding.

到咗 2020 年尾、2021 年， COVID 令全世界困喺屋企。 股市升。 加密貨幣升。 連平時唔講投資嘅人，都開始講股票同幣。

「To the moon」 由一個小圈子笑話， 變成 Reddit、狗狗幣、 Gamestop、 Elon Musk 共同催生嘅全民口號。 打開手機，係火箭 emoji。 打開群組，係「To the moon」。 短短一年， 變成一整代人嘅集體信仰。

同一時間， 美國有個九十幾歲嘅老人家， 喺公開場合一次又一次咁鬧。 查理・芒格唔鍾意加密貨幣。 非常唔鍾意。 佢話，呢類嘢像一場 莊家有近乎百分百優勢嘅對賭。

佢仲話:「有人賣**俾你，你都仆住去搶。 」

嗰陣我聽完， 覺得呢個老人家係咪唔識新科技。 全世界都數緊錢。 你講啲咁嘅嘢？

後來我問咗自己一條問題： 去到月球之後，然後呢？

我先察覺，「To the moon」其實只成功咗一半。

因為去月球， 只是任務的前半段。 太空人真正嘅任務， 永遠有個更重要嘅後半段： 安全返返嚟。 你去得幾遠，去得幾快， 如果冇一條返轉頭嘅路， 個任務就係失敗。

芒格當年畀好多人笑 「死老嘢唔識新科技」。 大家只顧衝去月球， 好少人問： 然後呢？ 更少人準備好回程嘅燃料。

後來 FTX 爆煲， 加密市場入咗寒冬。 好多以為自己搭上火箭嘅人， 先發現架船原來 冇留燃料畀佢返嚟。

到嗰陣，我先明白芒格點解咁口硬。

BECAUSE TIME WILL ALWAYS TELL.

佢最鍾意講： Invert, always invert. 反過來想。

唔問 「呢樣嘢可以升到幾高？」 先問 「行咗呢步，返唔返到轉頭？」

LEGEND之所以係LEGEND， 唔係因為佢講過幾多金句。 而係佢揀中咗一條問題， 可以跨過好幾個周期， 仍然適用。 人可以走。 問題會留下來。 唔止 to the moon， 仲要記得 back to home。 去程。 回程。 兩樣都要識。

中環呢個資本江湖， 好多人出事，未必係因為一次大錯。 而係因為太耐冇問過佢自己：行咗呢步，然後呢？ 返唔返到轉頭？

當年Bitcoin 嘅狂熱，結果呢？
今日 SpaceX , SK hynix, 緊隨其後。
OpenAI來勢洶洶 然後呢？

Time will tell.

《Lattice and Moat》

Lattice是乜嘢

Charlie Munger有一套理論，叫「latticework of mental models」。佢認為，你要真正睇通一件事，唔可以淨係靠一個學科嘅知識，而係要將心理學、經濟學、歷史、物理學嗰啲模型，編織成一個網，用嚟交叉檢視現實。單一角度好容易被表面現象騙到，但當你用幾個唔同嘅框架去照同一件事，隱藏嗰部分先會浮出來。

Lattice唔係抽象理論，係一套工具。過去幾集《表裡見聞錄》，我哋其實已經悄悄用咗好幾個lattice：機會成本、to the moon、專業投資者（PI）、Tiffany櫥窗。今日，我想小試牛刀，直接用呢幾個lattice拆開最近一星期最震撼嘅事件，南方東英SK海力士每日槓桿（2x），股票編號7709。

一個閃耀嘅商品華麗出場

上個月，韓股掀起一場史詩式暴走。SK海力士，靠HBM高頻寛記憶體嘅AI敘事，股價一路坐火箭，帶動整個南韓槓桿ETF市場陷入癲狂。市場公開報道，超過一百萬名韓國散戶捲入呢場槓桿風暴，有啲俾人追孖展，有啲甚至被迫強制平倉，損失慘重。而喺香港，南方東英發行嘅7709，正正就係呢場敘事嘅香港版本。

先簡單介紹下7709究竟係乜。毫無疑問，佢係今年全港最多人炒賣嘅股票，佢都係今年股市嘅Tiffany櫥窗，最閃耀、最令人心動嗰件展品。
原理超級簡單：將SK海力士股價每日放大2倍去賭。就係咁簡單一個結構，做到每日成交過百億，短短半年多啲時間，由10蚊一路飆到接近200蚊，然後又急速插水跌返落20蚊，再喺噚日單日升接近一倍。呢個過程之中，有人一夜之間變成百萬富翁，亦有人傾家盪產。同時間，亦養肥咗唔少基金經理。係呀，佢係一隻ETF，代表背後始終有人穩穩收管理費，贏輸都同佢無關。快

記得《珠光寶氣》嗰個開場畫面：Holly站喺珠寶店門外，望住櫥窗入面嗰啲鑲滿寶石嘅頸鏈，渴望躋身入嗰個世界。今日嘅散戶，望住手機屏幕上7709嗰條升到痙攣嘅K線圖，眼神同Holly望住Tiffany櫥窗，其實冇兩樣。同一種渴望，同一種距離感，同一種「我想擁有」嘅衝動。分別只係，Holly清楚知道自己冇能力擁有Tiffany嗰啲寶石，但好多買7709嘅投資者，未必知道自己冇能力承受呢件商品背後嘅風險。

Tiffany櫥窗嘅另一面：PI限制

呢一點，帶出第二個lattice，專業投資者（PI）。

喺金融監管邏輯入面，槓桿倍數越高、波動越大嘅產品，理應設有更高嘅門檻。事實上，證監會今年七月已經修訂通函，容許槓桿產品發行商喺市況急劇波動時，主動調低槓桿倍數，變相承認呢類產品本身承受咗超出一般散戶理解嘅風險。換句話講，監管機構自己都明白，呢類每日再平衡嘅2倍槓桿產品，設計目的從來唔係俾一般散戶長期持有，理論上應該更貼近專業投資者（PI）嘅使用場景。

但現實係，7709最終被包裝、宣傳、推廣成一件人人都可以買、人人都應該買嘅產品。有網民公開質疑，今年六月中曾經有一批財經KOL大力宣傳7709，見到產品其後大跌，會唔會出嚟交代。呢一句質疑，其實正正指出咗問題核心：一件本應限制流通嘅高風險商品，透過社交媒體嘅放大效應，變成人人觸手可及嘅「入場券」。

To the moon，然後呢？

第三個lattice，「to the moon」，呢句我哋喺加密貨幣狂潮時期都見識過嘅羊群心理，今次原封不動搬到7709身上。SK海力士本身股價喺AI敘事下升到飛天，而經過2倍每日槓桿嘅放大效應，7709一度升幅十倍。社交媒體上，一張又一張截圖，展示住短短幾個月由幾千變幾萬嘅「戰績」。

問題係，to the moon之後，然後呢？

每日槓桿產品有一個殘酷嘅數學結構。佢跟蹤嘅係「每日」回報，而唔係「總」回報。當標的資產劇烈波動，升完又跌，跌完又升，每日再平衡機制會產生「損耗」（decay），令長期持有嘅槓桿ETF回報，同標的資產嘅實際走勢出現巨大落差。呢個結構性缺陷，喺平穩上升市場入面唔會被察覺，但一旦標的資產劇烈震盪，殺傷力會被放大好多倍。

韓國嗰邊超過一百萬人捲入呢場風暴，被追孖展，強制平倉。好多人喺股價急跌時先發現，自己手上嗰隻槓桿產品，跌幅遠遠超乎想像。我相信喺香港，一樣有幾萬人正在經歷同一場劇本。分別只係地點由首爾換咗做中環，語言由韓文換咗做廣東話。

每個投資者背負咗嘅機會成本

第四個，亦係最重要嘅lattice，機會成本。

買7709嘅人，唔係傻。佢哋大部分都做咗選擇，用一筆錢，換取一次快速致富嘅機會。但每一個選擇背後，都有一條隱藏嘅經濟學賬單。

如果你將一筆本金放咗喺7709，你放棄咗嘅，可能係一個更穩健、複息效應更強嘅長期倉位，可能係一次冷靜研究SK海力士基本面、決定直接持有正股嘅機會，甚至可能係一次冷靜觀察市場情緒過熱、暫時離場等待更好時機嘅選擇。

機會成本嘅殘酷之處，在於佢從來唔會即時顯現。你贏嘅時候，唔會覺得自己付出咗任何代價。你輸嘅時候，先會發現，自己不只輸咗本金，仲輸埋本來可以行嘅另一條路。

呢個現象，仲有一個更隱蔽嘅層面，好少人留意到。當全城嘅注意力、資金同情緒，全部被7709呢類閃耀商品吸走嗰陣，市場另一邊，往往會發生一件完全相反嘅事。譬如阿芒果隻愛股，平時成交量唔多，冇人講，冇KOL炒作，股價一路穩打穩紮。但就係過去一星期，市場突然出現咗幾個大額賣盤。有趣嘅係每次有呢啲賣盤出現，都會有人喺幾分鐘內，靜靜地將貨掃走。

呢個先係機會成本最諷刺嘅一面。當大部分人將注意力同資金，投入去追逐一件會發光嘅商品嗰陣，真正嘅長期投資者，反而喺無人問津嗰個角落，靜悄悄咁完成佈局。你放棄咗嘅，唔止係另一種投資方式咁簡單，你可能仲放棄咗一次「趁全城冇留意」入手優質資產嘅時機。市場嘅注意力係一種稀缺資源，佢流向邊度，往往決定咗邊個先摸到便宜貨。

大家都望住閃耀商品嗰陣，好多冷清嘅細舖D貨反而悄悄轉咗手。

六十五年前嘅歷史與文化，今日一字不改

1961年，《珠光寶氣》上映嗰陣，冇人講HBM、冇人講AI敘事、冇人講每日槓桿再平衡。但戲裡面嗰個核心命題，一個人願意用幾多代價，去換取靠近財富嘅感覺，六十五年後，一個字都冇改變。

柯德莉夏萍用身份、用青春、用一次又一次嘅社交表演，去換取一種「屬於上流」嘅幻覺。今日買7709嘅人，用本金、用槓桿、用一次又一次 reload App嘅衝動，去換取一種「跟上時代」嘅幻覺。工具變咗，敘事變咗，但人性入面嗰條方程式，完全冇變。渴望永遠跑得比理性快，代價永遠喺你贏嗰陣睇唔見，輸嗰陣先現形。

1961年嘅觀眾，望住Holly企喺珠寶店門外食早餐，覺得浪漫。2026年嘅投資者，望住7709條K線飛天，覺得自己搭上咗時代嘅順風車。兩個畫面，隔咗六十五年，講嘅其實係同一句話。真正嘅風險，從來唔在於你有冇眼光揀中飆升嘅資產，而在於你有冇算清楚，自己為咗呢個飆升，究竟押上咗幾多樣本來屬於你嘅選擇。

劇本，
六十五年前已經寫好。
今日只不過換咗戲名。

Disclaimer： 本人從來沒有，亦永遠不會持有7709。以上內容純粹作為觀察同思考框架分享，唔構成任何投資建議。