I Tracked Every Stock Call From China's #1 AI "Guru" — 167 Bullish Picks, 56 Bearish Calls, All Mapped

◆ Serenity Flash

The AI Oracle With the Most Paid Subscribers on X — 167 Bullish Calls, 56 Bearish, I Mapped Every Single One

If you've been following Serenity, scrolling through dozens of her tweets daily but still can't piece together which companies she actually likes most, this article is for you.

One night, I was staring at a tweet from the white-haired oracle. The more I looked, the more something felt off.

Here's what happened.

If you trade AI or follow semiconductors, you've probably come across a white-haired avatar called Serenity, @aleabitoreddit on X. Everyone in the circle calls her the White-Haired Oracle. She's a frontline analyst covering AI and semiconductor supply chains. On June 4 this year, her subscriber count overtook Elon Musk's — making her the #1 paid account on all of X.

Number one. More than Musk.

Here's how intense she is: dozens of tweets a day, from TSMC's CoWoS capacity to yield rates at some obscure Taiwanese optical module shop to when CPO will actually ramp. The density is absurd. I follow her daily. It's exhilarating. It's exhausting.

She posted one saying a certain company (I won't name names) was overvalued by the market. Something felt wrong immediately — I distinctly remembered her being bullish on this same company not long ago.

X is searchable, so I typed in the company name. That search only confused me more. Dozens of tweets popped up: bullish here, bearish there, stance shifting every few weeks. I went through them one by one, and by the end still couldn't tell — what's her actual position on this company right now?

That's when it hit me, something utterly absurd: I follow the most-subscribed analyst on the entire internet this closely, yet I have no real grasp of her judgments.

I know what she said today. I don't know what she actually thinks.

Those two are worlds apart.

After that night, I started wondering if this could be solved. I use AI daily to help others with information overload — getting spun around by one Twitter account felt embarrassing. Slow as I am, I spent over a month hacking away. And I actually cracked it.

It's called Serenity Flash. Today I'll walk you through how I turned following the White-Haired Oracle from suffering into something genuinely enjoyable, pitfall by pitfall.

Honestly, you've probably stumbled into these same traps.

First pitfall — and this is the killer — no big picture.

Think about it: someone posting dozens of times a day, you scrolling chronologically, forever seeing isolated dots. NVIDIA today, some Taiwanese shop you've never heard of tomorrow, robots the day after. One night of scrolling and your head's full of fragments that won't assemble into any coherent map. What board is she actually playing on? You have zero sense of it.

So the first thing Serenity Flash does is lay that board out for you.

Everything she's ever discussed, organized into three categories: companies, industries, concepts. 786 companies, 44 industries, 96 concepts, built on top of 6,200+ tweets — fully collected, not a single one missed.

Those numbers on the right are my personal favorite: bullish on 167, bearish on 56, neutral on 111.

Think about what this means. For the first time, an analyst's fuzzy mental long-short map gets quantified and laid out in front of you. One glance tells you she's broadly optimistic — bullish calls triple the bearish ones. You could scroll ten thousand tweets and never get this feel.

Back to my late-night failure.

My real problem wasn't finding that one tweet — it was wanting to know everything she'd ever said about a particular company. This is pitfall two: single points of information with no aggregation.

A company's judgment scattered across months and thousands of tweets — human brains can't assemble this.

How does Serenity Flash solve it? I turned every company name, every industry, every concept in her tweets into clickable tags. Green for companies, gold for industries, purple for concepts, with dotted underlines — see it, click it.

Click NVIDIA and see.

A panel slides out. One line at the top tells you what the company does. Below that, the key part: a Serenity opinion summary.

This summary isn't something I made up — I had DeepSeek chew through every tweet she's ever written about this company and spit this out. Overall judgment first, then dated bullet points: what she said June 15, what she said June 13, lined up clear as day. Below that, every original tweet mentioning this company, all here.

I'm telling you, when I first saw this panel auto-generate, I felt something like vertigo.

What used to cost me an entire night, scrolling until my hand cramped, still incomplete — now one click, all there.

And notice the top right corner: a green tag saying "Bullish."

Which brings us to pitfall three — the one that most affects your actual decisions: long-short confusion.

On the same company, she might be bullish this month and flip bearish next. If you only catch one of those tweets, it's easy to get skewed into thinking that's her stance. But it's really just one sentence at one moment in time.

So I gave every company, every industry, every concept a position tag: bullish, bearish, or neutral. This tag comes from DeepSeek reading through her complete views, not just one tweet.

No more reading between the lines yourself — whether she loves or hates this thing, one color tells you.

Speaking of which, there's an even sneakier pitfall: concepts.

The most headache-inducing thing about the White-Haired Oracle for normal people is her wall of jargon. CPO, HBM, CoWoS, Active LSI — three or four acronyms in a single sentence. If you don't know these terms, you can't even understand the tweet, let alone judge whether she's optimistic or pessimistic about that technology direction.

I was clueless myself at first, looking things up as I went.

So I made concepts clickable too. Take CPO.

Click in: first, one plain-English sentence telling you what CPO is (next-gen interconnect paradigm, sticking optical engines right next to chips). Then the same structure: opinion summary, position, all related tweets.

Technical terms stop being roadblocks and become entry points you can click, read, and see her stance on. This is the part I personally use most — like getting a free pocket semiconductor jargon translator.

Last pitfall, and maybe the most counterintuitive one.

She mentions tons of companies daily. The ones she mentions most often are where she's actually placing her bets.

But scrolling chronologically, you can't feel frequency. A tiny company mentioned once today looks identical in the feed to her beloved favorite mentioned eight hundred times over six months — just another flat line in the stream.

This drowns out the most important signal.

So I built a directory view, ranking companies, industries, and concepts by mention frequency. And helpfully grouped "bullish" separately from "neutral/bearish."

Look at this bullish leaderboard — the top spot isn't NVIDIA, it's a cloud company you probably haven't heard of, Nebius, which she's mentioned 729 times. NVIDIA at 478 has to stand back. Same for industries: semiconductor equipment and core materials at 1,513 mentions, high-speed interconnect at 1,246, leading the pack.

The ones at the top, mentioned most frequently — those are her real focus areas. This leaderboard essentially maps out an analyst's attention in plain sight.

She never said "this is my top pick," but her mention count has spoken for her.

Alright, five pitfalls covered.

Overview for the big picture, click company names for judgments, position tags for long-short direction, click concepts for jargon translation, leaderboard for core themes.

Oh, and a few bonus features I threw in. The whole site is bilingual side-by-side: Chinese translation on the left, English original on the right — judge the translation quality yourself. New tweets sync in about a minute or two, basically real-time. Every company and concept has its own URL — see something good, just drop the link in your group chat.

This month-plus of hacking was honestly exhausting: entity recognition, summary generation, position tagging — each step had its share of grunt work. But using it daily since? Genuinely satisfying.

I later figured something out.

Information asymmetry was never about not seeing the information. The White-Haired Oracle's tweets are visible to everyone, free. The real gap is: faced with the same pile of information, some people can organize it into a battle map, while others just scroll themselves into ever-deepening anxiety.

What us followers lack was never information — it's the step that turns information into judgment.

What Serenity Flash tries to do is transform that step from a grind requiring late nights, good memory, and domain expertise — into one click.

She outputs judgments. I make sure those judgments don't get washed away by the flood of information.

If you follow her too, if you've ever scrolled through her tweet waterfall until your hand went numb — this site might be built for you.

Link's below, works on desktop and mobile. On phone, too lazy to type the URL? Just hit "Read Original" at the bottom of this article, jumps right in.

https://aichainmap.com/serenity/

Go dig around. Find that company you've been watching, click in and see whether the White-Haired Oracle loves or hates it.

That feeling of what used to take all night now arriving in one second — try it once and you'll get it.

That's all. If you've made it this far and found this somewhat useful, a quick like, bookmark, share — and feel free to forward to that friend of yours who's also following the White-Haired Oracle until they're pulling their hair out.

Thanks for reading. Until next time.

Serenity Flash · Real-time Updates

Go find that company you've been watching

See whether the White-Haired Oracle loves or hates it — that feeling of what used to take all night, now one second away. Try it once and you'll get it.

aichainmap.com/serenity

Desktop & mobile ready · Full archive · Bilingual · Real-time sync