Deconstructing the Investment Methodology of the 'Stock God Serenity' in One Article

Serenity's bottleneck point investment method, in simple terms, involves first confirming a major trend of high certainty, then deconstructing the industry chain to find the hardest-to-replace upstream link, and finally placing a bet before the market fully prices it in.

For example, when the market hasn't yet realized that the optical interconnect upgrade in AI data centers will turn an upstream material, laser, or test equipment into a scarce asset, then this small link could undergo a valuation re-rating far exceeding its current fundamental revenue.

It's like a restaurant where the main course is the most expensive, but what truly holds back operations might be a certain niche seasoning; if this seasoning runs out, all the main courses can't be made.

Essentially, the bottleneck point methodology can be broken down into a five-factor model:

Demand must be certain enough, supply must be narrow enough, market perception must lag, potential value needs to be clear enough, and there must be verifiable event catalysts in the future.

When all five conditions are met, a small company is more likely to generate excess returns.

First Layer: Certain Demand.

AI data center expansion, cloud vendor ASICs, self-designed chips, inference demand, bandwidth demand—these constitute the macro-demand backdrop.

Serenity repeatedly mentions AMZN Trainium, MSFT Maia, Google TPU, NVDA driving 800V DC, etc., indicating he doesn't view small companies in isolation but places them within the context of giants' capital expenditures and architectural shifts.

For example, in his tweets regarding AAOI /LITE, he wrote the logic: the market rewarded the Google TPU supply chain but may have underestimated the optical interconnect demand from AMZN Trainium and $MSFT Maia.

Second Layer: Constrained Supply.

The so-called bottleneck point is not a trivial "this thing benefits too," but rather "it's indispensable" and "hard to replicate in the short term."

For example, InP substrates, CPO external light sources, CW DFB lasers, SOI wafers, optical transceiver test equipment, etc., all sound very niche, but once AI data centers shift from electrical to optical interconnect, these links become bottlenecks in terms of capacity, yield, certification cycles, and customer qualification.

Take InP substrate as an example. InP plays a critical role in high-speed optical communication lasers, detectors, and some photonic devices, offering significant advantages especially in scenarios requiring direct bandgap, luminescence efficiency, and high-speed modulation.

Meanwhile, constrained by long certification cycles, extended equipment lead times, high production process barriers, inability for capacity expansion to keep pace with surging demand, and structural shortages, it's difficult to replicate at scale in the short term.

Third Layer: Low Attention.

Low Attention = True Price Inefficiency.

Many of Serenity's picks are not at the center of mainstream narratives; places with "little institutional coverage, retail investors don't understand, media hasn't written deeply" are more likely to see mispricing.

Fourth Layer: Value Capture.

Pricing power, gross margin space, customer lock-in, supply share.

Turning a real bottleneck into excess returns still depends on several conditions: whether the company can secure capacity, set prices, avoid customer price pressure, avoid dilution from financing, whether gross margins can materialize, and whether demand has already been front-run by the stock price.

Fifth Layer: Catalyst.

Long-term potential is important, but short-term catalysts are also price engines.

Short-to-medium term triggers: earnings reports, customer mass production, Jabil fireside chats, CHIPS Act, index inclusion, Nasdaq dual-listing, M&A, short squeeze, capital flowing from local markets to US investors, etc., are all good clues and catalysts.

1.$AXTI: The most classic bottleneck point case.

Serenity was once banned on Reddit for analyzing AXT, Inc. (AXTI) early on. Why?

At that time, AXT was a small-cap, niche business focusing on InP substrates, seen as "pumping a small stock." But Serenity's understanding was that AI data center optical communication needed foundational materials like InP, and if supply was constrained, the entire photonics supply chain would be affected.

Subsequently, $AXTI rose nearly 10x from around $14, further proving the core competency: not looking at whether the stock price will rise first, but first judging whether this link would change from a "niche material" to a "strategic bottleneck."

2.$RPI: Small-cap companies are extremely sensitive to marginal demand.

The same demand change might be just a 1% revenue fluctuation for a large company, but for a small company, it could be a valuation system re-rating.

For example, increased demand for AI hardware, development boards, and edge devices has limited impact on a giant like Apple, but for a smaller hardware company like $RPI, it could directly alter the growth curve.

Serenity's bullish view on $RPI is: if AI agents require a large number of low-cost local nodes or edge orchestration hardware, then this "small computer" might suddenly become a type of infrastructure for AI application diffusion.

3.$AAOI /$LITE: Expanding from a single bottleneck point to a supply chain map.

Serenity placed LITE in the TPU / OCS beneficiary chain, placed AAOI in the MSFT Maia and AMZN Trainium ramp-related chain, and suggested InP might become a bottleneck in 2026, similar to HBM.

The bottleneck point isn't just about looking at a point, but thinking about that point within the line and plane: after the Google TPU chain is rewarded by the market, the next step might be the discovery of optical interconnect companies related to AMZN and MSFT's self-developed ASICs.

Copying tickers is easy; learning the thinking process and executing it is hard. To truly hold onto good picks, one must form their own knowledge system.

So how can we better apply Serenity's thinking process? There are six steps.

Step 1: Find the Major Trend: Has the demand been validated?

First, judge the trend well; don't look for stocks first.

For example, AI computing power expansion, CPO optical interconnect, 800V DC, humanoid robots, stablecoin payments, RWA tokenization—these are all trends.

If the trend itself is uncertain, subsequent supply chain analysis is nonsense.

Step 2: Draw the Map: What links are there from end-users to upstream?

Draw the industry chain.

Take CPO as an example. We can't just know $NVDA, but also need to know ASICs, switches, optical modules, external light sources, lasers, InP/SOI materials, packaging, testing, fiber arrays, microlenses, etc.

Serenity himself mentioned that if one cannot explain the optical communication industry chain from upstream InP substrate all the way to downstream optical modules, it means they haven't read enough.

Step 3: Find the Bottleneck: Which link is hardest to expand/produce/substitute?

Distinguish between a "real bottleneck" and a "pseudo bottleneck."

Real bottlenecks usually have several characteristics: concentrated supply, long qualification cycles, high customer switching costs, difficult technical yields, slow capacity expansion, and reliance on giants' roadmaps.

Pseudo bottlenecks are usually just "in the industry chain" but lack scarcity, can be done by anyone, and have weak pricing power.

Step 4: Find Evidence: Are there customer, qualification, capacity, order clues?

Use evidence, not emotion, to build conviction.

Evidence can include: customer clues in annual reports, management meeting notes, supplier qualification, CHIPS Act/government funding, index inclusion, patents, hiring, capacity expansion, partnership announcements, customer product roadmaps, competitor capex.

The highest level is company announcements, regulatory filings, earnings reports/conference calls; the middle level is customer websites, hiring, patents, supplier lists, government projects; the lowest level is peer mapping, AI inference, social media rumors. These three types of evidence must be separated, otherwise it's easy to mistake inference for fact.

Step 5: Do Risk Control: If you're wrong, where is the mistake?

Always create a "contrarian list."

Boldly hypothesize, carefully verify. It's not something you can just buy and forget.

If the customer doesn't ramp up volumes, when will revenue be disproven? If a competitor substitutes, does the bottleneck disappear? If the valuation has already front-run the earnings, can the stock price withstand an earnings gap? If over-communication leads to over-crowding, who takes the last baton? If the company raises capital, dilutes, or restates finances, does the bull case change?

Step 6: Match position size with research depth.

If you've only read others' summaries, the position should be very small; if you can draw the industry chain yourself, read annual reports, analyze customers, and create scenario valuations, the position can be larger.

While learning the methodology, we must also pour a bucket of cold water to stay sober. Because even the best method has limitations.

1. Inferences are prone to overfitting.

Serenity is very skilled at piecing together regulatory filings, partnership announcements, customer websites, and earnings report wording, but this method inherently carries the risk of misjudgment. A customer removing a supplier from their website, a company appearing on a blueprint, a partner having a relationship with a hyperscaler—these can all be strong clues, but they can also be just noise. It's necessary to clearly distinguish inference from fact.

Boldly hypothesize, carefully verify.

2. When early-stage financials aren't attractive, there's no valuation anchor.

For picks like SIVE, XFAB, AAOI, Serenity often looks at future 2027–2029 revenue ramp, architectural migration, and potential M&A, not current profits.

This approach has a high payoff if the direction is correct, but it's easy to misjudge if the direction is wrong.

3. Liquidity Reflexivity Risk: Serenity has become a market variable.

Serenity is no longer an ordinary researcher but a market participant with hundreds of thousands of followers, high subscription numbers, and media citations. Once he publicly expresses optimism about a small-cap stock, follower capital can directly push up the price, directly impacting the payoff.

4. Looking dialectically, there is also a certain degree of survivorship bias.

The 4500%+ return, while the logic is worth learning from, is also largely due to catching the major AI computing bull market.

Serenity is indeed impressive, but we must also remain cautious.

Past experience may not apply in the future; will giants find ways to bypass current bottleneck points later?

Furthermore, Serenity's success, besides powerful analytical skills, requires a constant stream of first-hand information sources and a strong stomach to withstand drawdowns—all three are indispensable.

As the saying goes, boldly hypothesize, carefully verify. Be responsible for your own positions.

 

That said, the bottleneck point investment method works because the market often prices the big narrative first, then the tier-2 suppliers, and finally realizes the truly scarce materials, components, testing, and capacity links.

But the most dangerous aspect of this method is here: it highly relies on professional judgment, information piecing together, tolerance for non-consensus, and position sizing discipline.

What we should truly replicate is not Serenity's holdings, but his research sequence: first find the certain trend, then find the bottleneck, then find the evidence, then look at valuation, then wait for catalysts, and finally place bets with capital you can afford to lose.

In the end, after seriously studying Serenity's methodology, only three words remain in mind: Enter through the narrow gate.

In major trends like AI, don't buy the most obvious hot stocks, but drill down the industry chain to find the hardest-to-replace bottleneck points in future architectural migrations, and place early bets while old financials, old valuations, and old regional biases still suppress the price.

This is the narrow gate of investment, and it can also be the narrow gate of life.