Dismantling the three deadly variables behind the American Historic Philosophy

Author:Black Analytics Ltd XMerkle 3s Capital

 

Let's start with an unusual picture。

The moment the newspaper released, the numbers were so beautiful — record numbers, Māori rates so high for years, and each share of the proceeds so far above Wall Street’s most optimistic mouth. There was a cheer in the social media, and the family began to intercept their hold, and the title party prepared the word "money-printer"。

And? The first reaction to smart money is to slow down。

That's not what we assumed. That's what really happened last quarter. As a result, stock prices did not rise or fall, and nearly 30 per cent fell back in about a week. An epic report card, in exchange for an epic sale。

Why? Because the market never buys "how much has it made in the past" but "can we keep doing it in the future"? For a company that is labeled as a "cycle share", the best financial returns tend to occur before the worst point. The moment when profits peak is also the most dangerous moment in valuation logic — the fear that lies in the bones of the semiconductor industry。

So this time, we're in no hurry to celebrate。

We would like to ask a more serious question: is the windfall of this round of storage a return to another cycle or a real paradigm shift? Is it an old story of rising tides and retreating, or is the sea level raised forever

The market is still priced with the old ruler, "cycle share." But if the ruler itself is wrong, cheapness may be a trap and expensive may be the starting point。

This article, we have three rulers, one by one. Put the financial figures on the table, then devalue them, and lastly look at the variable that's not in the financial report, but that might determine everything。

The numbers are clear: put the money on the table

Before evaluating, let us do one of the simplest things — to set the figures straight. No emotion, no conclusion, but the truth。

THIS QUARTER (FY2026 Q3), US$41,456 MILLION IN CORE DATA WAS HANDED OVER BY US$41,446 MILLION, COMPARED TO +346 PER CENT, RING +74 PER CENT; MĀORI RATE 84.9 PER CENT; EARNINGS PER SHARE $25.11; FREE CASH FLOWS $18.3 BILLION; AND NET PROFITS $28,860 MILLION。

It's enough to look at this line — it's not a "good" financial paper, it's a financial paper that redefines beauty at a quantitative level。

To understand the weight of these figures, it depends on what kind of comparison they are. Whether or not a financial report is "super-foreseeable" has never been compared with last year's, but rather against three anchor points: prior guidance from the company itself, market expectations on Wall Street bets, and last quarter's ring comparison. We put this three-dimensional in a table。

It's really worth watchingMāori rateThis line。

A record record is not unusual for a storage plant in its previous cycle — upscaled, upscaled, and naturally well-paid. But the Māori rate is different. The Māori rate measures "the net drop of each $100 chip sold", which directly reflects pricing rights. Can a company keep profits in its pocket while increasing prices instead of being squeezed off by upstream equipment and downstream customers。

Historically, the Māori rate of beauty is famous for the roller coaster. At one of the worst times in the industry, it was sold at a loss, and the Māori rate was even negative; at the best, it rarely stood steady. So 84.9% of this season, if it's real, it means much more than "more" -- it's answering a deeper question: Is there any real price-fixing in Hong Kong this time。

In addition to the master table, there are a few numbers that we suggest you write down separately。

• Progress of HBM operations: As the core of this round of AI narratives, HBM's share of revenues, delivery tempo, and HBM4's production schedule are key to determining how long "AI dividends will last". This season, HBM4 12-high has reached twice the rate of HBM3E 12-high, and it is expected that the rate of maturity will be much faster than HBM3E, and HBM4 has received more than $1 billion in goods this season。

• GUIDANCE FOR THE NEXT QUARTER: NEXT SEASON (FQ4-26) TAKES 50 BILLION US DOLLARS (US$ 1 BILLION), WITH A MĀORI RATE OF ABOUT 86 PER CENT, EPS 31.00 US DOLLARS (US$ 1) AND CAPITAL EXPENDITURE OF ABOUT US$ 10 BILLION. MARKETS WILL ALWAYS LIVE IN THE FUTURE, AND THIS SEASON WILL BE BETTER THAN NEXT SEASON'S MANAGEMENT RHETORIC。

• CAPITAL EXPENDITURE ARRANGEMENTS: FQ3 NET CAPITAL EXPENDITURE $7.1 BILLION, FQ4 IS PROJECTED TO BE AROUND $10 BILLION, AND FY2026 ABOUT $27 BILLION FOR THE YEAR. LATER ON, WE WILL SAY SPECIFICALLY THAT IT IS PROBABLY THE MOST NEGLECTED, BUT THE DEADLIEST, LINE IN THIS NEWSPAPER。

The numbers are done. Objectively, it is a report card that is enough to be included in the history of beauty。

But we must give you a sobering shot in advance: the movement of the last eight quarters has proved time and again that it was never the performance itself that determined the price of the United States share. There is almost a difference between record harvests, expectations and post-fiscal stock prices. We left this in the chapter "A valuation trap" and told you in eight real newspapers。

The more beautiful the report card, the more vigilant we are..It never translates automatically into an epic return. What's at stake is what the market prices it with。

Next, we'll tear this ruler down。

The market got the wrong ruler

IF YOU TURN ON ANY OF THE GO-LIVE SOFTWARE, YOU'RE GOING TO SEE THE BEAUTY OF THE FORWARD-MARKETING RATE -- IN TERMS OF FUTURE PROFITS, IT'S PROBABLY ONLY 10 TIMES HIGHER。

FOR MOST STOCKS, TEN TIMES THE PE IS A "CHEAP." A TECHNOLOGY COMPANY, WHOSE PERFORMANCE WAS STILL BREAKING OUT, VALUED 10 TIMES AS MUCH AS IT SOUNDED LIKE A PIE IN THE SKY. THAT'S HOW MANY PEOPLE ARE ATTRACTED。

But this is precisely the most dangerous signal for the periodic stock。

We have to be clear about a counterintuitive thing:THE PERIODIC LOW PE IS NOT CHEAP, IT'S A PROFITABLE ALARM。

HERE'S THE LOGIC. PE EQUALS THE SHARE PRICE DIVIDED BY THE PROFIT PER SHARE. WHEN A CYCLE COMPANY IS AT THE TOP OF THE LANDSCAPE, THE DENOMINATOR -- PROFIT -- IS PUSHED TO THE HIGHEST IN HISTORY, AND EVEN IF THE STOCK PRICE IS NOT LOW, THE CALCULATED PE IS VERY SMALL. THE MARKET IS NOT A FOOL. IT LOWERS THE VALUATION MULTIPLIER PRECISELY BECAUSE IT KNOWS THAT THIS HIGH PROFIT WILL NOT LAST, AND THE RATE WILL PROBABLY FALL NEXT YEAR AND YEAR AFTER. LOW PE IS THE MARKET'S EARLY DISCOUNT ON "PROFIT COMING DOWN"。

ON THE OTHER HAND, AT THE BOTTOM OF THE CYCLE, COMPANIES MAY LOSE OR MICRORATE, THE DENOMINATOR CLOSES TO ZERO OR EVEN NEGATIVE, AND THE CALCULATED PE WILL BE OUT OF PROPORTION, OR EVEN 100 TIMES HIGHER. THE STOCKS LOOKED EXPENSIVE WITHOUT FRIENDS, BUT OFTEN THE GOLD PIT。

We don't believe we look at our own history。

IN 2018, STORAGE INDUSTRY TOPS, LUMINOUS, FORWARD-LOOKING PE WAS CRUSHED LESS THAN THREE TIMES - ALMOST ONE OF THE MOST CHEAP HOLDINGS IN THE WHOLE MARKET. AND? OVER THE NEXT YEAR, THE STOCK PRICE FELL FROM THE TOP TO ABOUT 70 PERCENT. THE MAN WHO BOUGHT THAT "CHEAP" THREE TIMES THE PE LOST HIS LIFE IN DOUBT。

AROUND 2024, THE INDUSTRY JUST CAME OUT OF ITS DEEPEST LOWS, AND THE PROFITS WERE NOT REPAIRED, AND THE PE WAS COUNTED OVER A HUNDRED TIMES, AND IT LOOKED RIDICULOUSLY EXPENSIVE. AND? IN THE NEXT PARAGRAPH, THE STOCK PRICE IS ALMOST TENFOLD OUT。

IN THE WORLD OF PERIODIC SHARES, THE PATTERN IS COLD: BUYING AT HIGH PE, SELLING AT LOW PE。

It's the opposite of all the investment books that taught you to buy cheap, sell expensive. Because the textbook speaks of a stable profit-making company, and the profits of the cycle are in themselves breathable -- what you think is expensive, it breathes at the bottom of the valley; what you think is cheap, it holds its last breath at the top of the mountain。

SO, OUR FIRST REACTION TODAY WAS NOT TO BE EXCITED, BUT TO BE ALERT WHEN SOMEONE POINTED TO THE LIGHT 10 TIMES AS FORWARD-LOOKING AS PE SAID, "SEVERAL UNDERESTIMATION, CLOSE-EYE BUYING." IF MIGWANG IS STILL THE OLD CYCLE STOCK, THEN THIS LOW PE IS A REPLICA OF 2018 -- THE MARKET IS PRICING A PROFIT COLLAPSE AHEAD OF SCHEDULE。

But -- and that's the key turn of the article --What if Mi-go isn't the old-timer

IF THIS ROUND OF HIGH-PROFITS IS NOT A FLOWER AT THE TOP OF THE CYCLE, BUT IS TIED UP BY A LONG-TERM, AI DEMAND, SUPPLY PATTERN, THEN THIS TEN-FOLD PE IS NO LONGER AN ALARM, BUT A REAL CHEAP PRODUCT THAT WAS ACCIDENTALLY KILLED BY THE OLD FRAME。

SAME NUMBER, TWO FATES. TEN TIMES MORE PE IS SWEET OR VERY TOXIC, NOT DEPENDING ON THE NUMBER ITSELF, BUT ON WHICH KIND OF COMPANY YOU JUDGE。

AND TO MAKE THIS JUDGMENT, JUST LOOK AT THE PE IS USELESS. WE HAVE TO TAKE OUT THREE SHARPER FEET TO MEASURE ITS BONES -- THE LENGTH OF THE ORDER, THE RESILIENCE OF PROFITS, THE PATTERN OF SUPPLY。

First ruler, first order。

First ruler: locked orders. Can you lock the cycle

Why does the storage industry write the word "cycle" on its face

'Cause it used to be a " spot business." Chips, like bulk commodities, are available today at whatever price the customer buys and the manufacturer sells. Demand is hot, people are struggling to expand; energy is running out of supply, prices jump; prices collapse, producers cut down and investments cut off; and prices rise again after supply shrinks — again and again, after six months, no one can see. This business model of "not seeing the future" is a cyclical source。

BUT THERE'S ONE THING THAT'S QUIETLY CHANGING THE RULES OF THE GAME — THE LTA。

LET'S EXPLAIN IT IN ONE SENTENCE: THE LTA IS THAT THE CUSTOMER NO LONGER BUYS ON HIS OWN, BUT SIGNS LONG CONTRACTS WITH THE MANUFACTURER ONE YEAR OR MORE, LOCKS THE AMOUNT, LOCKS THE PRICE AND SOMETIMES MAKES AN ADVANCE PAYMENT. TO BE HONEST, IT'S CHANGING THE SPOT STAND TO A BUN-YEAR CAFETERIA。

BUT WHAT REALLY DESERVES IS SOMETHING MORE THAN THE LTA -- STRATEGIC CLIENT AGREEMENT。

IT'S THE DIFFERENCE BETWEEN A REGULAR LTA, NOT JUST A NAME THAT'S MORE SCARY. TRADITIONAL LTAS ARE OFTEN SHORT-TERM, RELATIVELY LOOSE "INTENTIONAL LOCKS" AND SCAS ARE MULTI-YEAR DEEP BINDINGS, SOME SIGNED FOR UP TO FIVE YEARS. MOVING THE BUSINESS ONE STEP FURTHER FROM THE YEAR'S CAFETERIA, THE SCA IS MORE LIKE A FIVE-YEAR ENGAGEMENT BETWEEN A CLIENT AND A MANUFACTURER -- NOT MORE THAN A YEAR, BUT A LONG WAY TO GO。

BUT THE WORD "MULTI-YEAR" HAS A VERY DIFFERENT AMOUNT OF GOLD. A REALLY HARD SCA, AND AN EMPTY "WE LOOK AT LONG-TERM NEEDS" THAT'S WORSE THAN THE DETAILS HIDDEN IN THE TERMS OF THE CONTRACT. SO THIS FINANCIAL REPORT, WHEN MANAGEMENT TALKS ABOUT THE LONG TERM, WE'RE GOING TO DO THE FOLLOWING THINGS WORD FOR WORD:

• is there a clause to take-or-paythis is the most crucial one. take-or-pay, it's a "take or pay" translation - the customer has to pay even if he doesn't take everything. with this one, it's from "i'm going to buy" to "i'm going to have to pay" and from "i'm going to buy" to "i'm going to pay for it."。

• Do you have a spareDid the client lock up a certain line, a certain percentage of the capacity, and leave it to himself? Once retained, this supply is removed from the market。

• Any advance paymentsWould the client like to pay for the pit first? The size of the advance is the most honest thermometer of demand — strong and worthless, paid first。

• is there a price guaranteeEven if the future spot price falls, is there a floor price in the contract? With a price floor, the Māori rate would not be hit back by a swing in demand。

• Did you cancel the penaltyWhat's the price of a client trying to break a contract in the middle of a trip and cut a bill? The heavier the penalty, the more genuine the order is。

These five questions come together, essentially asking one thing:This long contract, is it a gentleman's agreement or a contract with teeth

Why are we staring so hard? This directly determines how high the market is willing to give a valuation to beauty。

It's very simple. If management can convince the market that a significant proportion of the profits in 2027 are "contracted in black and white" – taken-or-pay to the bottom, priced floors, broken contracts, and paid for – the certainty of that share of the profits is no different from the subscription income of a software company. The market has never paid more for certainty than for cyclicality. Once the concept of "contracted profits" is in place, ten times the PE, this old ruler is completely dead。

If, in turn, management turns around and says "very strong demand, very large clients" without ever disclosing any particular article, then we have to reap the excitement — the vague optimism, the logic of overestimation。

It was never a "strong demand" that could hold up a higher valuation, but rather the terms of the contracts with teeth: either to collect the goods or to pay for them。

This is much more lethal than it looks。

FIRST, IT MAKES THE MANUFACTURER "SEE THE FUTURE" FOR THE FIRST TIME. IN THE PAST, AMERICA USED TO PLAN AND SEE AT MOST THE NEXT QUARTER; NOW, HBM, A HIGH-END PRODUCT, HAS BEEN RAISED FAR BEYOND THE REACH OF THE ORDER -- US LIGHT 2026, HBM IS REPORTED TO BE NEARLY 100% SOLD OUT, AND SOME OF THE ORDERS HAVE EVEN BEEN SCHEDULED TO 2027. THIS MEANS THAT THE FACTORY IS NOT WORKING, AND THE NEXT YEAR OR TWO ARE ALREADY BOOKED。

SECOND, IT'S TURNING THE GAME UPSIDE DOWN. IN THE SPOT AGE, IT WAS THE PRODUCERS WHO WANTED TO BUY IT; IN THE SCA AGE, IT WAS THE CUSTOMERS WHO WERE AFRAID THEY COULD NOT BUY IT AND WHO WERE WILLING TO PREPAID IT. THE SUPPLY AND DEMAND INITIATIVE SLIPPED FROM THE BUYER TO THE SELLER。

THIS CHANGE IS NOT UNIQUE TO BEAUTY. THE SIGNAL WE'RE SEEING IS INDUSTRY: HERCULES WAS PASSED ON TO A MEGA-CLIENT LIKE MICROSOFT TO SIGN A MULTI-YEAR DDR5 SUPPLY AGREEMENT, AND SAMSUNG WAS BOUND BY A LONG CONTRACT WITH A BIG CUSTOMER. WHEN THE PLANT'S CAPACITY IS DIVIDED IN ADVANCE BY A MULTI-YEAR AGREEMENT, THE CLASSIC VERSION OF "CLOSE WHEN THERE IS EXCESS CAPACITY" IS BEING REWRITTEN, AT LEAST IN HIGH-END PRODUCTS。

Long-term binding is much more than just price and quantity. Here is a particularly interesting detail: Mizuki is working with Anthropic to optimize the bandwidth of the memory subsystem - HBM - for Claude's training and reasoning - and how it's stretched, how it's working, how it's working, how it's working。

What does that mean? The relationship between the top client and Mizuang is not a trade relationship between me and you, but a co-research relationship between us and my model. Once the memory subsystem is perfected for your model's depth, you want to switch suppliers, the costs and risks are so high。

It's never just a price. When memory begins to be customized for a family model, it is locked in a technology, a moving wall, a deep binding that can't walk。

This brings up the core of the first rulerLocked orders. Can you lock the cycle

We tend to split the answers in half, not in black or white。

THE HALF THAT CAN BE LOCKED IS HIGH END. HBMS, SUCH AS HIGH BANDWIDTH STORAGE FOR AI TRAINING, HIGH TECHNICAL BARRIERS, SCARCE CAPACITY AND HIGH CONCENTRATION OF CUSTOMERS (FOR EXAMPLE, A FEW GIANTS OF AI CHIPS). SUCH PRODUCTS ARE NATURALLY SUITABLE FOR LONG-TERM — THERE IS AN INCENTIVE FOR PRODUCERS TO RECYCLE LARGE AMOUNTS OF R & D AND CAPACITY INVESTMENTS FOR FEAR OF LOSS OF SUPPLY AND FOR BOTH SIDES TO PERPETUATE THE RELATIONSHIP. IN THIS AREA, IT IS TRUE THAT THERE HAS BEEN A STRUCTURAL SHIFT IN PRICING POWER。

THE HALF THAT CAN'T BE LOCKED IS STANDARD. ORDINARY DDR5, STORED IN MOBILE PHONES AND PCS, REMAIN RELATIVELY STANDARDIZED COMMODITIES. IT CAN BE LOCKED IN A LONG SECTION, BUT IT IS DIFFICULT TO COMPLETELY DISENGAGE FROM THE GRAVITY OF THE SPOT PRICE. ONCE CONSUMER E-DEMAND IS WEAKER, THE CYCLICALITY OF THIS COMPONENT WILL RETURN。

SO, INSTEAD OF ELIMINATING THE CYCLE, THE SCA HAS "SCATTERED" THE CYCLE -- THE HIGH END IS ABOUT TO BE HELD UP, AND THE STANDARD IS STILL IN THE WATER。

And that's why we have to read every word in this financial paper about the compact word by word. On this occasion, the details given by management were hard enough than expected: Misaki has signed 16 strategic client agreements (SCAs) covering the three main areas of data centres, consumer electronics, automobiles, usually five years (2026-2030), and automobiles for about three years. Together, these agreements cover about 20 per cent of DRAM production and about one third of NAND production; of these, 14 SCAs have a cumulative minimum price income of about $100 billion and client advances and pledges amount to $22 billion. More crucial is the structure - these agreements use take-or-pay (either to take delivery or to pay for), and the largest agreement has both price caps (locking to the current market price of CQ2) and price thresholds. Management expects that about half or more of the company ' s revenues will be covered by the SCA once all the SCAs have landed。

what were we supposed to do word for word — what proportion of capacity is covered by the long term? is it the number or the price? do you have any of these teeth? how clear is the visibility in 2027? has the size of the advance increased? this time, almost every article was answered positively:

the capacity is locked in one-fifth to one-third, the price is up and down, take-or-pay bottom, $22 billion in advance cash, and can be pulled directly to 2030I don't know. When "a substantial proportion of the profits have been contracted out" went from a slogan to a $100 billion contract amount, that's one of the toughest proofs of a paradigm shift, not a vague "demand strong."。

The order is long, this ruler, and it's "Can you see in the future."。

But seeing orders does not mean making profits. The second ruler, the other thing we're going to measure -- this round of wind is "the color."。

Second ruler: supercycle, or recycle

The most noisy word on the market now is "super-cycle."。

IT MEANS THAT THIS TIME AI BROUGHT STORAGE NEEDS THAT WERE NOT JUST A GENERAL UPSWING, BUT A "SUPER" UPS AND DOWNS THAT WERE BIGGER AND LONGER ENOUGH TO BREAK OLD PATTERNS. SOUNDS BEAUTIFUL. BUT WE MUST FIRST SET A STANDARD OF JUDGEMENT, OTHERWISE THE DEBATE WILL BE RHETORICAL。

Our standard is simple and harsh:The super-cycle is also a re-cycle, depending on how much price rises, and the Māori rate falls。

Why the Māori rate, not the price

Because price increases happen in any ordinary cycle, it's not unusual. It's really rare, it's really possible to distinguish between "species" and "backlights" by the ability of the manufacturer to hold high profits when demand retreats. If this round is really structural, then even if demand fluctuates during a quarter, the Māori rate should be resilient and not return overnight to pre-liberation. In turn, if the Māori rate drops at the sight of the wind, it is essentially the same old cycle, but the wave is a little higher。

So we're looking at a figure -- 84.9 percent of the Māori rate, and how it's going in the next few quarters。

Look at the evidence that sustains the supercycle。

• DATA CENTRES ARE SWALLOWING MOST OF THE PLANT ' S CAPACITY. AI HAS SEVERAL TIMES MORE APPETITE FOR MEMORY THAN TRADITIONAL SERVERS, AND INDUSTRY-RELATED DATA CENTRE-RELATED NEEDS CONSUME A SIGNIFICANT PROPORTION OF DRAM CAPACITY (ABOUT 70% LEVEL). THIS SEASON'S DATA CENTRE ALONE HAS COLLECTED MORE THAN $25 BILLION, AND THE ANNUALIZATION HAS EXCEEDED $100 BILLION -- AN AMAZING GROWTH CURVE IN ITSELF。

• THE ARRIVAL RATE OF THE AI SERVER HAS INCREASED ALARMINGLY, MORE THAN DOUBLED A YEAR (NEAR THE +180% SCALE), AND EACH OF THE AI SERVERS IS STORED AS A GOLD SWALLOWER。

• HIGH-END PRODUCTS SUCH AS HBM, ALTHOUGH NOT THE LARGEST SHARE OF HARVESTS, ARE EXTREMELY "EATING THE CRYSTAL ROUND" - THE SAME CRYSTAL ROUND AS HBM, WITH OUTPUT CAPACITY WELL BELOW NORMAL DRAM, WITH HBM IN THE INDUSTRY ACCOUNTING FOR ABOUT 20% (~23%). IT'S EQUIVALENT TO DISGUISING STANDARD DRAM SUPPLIES, SUPPORTING THE OVERALL PRICE。

• There is also the most intuitive thermometer — stock rotation. In a healthy state, the storage plant ' s stock is equivalent to approximately 8 to 12 weeks ' worth of deliveries; now, some high-end product stocks are under pressure for 3 to 5 weeks. It is the most honest signal of supply and demand that the goods are stolen as soon as they come out。

PUTTING THESE TOGETHER, THE "SUPER-CYCLE" STORY REALLY MAKES SENSE: AI CREATED A PREVIOUSLY NON-EXISTENT, RIGID, LONG-TERM DEMAND FOR STORAGE, WHICH STRUCTURALLY RAISED THE INDUSTRY'S PROFITABILITY。

But when it comes to this, it is necessary to dispel the most widespread misunderstandingA LOT OF PEOPLE THINK THIS ROUND IS THE PRODUCT OF HBM, NOT REALLY。

A COUNTER-INTUITIVE FACT, REVEALED BY THE MANAGEMENT ITSELF, IS THAT GENERAL DRAM, OR GENERAL DDR5, IS NOW MORE PROFITABLE THAN HBM. IN OTHER WORDS, THE HBM, WHICH HAS BEEN WIDELY REPORTED, IS NOT YET THE MOST PROFITABLE PIECE OF PROFIT ALONE。

WHAT DOES THAT MEAN? IT MEANS THAT THE BLINDLY PADDY, MUCH OF WHICH IS NOT THE CONTRIBUTION OF HBM AT ALL, IS A GREAT STRAIN FROM TRADITIONAL DRAM. AS STATED EARLIER, THE THREE GIANTS HAVE MOVED THE ROUND TO HBM, AND THE SUPPLY OF ORDINARY MEMORY HAS BECOME LESS AND LESS EXPENSIVE, WHILE THE PRICE AND PROFITS OF STANDARD GOODS HAVE BEEN PUSHED TO HISTORIC HEIGHTS。

This is precisely the most undervalued piece of evidence of the "supercycle": the dividend of this roundIT'S NOT ABOUT HBM, IT'S ABOUT A MONOLOGUE, IT'S ABOUT THE WHOLE STORAGE BUSINESSI DON'T KNOW. FROM THE TOP HBM TO THE MOST COMMON DDR5, ALMOST EVERY SHELF IS UP AND MAKING MONEY. A SINGLE-STAR-BASED PROSPERITY IS FRAGILE, AND A WHOLE CLASS OF PROSPERITY BENEFITS FROM IT AT THE SAME TIME。

So, what's the real meaning of HBM? The answer lies in a much lower technology logic -- Token Economics. And that's one of the hardest technical arguments that we think "this might really be different."。

Let's be clear in white. In the age of Agenic AI, AI is no longer just answering one sentence, but is going to think, to call tools, to run a lot. To measure the usefulness of this AI, the core indicator has become a word: token throughput -- in short, the speed of AI "publish" -- how much it can process and generate every second. AI, think fast, how many users can be served in a second。

and what determines the speed of the token is a simple, almost brutal formula:

Token speed = HBM capacity x HBM bandwidth。

TRANSLATION INTO HUMAN LANGUAGE: THE AI MODEL HAS TO THINK FIRST BY MOVING THE LARGE AMOUNT PARAMETERS AND CONTEXT INTO THE COMPUTING CHIP FROM THE MEMORY. THE CAPACITY OF HBM DETERMINES THE SIZE OF THE MODEL AND THE LENGTH OF THE DIALOGUE; THE BANDWIDTH OF HBM DETERMINES THE SPEED WITH WHICH DATA ARE MOVED. THESE TWO, THEY'RE JUST STUCK IN THE AL "PUSHING" CEILING。

A LOT OF PEOPLE WOULD ASK, "CAN'T SOFTWARE BE OPTIMIZED?" THE ALGORITHM IS GETTING SMARTER. CAN'T IT SAVE A LITTLE MEMORY? YES, BUT THERE ARE LIMITS. SOFTWARE OPTIMIZATION CAN MOVE UNDER THE CEILING, BUT NEVER BREAK THE HARDWARE-DECIDED CEILING. AS LONG AS AI TRIES TO RUN FASTER AND SERVE MORE PEOPLE, IT CAN ONLY PILE MORE, FASTER HBMS. IT'S NOT A BUSINESS MODEL CHOICE. IT'S A PHYSICAL LAW。

AND THAT'S THE HARD POINT OF HBM'S NEED, WHICH MAY REALLY BE DIFFERENT: IT'S NOT DRIVEN BY A SINGLE-TIME CAPITAL EXPENDITURE, BUT IT'S LOCKED DOWN BY THE BOTTOM FORMULA OF AI PERFORMANCE. AS LONG AS AI GETS STRONGER, THE THIRST FOR HBM WILL NOT STOP。

THERE'S AN EQUALLY NEGLECTED PUZZLE OF SSD AND NAND IN THE DATA CENTRE。

Your eyes are focused on HBM, but very few people notice that AI is also whispering the needs of corporate-level flashlights. AI's reasoning is to access big data, the vector database is to store astrometric vectors, and the so-called KV Cache offload (in short, to remove from expensive memory to SSD for temporary storage) — all of this is a crazy drain on enterprise-level SSD. This season, SSD has collected more than $5 billion, double the ring。

IN OTHER WORDS, WHO ATE AN AI DIVIDEND, NEVER JUST HBM. FROM HBM, TO GENERAL DRAM, TO DATA CENTRE SSD/NAND, THE ENTIRE REPOSITORY IS RISING TOGETHER. THIS FURTHER REINFORCES OUR JUDGEMENT THAT THIS ROUND IS AN INDUSTRY-WIDE STRUCTURAL BOOM, NOT A SINGLE PRODUCT FLOWER。

But as we invest, we cannot just listen to one word. The empty one, the logic is equally calm and sharp。

Their core argument is only one sentence: Māori is itself a cyclical reverse seed。

IT'S ALMOST CRUEL -- WHAT DO THEY DO WHEN MIZUKI, HERCULES, SAMSUNG DISCOVERS THAT HBM AND HIGH-END DRAM PROFITS ARE AMAZING? IT'S A MASSIVE EXPANSION, OF COURSE. AND THERE'S A TERRIBLE TIME LAG FOR THE SEMICONDUCTOR EXPANSION: TODAY IT'S DECIDED TO BUILD A PLANT, BUY EQUIPMENT, AND THE REAL RELEASE OF CAPACITY IS A YEAR OR TWO LATER. BY 2028, 2029, THE WAVE SURGED, THE SUPPLY SURGED, AND PRICES AND PADDY RATES COULD TURN DOWN - MANY OF THE NEW PLANTS HAPPEN TO BE PACKED IN THAT WINDOW AT THE SAME TIME, THE HIGHEST RISK OF EXCESS CAPACITY. THIS IS THE STANDARD SCRIPT FOR THE COLLAPSE OF EACH CYCLE IN THE PAST — NOT THE DISAPPEARANCE OF DEMAND, BUT OVERSUPPLY。

Thus, some investors pushed for a "turn in profits -- build-up -- overcapacity -- fall in prices." This logic is well established: the current high-māori is not sustainable and markets will be re-pricing at the "top of the cycle" sooner or later。

BUT IT'S INTERESTING THAT THIS EMPTY LOGIC ITSELF IS RELAXING. EVEN THE MOST CONSERVATIVE VOICES OF THE PAST, ONCE INSISTING ON PRICE-FIXING AT THE TOP OF THE CYCLE, HAVE RECENTLY BEGUN TO CHANGE COLLECTIVELY — THEIR LATEST MODELS NO LONGER BET ON "THE NEXT YEAR'S SURPLUS" BUT PREDICT THAT THE SUPPLY OF DRAM WILL CONTINUE INTO 2028. WHEN EVEN THE HARDEST EMPTYS BEGIN TO RAISE THE SUPPLY AND DEMAND GAP AND PUSH THE TURNING POINT, THIS IS IN ITSELF ONE OF THE MOST POWERFUL BYSTANDERS OF THE PARADIGM。

The differences between the two factions are essentially two narratives of the same set of facts. We'll put it in a comparison, so you can take care of it。

What do we think

We don't stand on the team, but we give a frame of judgement:This round is about "half supercycles" -- the high end is the paradigm, the standard is the cycle。

The HBM part, with technical barriers, long-term, focused AI clients, and, worse still, a physical formula for Token speed behind it, is closer to structural paradigm shifts, and the Māori rate is truly resilient. And the standard DRAM part, which is now more profitable than HBM because of its extreme tightness, is still gravitational and cyclical。

So what really needs to be looked at is not "prices can't be too high" but, when demand disturbances in the coming quarter, the Māori rate slows down gracefully or jumps off the cliff。

Prices lie, stocks lie, but Māori rates don't -- it's the only polygraph that's going to be super。

IF, IN THE NEXT FEW QUARTERS, THE MAURICITY OF U.S. LIGHT CAN STAND AT A HIGH LEVEL, EVEN IF DEMAND FLUCTUATES, IT'S JUST A MILD RETREAT, THEN THE PARADIGM WILL WIN. IF IT CRASHES, THEN I'M SORRY, IT'S ANOTHER OLD CYCLE IN AI'S NEW SUIT。

The second measure is the resilience of profits, depending on whether the supply is out of control。

And the biggest variable on the supply side, it's exactly what happened in this newspaper. The third ruler, we're going to look for answers outside the books。

Third ruler: long-stamped storage, the variable that didn't appear in the paper

The storage industry today has such a strong pricing power based on a word — oligarch。

THE GLOBAL DRAM MARKET IS LARGELY DIVIDED BY THREE GROUPS: SAMSUNG, HERCULES AND LIGHT. THREE PLAYERS, LIKE THREE PEOPLE WHO WORK TOGETHER TO OPEN A CASINO, CAN KEEP THEIR PROFITS HIGH AS LONG AS NO ONE LIFTS A TABLE AND DROPS PRICES. THIS "CONCRETE RESTRAINT" IS THE INVISIBLE FOUNDATION OF THIS ROUND OF HIGH MAORI。

But any oligarchy pattern, the worst fear is the same thingThe door came to the fourth player。

FOR THE STORAGE OF THE TROIKA, THE GREATEST RISK OF PRICING RIGHTS WAS NEVER ON THE DEMAND SIDE, BUT ON THE SUPPLY SIDE THE GROWING SPOILER. THIS MESSER, CXMT, FROM CHINA。

It does not appear in the financial statements of the beauty, but may be the variable that determines the maximum valuation of the beauty。

And this variable is going through a critical identity shift. In the past, we talked about "replacement of national production" -- can it be built, used, imported less. But now its story is moving from "pure substitution" to "sizing, profitable, high utilization". The three words are extremely heavy: size, which means that its capacity is no longer piecemeal; profit, which means that it no longer relies on subsidies and has the ability to make its own blood; and high utilization, which means that the lines it produces are running at full capacity and the goods are actually sold. A rival who earns his own money and continues to expand, and an opponent who survives on blood transfusions, is a total threat of two magnitudes。

MORE CRUCIAL IS CAPITAL. CHANGSUN IS ALREADY WORKING ON THE SPRINT IPO, PLANNING TO RAISE NEARLY 30 BILLION YUAN, AND THERE IS LITTLE HOPE THAT THE MONEY WILL GO — TO CONTINUE TO EXPAND. THE IPO IS NOT JUST A LUMP SUM; IT IS AN OPEN "REVALUATION": ONCE THE MARKET HAS BEEN SUCCESSFUL AND THE MARKET HAS GIVEN A BRIGHT MARKET VALUE, LONG-STANDING AMMUNITION IS AVAILABLE TO BUY EQUIPMENT, BUILD PRODUCTION LINES, DIG PEOPLE. THE FOURTH PLAYER IS NO LONGER JUST STANDING AT THE DOOR, AND IT'S ALREADY KNOCKING. ONCE THE DOOR IS PUSHED OPEN, THE HIGH-PROFIT TABLE, BACKED UP BY THE THREE CONCORDS, WILL BE REWRITTEN。

AND ITS REAL PROGRESS ON THE STANDARD DRAM IS ITSELF A TEST OF "DOMESTIC COMPETITION HAS MOVED TRUE". IT'S NO LONGER A VISION ON THE PPT, IT'S A GOOD RATE AND PRODUCTION THAT'S RUNNING OUT ON THE PRODUCTION LINE。

Let us first make clear the current pattern of supply, because there is a subtle chain reaction to price increases。

The three giants are now doing the same thing: leaning on HBM. Because HBM's profit is high, everyone wants another one. But the crystal circle is limited, and more HBM means less standard DRAM. So there's an anti-intuitive situation -- the fire in AI, the supply of ordinary memory is shrinking. Standard DRAM, with a shortage of goods, the price went up, so the price went up from the high end of "smuggling" to the whole class. This season's ASP increase is the most direct footnote - DRAM has increased its average selling price by a low-60s per cent (60% out) and NAND by a mid-80s per cent (about 85%). And that's the bottom mechanism for storing full-line price increases now, and we're judging that demand and supply stress will continue beyond 2027。

Now, put the Zen on this board。

THE LONG-TERM STRATEGY IS CLEAR: INSTEAD OF TOUCHING THE WORST HBM HEIGHTS, IT IS MOVING FROM STANDARD DRAM TO FAST-CALCULATION BASED ON COST AND LOCAL DEMAND. THIS CREATES A CRITICAL TIME GAP..

• ON THE STANDARD DRAM LINE, CHINESE MANUFACTURERS ARE CATCHING UP FASTER THAN MANY EXPECTED. ONCE THE STANDARD PRODUCTION CAPACITY OF LONG SORGHUM IS RELEASED ON A LARGE SCALE, IT IS POSSIBLE TO FILL THE SUPPLY GAP LEFT BY THE CONVERSION OF THE TROIKA TO HBM. THE LOGIC OF "AI CROWDING OUT STANDARD SUPPLIES AND DRIVING FULL-LINE PRICE INCREASES" IS WEAKENED。

• BUT ON THE HBM LINE, THE BARRIER IS ANOTHER SCALE. HBM INVOLVES ADVANCED SEALING, STACKING PROCESSES, DEEP BINDING AUTHENTICATION WITH AI CHIPS, AND IS NOT QUICK TO PASS THROUGH MONEY AND TIME ALONE. THIS HIGH-END MOAT IS STILL DEEP IN THE SHORT TERM。

SO THE THIRD MEASURE OF THE RULER IS A "SEPARATED" IMAGE: THE IMPACT OF THE LONG-SPRING ON THE STANDARD DRAM IS REAL AND NEAR, BUT IT'S A THREAT TO THE HBM HIGH END OF THE SPECTRUM. THIS COINCIDES WITH THE CONCLUSIONS OF OUR FIRST TWO RULERS — THE HIGH-END PARADIGM, THE STANDARD PRODUCT IS THE CYCLE, AND THE LONG-SPRUCE IS THE CATALYST THAT MAKES THE STANDARD CYCLICAL "EARLY RETURN"。

It also adds to the story of the "super-cycle" one of the most easily ignored puzzles: judging how far this round can go, not just by looking at beauty, Hercules’s financials, but also at the Chinese variable, which does not produce the dollar, but is quietly changing the supply equation。

Speaking of which, the meaning of the ruler has become even deeper。

TURNING BACK TO THE FIRST RULER -- WHY IS THE CLIENT SO ANXIOUS TO SIGN THE SCA, PAY THE ADVANCE, LOCK THE ORDER TO 2027? WE USED TO UNDERSTAND "LOCK PRICE, NUMBER OF LOCKS." BUT IN THE CONTEXT OF SUPPLY CHAIN SECURITY, IT ALSO HAS A THIRD DIMENSION: LOCK CAPACITY, LOCK SECURITY。

WHEN GEOPOLITICS MAKES THE SUPPLY CHAIN FRAGILE, WHEN STORAGE IS SEEN AS "STRATEGIC GOODS" OF THE AI ERA, WHAT A SUPER-CLIENT WANTS IS NOT JUST A GOOD PRICE, BUT A "WHATEVER HAPPENS, I HAVE THE CERTAINTY." LONGER TERM, WHICH IS ESSENTIALLY A PREMIUM PAID BY CUSTOMERS FOR SUPPLY CHAIN SECURITY。

What seems to be locked behind the long list of locked prices is capacity, security, and the most expensive thing in an age of uncertainty — certainty。

This layer is exactly what the old cycle framework could not capture at all. In the old framework, storage is for large commodities at low prices; in the new pattern, storage is for strategic resources, with kings of cargo. If this shift is true, then the rise in pricing rights is not a matter of one or two quarters。

But we also have to be honest — the logic of supply chain security is a double-edged sword. While raising the price-fixing power, it is also pushing China to accelerate its domestic replacement, and the growth of the long-term herds is itself a product of geo-anximity. In the short term, it locks up the customers, and in the long term it nurtures the one who lifts the table。

THREE FEET ARE DONE. THE LENGTH OF THE ORDER, THE RESILIENCE OF PROFITS, THE PATTERN OF SUPPLY - EACH OF WHICH POINTS TO TWO SIDES OF THE SAME CONCLUSION. NOW, WE GO BACK TO THAT NUMBER THAT INITIALLY SEDUCED YOU: 10 TIMES THE PE。

VALUATION TRAP: 10 TIMES PE SWEET AND HIGHLY TOXIC

We've gone around a big circle, and now we can answer the initial paradox in a positive way — why a bursting financial paper would be sold in exchange。

Throw a number that will cool your back。

Over the past eight quarters, eight financial papers have been delivered. The vast majority of the profits are higher than Wall Street's expected, and even a few are epic. But only two times — the remaining six — will be able to fall on the spot or turn back and throw back the increase。

Eight to two. A firm with a revenue of $6.8 billion went all the way to $23.8 billion with a high performance rate, with only two real increases after the financial report, with the remaining six falling either directly or rushing back. In other words, in the past two years, if you believe in the simple common sense of "good performance and stock price increases," you're probably slapped by the market every quarter。

On this stock, "good money" and "good stock price" are two things that are almost irrelevant。

If you don't believe me, we'll put back some of the most typical footage, and you'll see what a cold pattern there is。

ACT I, 2024, AT THE END OF THE YEAR. THE REVENUES WERE LARGELY IN LINE WITH EXPECTATIONS, AND THE PROFITS WERE A LITTLE TOO SMALL — THIS SEASON ALONE, NO PROBLEM. AS A RESULT, STOCK PRICES FELL BY 16 PER CENT THE NEXT DAY, THE LARGEST SINGLE-DAY DROP SINCE MARCH 2020. THE FUSE FOR SMASHING IS NOT "THE PAST" BUT "THE FUTURE": THE NEXT QUARTERLY GUIDANCE GIVEN BY MANAGEMENT IS FAR BELOW MARKET EXPECTATIONS AND REFERS TO CONSUMER ELECTRONICS BEING STORED. THE PERFORMANCE ITSELF IS FINE, AND THE MARKET IS AFRAID OF "THE NEXT SEASON."。

ACT II, MID-2025, UNTIL AUTUMN (FY25 Q3, Q4). THE RECORD RECORD, THE EPS, THE MĀORI RATE WENT FROM 39 PER CENT TO 46 PER CENT - A PERFECT RESPONSE BY ANY STANDARD. AND? ONE JUMP (ABOUT 8 PER CENT AT A TIME, ABOUT 5 PER CENT AT A TIME) AFTER THE STOCK EXCHANGE DID NOT HOLD: EITHER THROW BACK ALL THE INCREASES THE NEXT DAY, OR EVEN TURN GREEN (ONE OF THEM WENT DOWN ALMOST 3 PER CENT DIRECTLY), OR FALL BACK AND FORTH. THE REASON IS ONLY FOUR WORDS: IT'S GOOD. STOCK PRICES HAD RISEN AHEAD OF SCHEDULE, AND THE MOMENT THE MONEY WAS CASHED, IT HAD BECOME A PROFIT-MAKING GUN。

ACT III, AT THE END OF 2025. THE SAME YEAR, THERE WAS A SURGE OF 57 PER CENT IN REVENUE COLLECTION, HIGH EXPECTATIONS, NEXT SEASON'S DETONATING WATCH, AND EVEN CAPITAL EXPENDITURE INCREASED — THIS TIME, THE STOCK PRICE FINALLY ROSE, AND THE NEXT DAY IT ROSE BY ABOUT 10 PER CENT. IT'S A "GOOD MONEY AND GOOD STOCK PRICE" EIGHT TIMES。

THE FOURTH ACT, LAST QUARTER, WAS THE MOST EXTREME. IT'S ALMOST THREE TIMES THE SAME RATE, 75 PERCENT OF THE MAORI RATE, AND THE EPS DOUBLE-THIRTY-THIRTY-TWISTY -- THE DATA CAN'T BE SO BEAUTIFUL. AS A RESULT, THE FINANCIAL STATEMENTS DID NOT RISE OR FALL, AND WITHIN ABOUT A WEEK, THEY WITHDREW FROM THE HIGH POINTS TO CLOSE TO 30 PER CENT. BUT THE SECOND PART OF THE STORY IS REALLY MAGIC: IF YOU HADN'T BEEN THROWN OUT OF THAT PANIC, THE PRICE OF THE STOCK WOULD HAVE DOUBLED FROM THE RUINS IN THE NEXT THREE MONTHS, AND IT WOULD HAVE GONE UP。

To look at these four scenes — indeed, eight quarters — together, three patterns that really work for you have surfaced。

The first rule, the most counterintuitive and systematic:Good money, good stock prices, it's not accidental, it's normal。Six out of eight failed to pick up and fall the hardest, and financial data are often the most beautiful. The market never buys "how much it made last quarter," but "can we continue to expect next quarter?" Stock prices are always determined by the gap between performance and expectations, not by performance itself。

The second rule, hidden in front of the financial statements:The more pre-fiscal stock prices win, the harder they fall。it doesn't make any sense -- if, in the days before the financial report, america was ahead of the board (e.g., 3% or more), the good news had already been stolen, priced in, and the financial report had only been "delivered + profited" ; in turn, if the financial report had been lost, no one had looked at it, the expectations had been kept low, and then the money would have gone up. it's an almost conditional reflection, and it's worth putting out a pair in front of every financial report。

The third rule is most useful to shareholders:It's almost the fixed rhythm of the newspaper。The rate dropped in the short term after the financial statements, but increased to one month, and most rose back. The most exaggerating version is last season -- down 30 percent, double three months later. In many cases, the short-term collapse was not a fundamental collapse, but rather a fighting, panic and dishwashing。

It's not the basic face of the company, but your mind. See, you don't hand out chips in the darkest week。

IT'S WORTH MENTIONING THAT THE SIX DROPS AND DROPS ARE DIFFERENT: SOMETIMES THE NEXT SEASON IS LESS GUIDED THAN EXPECTED, SOMETIMES THE DECLINE IN THE MĀORI RATIO IS A HEART ATTACK, SOMETIMES THE NET INCREASE IN PROFIT ENDS, AND LAST SEASON IS A RUMOUR THAT OVERVALUATION PLUS AN "AI MEMORY DEMAND MAY NOT BE AS STRONG AS EXPECTED". THERE'S ONLY ONE KERNEL — THE MARKET IS PRICING "FOR THE FUTURE" INSTEAD OF APPLAUDING "FOR PAST PERFORMANCE."。

Does that mean that the financial figures are completely irrelevant? What really ignites the sale is often a signal that a certain business is interpreted by the market as "round the clock." One of the most classic of these is Capital Expenditure。

Let's explain why this is Ray. Capital expenditure is the amount of money the company plans to spend on building plants, buying equipment and increasing capacity. For a general growth unit, greater investment is good — it means that it looks to the future. But for a company that is priced as a periodic share, Capex is the most impressive alarm. Because the market's condition is that you expand your production, and after a year or two, you're overcapacity, and the price collapses, and the cycle goes up. In the language of the cycle, "I'm going to expand" is about "this is going to end."。

And so there's this ridiculous and reasonable scenario: the more profits are expected, the more tight the market is, the more the Capex is, the harder it is. Smart money is not selling "good old" but "excessive future expectations"。

THAT'S 10 TIMES THE PE'S DOUBLE PERSONALITY..It's a sweet discount to believe that it's a growing stock; it's a periodic stock, it's a poison bait。

So this time, we're gonna have to look at two chutes, which are much more important than the numbers。

First: capital expenditure. This time, management spoke of a clear upward revision - FQ3 net capital expenditure $7.1 billion, FQ4 is expected to jump to about $10 billion, and the year-round FY2026 about $27 billion, which is higher than previously anticipated by $25 billion; and, more alarmingly, management also predicted that FY2027 quarterly capital expenditure would be higher than FQ4 and that more than half of the increase would come from construction expenditure. It's a real increase. Capex is the sharpest alarm in the old playbook of the cycle, and even if the money is better, prepare the market for the "top of the cycle" logic. But in other words, most of the money went for the long contract. I don't know. If the increase is "disciplined priority for SCA needs" rather than brainless spot prices, then it's not as devastating as the paradigm narrative。Repressive and order-based expansion is evidence of oligopoly; the problem is that the first reaction in the market tends to be "over". FightI don't know。

Second: The first reaction to the stock price after the paper was published was itself a poll ticket - It tells you which ruler the market uses. But this article was written at the first moment of the financial disclosure, and the reaction was too late to digest. We can only say that, if history rhymes, the rate is again volatile。

ONE OF THE TECHNICAL DETAILS THAT MANY HAVE IGNORED IS WORTH DOING IN ADVANCE: BEFORE THE FINANCIAL STATEMENTS, THE FUTURE POWER RATE OF U.S.-LIGHT (IV) WENT UP TO 114 PER CENT, AND THE MARKET HAS A SINGLE-DAY VOLATILITY OF UP TO 14 PER CENT IN REAL MONEY AND SILVER. TRANSLATION — THE MARKET KNEW IT WAS A BIG BET, BUT NO ONE DETERMINED WHICH DIRECTION THE BULLET WENT. AND YOU CAN EVEN GO ONE STEP FURTHER, IN CONJUNCTION WITH THE THREE PREVIOUS PATTERNS: LET'S LOOK AT WHETHER MISAKI WON THE BIG GAME A FEW DAYS BEFORE THE FINANCIAL STATEMENTS, OR WHETHER HE IS BEHIND — THE MORE HE WINS, THE HARDER HE FALLS。

What's the concept? It means that, regardless of whether it rises or falls, it is destined to be a big one or a big one. The market isn't guessing "How good is the light?" Everyone knows it's good; the market is guessing "Well, it should be priced 30 times the growth stock, or 10 times the cycle stock." This valuation framework is the real source of volatility。

So remember: When the financial paper is released, don't just watch the game of numbers with the collection and the EPS, let alone panic over a big shadow after the financial report. The true winner was hidden in Capex's mouth, management's wording of the long bargain, and the vote on the stock price. And history tells us that the short-term cavity is often just the first half of the script。

TEN TIMES THE PE IS NOT THE ANSWER, IT'S A CHOICE -- IT FORCES YOU TO ANSWER WHAT KIND OF COMPANY YOU BELIEVE IN。

And to answer that question, we don't have to predict, we just have to wait for a few signals to be verified。

Backlight or pattern

Here, you might want us to give you a straight conclusion: buy it or not。

But we're not going to give you an answer. We're giving you a set of coordinates. Because the question of "returning light or paradigm" should not be based on guesses, but on a few verifiable signals。

We suggest you keep an eye on three。

First signal: 2027. degreesI don't know。

If management and the LTA can lock the order clearly into 2027 or even further, it means that the business has really changed from a " spot stand" to a "pack-a-year canteen" and the cycle has been re-established. The longer the visibility, the color of the paradigm. If visibility is still only one or two quarters, it's still the same old cycle of food watching。

Second signal: 80% of the Māori rateI don't know。

We're using the Māori rate as a litmus test of profitability. Note that the key is not how high a season can be, but when demand fluctuates. Hold it, it's a paradigm; it's a disturbance, it's a return light。

Third signal: Capital spending restraintI don't know。

This is the most anti-intuitive and the most important. If the triads collectively maintain the discipline of expansion — not making money without brain gain, but giving a regulated priority to service long-term — the oligarch's “conciliation” is still alive, and the profits are high. Once the first person who can't breathe, the Capex group races, the end of the round is written, just a matter of time。

Put these three signals together and the logic of judgment is very clear:

Three signals are in place at the same time — 2027 with clear visibility, a steady Māori rate, Capex with restraint — and that's a paradigm shift, 10 times the PE is a cheap product that was accidentally killed by the old frame. But as long as any of them collapses, it's a return light, 10 times the PE, the poison in the sugar coat。

AND THAT'S WHY WE'VE BEEN REFUSING TO PRICE THE LIGHTS FROM THE BEGINNING WITH ANY SINGLE LABEL OF THE "CYCLE UNIT" OR "GROWTH UNIT". IT IS NOW A STOCK THAT LIVES IN TWO VALUATION FRAMES -- HALF OF ITS BODY IS IN THE AI PARADIGM AND THE OTHER HALF IS IN THE WATER OF THE STORAGE CYCLE. THE SHARP SWINGS IN THE MARKET, AND THE SIX "GOOD MONEY AND BAD STOCK PRICES" OVER AND OVER AGAIN, ARE ONLY THE RESULT OF THESE TWO SETS OF FRAMES BEING PULLED OVER AND OVER。

SO WHEN THOSE NUMBERS ARE FILLED IN, DON'T RUSH TO CHEER FOR THE HARVEST, AND DON'T BE TEMPTED BY 10 TIMES THE PE, LET ALONE BY THE FIRST K-LINE AFTER THE FINANCIAL REPORT. BACK TO THE THREE SIGNALS, ONE AGAINST THE OTHER。

And when the tide rises, everyone is swimming; and it is only when it recedes that you know whether the sea level is raised forever。