Gooaye EP696 Notes: The Customer Who Never Haggles, the iPhone Fold, and AI Infrastructure Meets Local Politics

A wine shop counter late at night, a warm lamp over a blank receipt, wine racks receding down a long aisle into cool blue shadow

A man without trustworthiness—I do not know how he can get on. A large cart without its crossbar pin, a small cart without its collar pin—how could either be made to go? — The Analects, “Wei Zheng” (Spring and Autumn period; my translation)

The pins in that line are the small pegs that join the shaft to the yoke. Without one, the ox can be as strong as you like and the cart still won’t move. Listening to Gooaye EP696 (2026-09-12), I kept coming back to that pin as the host moved from the price gap on a bottle of wine to stocks that get dumped the day after hitting limit-up.

What this episode covers

It opens light. He recommends Pepsi’s new “draft cola” and admits that in the past, if a friend served Pepsi with pizza, he’d walk to 7-Eleven for his own drink. Then he calls road cycling “the spinning top of middle age.” Three cycling friends trashed the Shimano 105, an entry-level pro groupset, and picked frames to meet the weight minimum of races they will never enter. He confesses he’s no different: at the batting cage he bought a pile of engraved aluminum bats, learned MLB players are only allowed wooden ones, and then ordered custom engraved wooden bats.

The main content has four parts: being overcharged by a wine dealer he trusted, which leads to “trust between people has disappeared”; Apple’s event and the foldable; what Oracle said around its earnings; and a view on closed AI models voiced at a Goldman Sachs conference. The Q&A has its human moments. His older son Noah begged for a friend to come over, got bored quickly, and asked Dad to send the friend home. When told that meant no more playdates, Noah answered “Yay.” There’s also his dog: its littermates were adopted to the US and Canada, and one shows up in photos skiing in full gear, which made him and his wife wonder whether they had held their dog back. His conclusion is that dogs don’t scheme against you, which ties back to the episode’s main thread.

Key points

1. The loyal customer who never haggles gets the worst price. He referred a friend to his wine dealer, and on the first purchase the friend got a lower price than he did. His orders start at seven figures, he never negotiates, and he pays the same day. When he asked about it, the salesperson said discounts require “accumulating a few more bottles.” He didn’t shame anyone online. He simply stopped doing business with them and told his family they would no longer let themselves be fleeced. He calls this a “refugee mindset,” and notes it shows up more among people who grew up comfortable, because they don’t treat other people’s money as real money and assume the customer won’t ask. The cost, as he puts it, is losing that customer plus a circle of friends with similar spending power.

Two side-by-side panels: on the left, a wine merchant is linked to one loyal customer, who is linked to five friends; on the right, the line between the merchant and the customer is cut and crossed out, the customer and all five friends turn into gray dashed outlines, and no one is left beside the merchant.

2. The stock market runs on the same thing. Stocks hit limit-up and fade the next day; companies use good revenue news as a chance to unload shares. Once participants learn this, everyone races to sell first and the environment turns into mutual scheming. He contrasts Nvidia and TSMC: plenty of good news, and the stocks kept rising. Companies that use every piece of good news to distribute shares end up “with no friends”; nobody wants to come back.

On one timeline with four pieces of good news, the blue line steps up a level after each one and trends higher overall, while the orange line pops briefly after each one and then falls even lower, trending down overall.

3. The iPhone Fold is priced below market expectations. Fully specced it runs well over NT$100,000, but the base price is only slightly above Samsung’s comparable foldable, and consensus had expected more. From that he infers sales will beat expectations and foldables could gain share of the overall phone market. To the “Android did it first” crowd he points to Nintendo: its consoles often use mature, cheap hardware and win on IP and experience. In tech, what counts is who does it best; nobody cares who did it first. Apple waits for technology to mature before entering, and the Vision Pro, which came too early, didn’t sell. The open question is China: luxury results show consumer downgrading there, and China is a key foldable market.

An S-shaped technology maturity curve: the gray dot at the bottom is Android going first, the orange dot midway up the slope is the Vision Pro entering too early, and the blue dot past the dashed "mature enough" line is the iPhone foldable.

4. Being Apple’s supplier is a thin business. The hinge is one of the hardest parts of a foldable, and he assumed first-generation suppliers would earn fat margins. Industry friends told him Apple squeezes hard. Related stocks rallied on the rumors and then went sideways. Apple will cultivate a second supplier against you; its in-house modem has moved from C1 to C2, and Qualcomm’s role will get squeezed out; some vendors now refuse Apple business altogether. On the other side, when memory prices rise, rivals lose money while Apple holds its margins and can use relatively affordable pricing to win new customers. His inference: everyday uses like asking for a recipe or about a bike derailleur don’t need frontier models, so large language models will commoditize at that layer, and value will flow to whoever owns distribution.

5. Oracle’s “no-money-down magic.” Customer prepayments and borrowing from customers; customers bringing their own hardware while Oracle charges hosting fees; plus vendor financing from Nvidia or equipment makers. Fine in good times, but if demand turns, demand inflated by one’s own financing will show as bloat. Another figure: rental prices for four-year-old GPUs actually rose 20%, echoing Jensen Huang’s point that GPUs are fungible, durable, and rentable, and weakening the bear case of rapid depreciation. He adds that if demand reverses, those prices can fall fast. Revenue grew about 30% year over year and remaining performance obligations (RPO) exceed $660 billion, yet in a choppy market good news doesn’t guarantee a rally.

Two bars compared: on the left, the data center's size in good times is built, from bottom to top, of one small self-funded block, three blocks of customer prepayments, three blocks of customer-supplied hardware, and three blocks of vendor financing; on the right, when demand reverses, the top nine blocks turn dashed and pull away, leaving only the small block at the bottom.

6. AI infrastructure has entered local politics. A 17-mile pipeline was rejected twice; one state demanded an A-grade credit rating or a $7 billion letter of credit; voters turned out in opposition, and Bernie Sanders and AOC have publicly opposed data centers. His read: a technology that makes only a few people rich will lose votes in a system where the majority wins, even if the opponents benefit from it. So big companies will turn to behind-the-meter power, building their own reciprocating engines, gas turbines, and solar. He mentioned rumors of Tesla placing large solar orders.

Two side-by-side panels: on the left, the line from the grid to the data center is blocked by an orange veto; on the right, the data center sits inside one dashed fence with gas turbines, reciprocating engines, and solar that supply it directly, while the grid outside has no connection.

7. Only one closed-model winner? He’s skeptical. At the conference, someone argued the closed-model winner would be an oligopoly, perhaps a single company, citing figures: $100 billion invested in closed models returns $130 billion, while the same in open-weight models returns $30 billion. The host thinks a duopoly trading the lead is more likely, and that each domain will have its own strong models; simpler uses will go to open models, and once hardware build-out costs flatten, value on that side will surface. He adds that if there really were only one winner, the collapse of even one other lab would wreck the whole market narrative, and the person making the claim would be the biggest victim.

Two groups of bars: closed models invest 1,000 and earn back 1,300, so the return bar stands taller than the investment; open-weight models also invest 1,000 but earn back only 300, a much shorter bar. A note marks this as a claim made at a conference.

Further thoughts

Why does a stock fall right after good news?

This is where I get stuck most. Good earnings, good headlines, and the next day it opens high and sells off, and I start wondering whether I misread it.

The host gave me an angle I hadn’t considered: it’s about whether the market has a memory. If a stock’s previous good news was used to unload shares, people learn that “good news means run,” and next time everyone tries to sell before everyone else. That kind of drop comes from expectations built up by participants and has little to do with how good this particular news is.

Three small price lines stacked top to bottom, all aligned on the same dashed "good news out" line; the first peak comes long after the news, the second comes sooner, and the third opens high and slides the moment the news is out, with an orange dashed line linking the three peaks and moving steadily left.

So how do I tell the difference? I turned it into two questions:

After this stock’s past rounds of good news, did the next quarter’s numbers follow through? Nvidia and TSMC were his examples because results kept arriving quarter after quarter, so the price had something underneath it.
Is this drop about the company, or is the whole market doing it? On Oracle he was clear: 30% revenue growth, over $660 billion in RPO, and still “good news doesn’t guarantee a rally,” because the market as a whole is choppy. That’s noise, unrelated to the company’s structure.

If both answers are “it followed through” and “it’s the whole market,” I treat it as noise. If the numbers stopped keeping up after earlier good news, I discount this round. It doesn’t guarantee I’m right. I’ve been shaken out by open-high-close-low days myself. But it means I don’t have to doubt myself every time.

The product is great, so why won’t the supplier’s stock move?

From this episode, foldables sell well and the hinge is a key part, so intuition says the hinge maker should soar. What the host described: a rally during the rumor phase, then sideways.

I checked it against the “bottleneck layer” framework I use, and here’s the catch: something being scarce or hard to make and being able to charge more for it are two separate things. However hard the part is, if there’s one buyer who keeps a second supplier and brings work in-house whenever it can, pricing power at that layer sits with Apple. He noted Apple and Tesla even make their own PCBs; same logic—brands keep the scarcest, most profitable pieces for themselves.

A two-axis chart with how hard the part is to make on the x-axis and pricing power on the y-axis; the key supplier sits at the far right of the x-axis, and while intuition places it in the top-right corner, it actually drops to the bottom-right, with a downward arrow labeled one buyer, funds a second source, and makes it in-house.

The second layer is capital. His analogy was a film release: however good your movie is, open against Christopher Nolan and your box office suffers. Money in the market is finite, and consumer electronics, however well it grows, may not draw capital away from AI themes that double in a year.

So next time I see “product X is a hit, suppliers benefit,” I’ll ask three more things: how many buyers are there? Can this supplier raise prices? Is something growing faster competing for the same money? After those three, a stock that doesn’t move makes sense.

Everyone’s calling a crash. Should I short a little to hedge?

A listener asked: if you moved into defensive stocks, aren’t you still fully long? When do you short?

I found his approach practical: how defensive he gets depends on how much he’s already made this year, recalculated every year and quarter. In a strong year he’s willing to hold longer and tolerate deeper pullbacks; in July he only took profits once prices hit the level he’d set in advance. If the year were up only 20% and the market kept falling, he’d turn defensive much faster. Shorts go on only after clear signals. In 2022 he waited until after Q1 and Q2 to add them, and the purpose was protecting his longs; profiting from a crash wasn’t the goal. He jokes that retail hedging usually adds risk: long on one side, short on the other, and whipsawed on both.

A line rising to the upper right, with this year's gain so far on the x-axis and how deep a pullback you can tolerate on the y-axis; the lower-left point at 20% is labeled defend early, the upper-right point at more is labeled hold longer, and the top-right corner notes that it is reset every year and every quarter.

In the same segment another listener, trading momentum on margin, saw year-to-date returns drop from over 300% to 15%. Along with his good wishes, the host added that a return curve like that means too much risk, and over time the odds of blowing up are high.

My own takeaway is that “hedging” easily becomes an excuse to guess the top every day. He puts it bluntly: people who claim they shorted the top have been squeezed countless times along the bull run. What I do now is write down in advance which signals would make me cut exposure, and do nothing until I see them. It won’t dodge every drop, but I don’t have to guess daily.

References

Gooaye EP696 (2026-09-12)
Apple’s fall 2026 event (iPhone Fold)
Oracle’s latest quarterly results and earnings call
Industry remarks at Goldman Sachs’ September 2026 technology conference

One thing to take with you

In any relationship, the person who never keeps score is the easiest one to overcharge; shortchange them once and you lose the whole circle around them.

One thing I tried: I listed three people who never bargain with me—a regular client, a colleague who always cleans up after me, and a family member who never complains. For each one I asked: are the terms I give them today better or worse than what I’d give someone I just met? If any answer was “worse,” I fixed it that week and wrote down how, next to the list.

Raw Markdown

This article is an educational discussion of investment method. It is not advice to buy or sell any individual security, offers no target prices, and does not analyze any current holding. Investing carries risk; make your own decisions or consult a qualified professional.