Go Figure: A Field Guide to Market Oddities
The Markets’ Favourite Party Trick
Global investing is a theatre of mirrors. Expectations are rarely universal. The same facts, spoken clearly in one corner of the world, are refracted, distorted, and even inverted in another. It is not just about earnings or margins; it is about the mood of the audience. A set of numbers that in New York triggers champagne toasts can, in Seoul, lead to a price target cut. One market will frame a statistic as proof of secular growth; another will label it a “cycle top” and start drawing doomsday charts. Often, you don’t even need two markets or even two people working under the same banner for such acrobatics: the same expert can, in the same note or podcast, apply entirely different yardsticks to different stocks without ever seeing the irony.
We can only write about what is said or observed in the open — corporate disclosures, analyst reports, public statements — but the gaps are there for all to see: in how valuation multiples are justified, in which datapoints are amplified and which are ignored, and in the time horizons people pretend to care about. These divergences are not accidents. They are shaped by local investing cultures, liquidity realities, and historical scars that dictate how much good news one is allowed to believe.
For us at GenInnov, these are not intellectual curiosities. They are price tags. Big ones. Many are not “opinions” in the polite sense at all, but ingrained reflexes shaped by history, liquidity, and client needs. We hunt them precisely because they persist, and because they can reveal opportunity, risk, or an important dynamic hiding in plain sight. At GenInnov, we extract the insights left unsaid by asking why markets assume what they do. We like the companies where the mismatch between what’s happening and what’s believed is wide enough to drive a truck through, and we like to watch when the headlights are blinding but half the crowd still insists the road ahead is blocked.
Below is a short roll call of recent oddities observed surrounding some of the companies we track. They are a mish-mash of conflicting valuations, selective narratives, and analytical double standards. They are illustrations of what catches our attention when we are not wrestling with another esoteric innovation claim or subject.
A Catalogue of Oddities, in No Particular Order
WiWynn’s Vanishing Act
So we got Q2 2025 revenue and profit growth of 150%+ year-on-year. What was more surprising was how many saw a positive in the top line, also despite the monthly sales that had already been published weeks ago. That’s perhaps okay, as many were waiting for the company to release earnings details and guidance before revising their numbers again. However, their reports likely triggered more action from short-term, newsflow-driven investors and machines, providing a boost to the share based on information that wasn't entirely new.
Now comes the real twist: the very analysts forecasting relentless double-digit growth in global data-centre spending next year are modelling WiWynn’s revenue growth at barely above zero. So, the world will apparently keep buying racks of servers… just not from the company currently selling them the fastest. Or maybe, in this case, the stock will only move months after the facts are in, never in anticipation of them. Frankly, this is fine by us for a stock growing at three digits and with valuations that cannot be easily described as even mid-teens.
Micron the Hero, Hynix the Ghost
Micron upgrades next-quarter earnings guidance; shares pop. SK Hynix, with roughly 70% share in the HBM market (vs. Micron’s low-teens), unveils a projected 30% CAGR in HBM revenues to 2030 on the same day, and its stock market yawns. All AI infrastructure roadmaps depend on HBM scaling, and yet the biggest supplier is treated like a minor footnote about to be devoured by the minnows miles behind.
Triple-Digit Growth? Please Sit Down
Hynix deserves a closer look, if only for the sheer variety of oddities orbiting it. This is a company heading into its third consecutive year of blistering EPS growth (high double digits or even triple digits), a record that, in most NVIDIA-linked corners of the supply chain, would inspire unqualified praise. Yet a major brokerage, cheerfully bullish on pricing power everywhere else in the data-centre ecosystem, chose this moment to downgrade the stock on fears of “likely pricing pressure” next year.
Their report is hardly alone in its collapsing earnings growth forecasts; it sits comfortably among peers in projecting a WiWynn-style collapse in bottom-line growth from high double digits now to low single digits for 2026, on the back of competitive and oversupply concerns. The stated trigger for the downgrade? A “historic high” 1.8x price-to-book, delivered with the solemnity usually reserved for structural collapse, while the single-digit forward PE is conveniently left out of the justification. One wonders how many booming global technology stocks would survive that particular yardstick. Go Figure.
Rule of 1500/15/1.5
Let’s continue with Hynix, shall we? If Hynix is doing so well, how does that square with the “average” bonus of just 1.5 months’ salary announced recently? Around the same time, OpenAI handed its top 1,000 employees an average of USD 1.5 million each. The two cases, of course, are not meaningfully comparable, except as a reminder of how differently stakeholders are treated depending on geography and corporate tradition. To be fair to Hynix, in early 2025, it paid the equivalent of 1,500% of the monthly salary in performance bonuses. Its labour unions found the payout inadequate and we got more.
Always on the Wrong Side of The Table
Hynix must feel like it is born unlucky, tortured by imminent competitors, analysts, labour unions, customers, and if anyone can believe it, even its suppliers. Earlier in 2025, looking to reduce reliance on its sole TC bonder supplier, Hanmi, Hynix shifted some business to Hanwha. Hanmi’s reaction was almost theatrical: prices up, services cut, and in the end, Hynix forced back into a deal with its original partner. One can only imagine the satisfaction in Hynix’s boardroom if it could even fantasize about treating its own customers, say NVidia, the way Hanmi treated them the moment a rival, like Samsung Electronics, got a look-in.
Same Message, Different Ears
One more before we move from our Hynix oddity obsession: The way IR messages are delivered — and more importantly, received — across analyst communities is a wonder in itself. Micron’s investors hear steady, confident assertions that HBM prices will remain high, backed by “sensible” supply behaviour, and treat them as gospel. Its stock goes soaring when this is officially declared in front of analysts. Samsung says the same thing, albeit with a touch of defensiveness, and suddenly both its stock and Hynix’s are in the penalty box. It is almost like nobody believes them to be anything but predatory and irrational in trying to win back the crown it has lost to Hynix.
Samsung’s Spotlight, Analysts’ Blindfold
Samsung may have convinced Apple and Tesla, not exactly easy customers, that it’s worth partnering with Samsung to do cutting-edge work, yet its analyst community remains welded to near-term memory and foundry datapoints, dismissing any hint of a turnaround. Tesla’s eight-year AI6 chip supply deal? Apple’s joint chip development announcement? Interesting, but apparently not headline material. The real news, for those keeping score, is that Apple’s iPhone 18 will use Samsung for its camera image sensors — breaking Sony’s long-standing monopoly. The only open question is whether Sony will pull a Hanmi. (See supplier drama above).
The Monopoly with Margin Anxiety
TSMC’s grip on advanced packaging CoWoS is as close to a monopoly as the industry gets. In fact, it is the global standard. One can understand chatter about the phase-down of older CoWoS-S or about utilisation hovering at 60%. But talk of meaningful pricing pressure for a monopoly is stretching it. Only in some corners of Asia could this be a serious worry: a fresh way to fear for margins when none seems readily available. A strong contender for “oddest downgrade rationale of the year.”
Adobe’s AI Monetisation Masterclass
Adobe is facing intense competitive pressure in the AI era, with challengers biting at the entire application layer. Management’s move? Raise prices by almost 18% under the banner of AI monetisation. The analyst community applauds, calling it a clever offset to volume pressure. Some of those analysts might consider covering WiWynn or Hynix. Or, maybe we should have the IR chiefs and price-setting executives of Adobe or Micron provide coaching services to a few Asian corporates.
ASML and the Sacred Order Book
Few technology companies fetishise their order book quite like ASML. Its coverage analysts, usually deep-tech specialists, turn into industrial-sector veterans when discussing it, parsing every quarterly figure. In January, ASML conceded the lumpiness made the data misleading and vowed to stop publishing it. Their obsession with the orderbook must be like ours with Hynix oddities: they still publish it. Even when they finally do not, one suspects the “conservative” guidance will continue to be propped up by selective leaks of the very numbers they swore off. Perhaps it’s no surprise they’re rarely seen on conference stages designed for crowd-pleasing, high-octane keynotes.
Alibaba: Innovation? Let’s Not Dwell
A fact one can use: if you have to determine whether anything is written or edited by an AI, search for words like “dwell”, “tapestry” and most importantly, the Em Dashes (“ — ”).
Sorry, for that diversion. Back onto the oddities.
One has to read notes on Alibaba to witness a unique form of AI indifference. Innovation stories are acknowledged politely, then swiftly moved aside in favour of near-term business pressures as the sole stock driver. Apply this same editorial discipline to Apple or Alphabet and watch the analyst rankings implode.
Tariff Talk: Two Markets, Two Speeds
When a US president hints at new tariffs, American analysts tend to shrug and wait for detail, maybe adding a bland macro footnote. In Asia, the same hint produces overnight scenario tables, exhaustive sensitivity models, and portfolio reshuffles before breakfast. Whether tariffs discussed are on imports or exports, fear travels faster outside the US, the pessimists have bigger canvases, and short-term traders know exactly how to use both.
Long Runs, Short Patience
If you’ve spent decades covering markets that rarely move in long, steady climbs, it becomes hard to trust a true secular run. Outside markets like the US and India, investors feel they can benefit far more by disbelieving any pretender to be a secular winner than backing them for compounding. Take the case of Xiaomi: despite forecasting 15–20% multi-year top-line growth and trading at 25x forward earnings, when it launches the YU7 — one of the most impressive products in recent memory — the reaction is effectively a market shrug. The same type of launch from Tesla or Apple for a product with massive revenue potential would trigger an avalanche of upward revisions on the same announcement for months and quarters.
In Asia, downgrades can appear for reasons that would barely register elsewhere — like capacity utilisation being “too full” in the near term. The cycle anxiety is even harsher for companies without their own branded products. Quanta, for instance, is on track for 40% revenue growth this year and next, yet some argue it doesn’t merit 13x 2026 earnings. Here, the suspicion isn’t whether the numbers are real; it’s whether the story is “allowed” to keep running.
Why We Care
One can go on and on. We do not plan to make our write-ups a gossip column of such oddities. The serious point is different.
We are old enough to know that stubborn valuation ranges and ingrained behaviours will not change simply because a similar stock trades differently in another liquidity pool. Cross-market arbitrages survive decades, as evident from the persistent premiums and discounts for the same stock across listings.
What changes and creates genuine opportunity is how analysts and investors interpret the same business drivers. Whether it’s a 5x PE Hynix being downgraded in 2024 or an 8x PE Hynix cut in 2025 over “historic cycle fears” or to generate trades for a fast-moving client base, the label matters less than the assumptions underneath. On the balance of probabilities, the idea that Hynix would suddenly stop growing requires quite a chain: Samsung producing a flawless rival product for a handful of buyers almost instantly; every yield problem solved on day one; Samsung slashing prices to take share and behaving contrary to the behavior from any major players across tech subsectors globally; Hynix making zero technological progress; and the entire industry happily benefiting from two Korean giants undercutting each other.
Anything is possible, of course. The AI cycle could stall, and the analyst predicting Hynix at half today’s share price might be proven right. But for that to happen while the same houses keep their bullish forecasts on US semis feels less likely than this writer finishing a 5km run in thirty minutes next year. The same probability logic applies to WiWynn supposedly dropping to zero growth in a world where the very same models assume robust data-centre expansion for NVIDIA and the utilities that power it.
Yes, TSMC could, in theory, inflict its own pricing pressure on CoWoS despite a monopoly position and healthy global demand. Xiaomi could, despite one of the most successful product launches in years, limp along forever under supply constraints. These are all non-zero probabilities. But on the balance of probabilities, these are exactly the kinds of oddities global investors like us at GenInnov watch for and, when the set-up is right, act on.
