To The Moon

Muthu boy is sharing a little update on the AI stocks. I’m adding to my position in Montage Technology. Will also share my analysts’ insights with you, okay[Happy]  [Miser]  

Muthu boy is now chilling at Paradox Singapore Merchant Court 😌✨ since yesterday. 

On February 9, Montage Technology debuted in Hong Kong and surged more than 60% on its first trading day. Priced at the top end (HK$106.89), it opened sharply higher and closed strong amid extreme oversubscription — retail demand reportedly exceeded 700x. That wasn’t just enthusiasm. That was conviction capital chasing AI bandwidth exposure.

And that momentum didn’t fade.

On the final trading day before Lunar New Year, Hong Kong markets only opened for a half-day — yet the rally continued, up another 14%.

$MONTAGE TECH(06809)$  

 The Hang Seng Index closed at 26,723.0 (+0.59%), while the Hang Seng Tech Index edged up to 5,367.28 (+0.13%).

AI application names were on fire. MINMAX-WP jumped 24%. 

Zhipu gained 4%. The non-ferrous metals sector surged — CMOC rose over 6%, while Zijin Mining and Lingbao Gold climbed more than 4%. Semiconductors strengthened further, with GigaDevice rallying over 9%.

$GIGADEVICE(03986)$  

It was broad-based risk appetite — AI apps, chips, metals — all bid.

Meanwhile in the U.S., software was bleeding.

Fireworks on one side of the world. Flames on the other.

This wasn’t random noise. Capital was clearly rotating — away from vulnerable subscription models and toward compute, bandwidth, and AI monetization pipelines.

The trigger?

 The release of enterprise AI agents like Anthropic’s Claude Cowork plugins in late Jan/early Feb 2026. These aren’t cute productivity helpers. They’re autonomous agents plugging directly into core workflows — reading contracts, running compliance checks, drafting legal summaries, analyzing sales leads, mapping product roadmaps, reconciling finance. Eleven enterprise-grade integrations in one sweep.

The broader software sector (S&P 500 Software & Services Index, iShares Expanded Tech-Software ETF IGV) has been hammered, shedding massive value in early Feb 2026—estimates range from $830B to $1T+ wiped out since late Jan, with sharp single-day drops (e.g., indices down 4-8%+ in sessions).

IGV itself plunged hard: From highs around $117 in late 2025/early 2026, it traded down to lows near $80-83 in mid-Feb sessions (e.g., closing around $82-83 on Feb 13 after volatile swings, YTD down ~20-23%).

Individual hits: $Salesforce.com(CRM)$  down ~26%+ in 2026 so far (hitting 52-week lows near $187), ServiceNow down sharply (e.g., 11%+ drops in sessions, YTD heavy losses), Workday facing similar pressure amid CEO changes and fears. Other names like $Adobe(ADBE)$  , Intuit, SAP, Data dog, etc., all got crushed in the "SaaSpocalypse" wave.

Adding fuel: Late Jan 2026 saw a sharp drop (24%+ in one session) tied to Google's Project Genie rollout—an AI prototype for generating interactive virtual worlds from prompts. Markets freaked that generative AI could democratize game/world creation, reducing reliance on complex engines like Unity. This sparked sector-wide panic, dragging Unity, Roblox (and peers) to multi-month lows.

SaaS’s 20-year golden model was per-seat pricing — revenue tied to headcount. Hire more people, buy more licenses. But what happens if AI agents take over even 20% of entry-level white-collar tasks? Companies cut seats. Revenue drops. Growth narratives crack. Valuations compress — and not linearly. 40–50% drawdowns suddenly make sense in that framework.

Even Salesforce quietly acknowledged the risk in its 2025 annual report: 

if AI reduces customer workforce needs or seat demand, operating results could be materially affected. That’s not noise — that’s a structural warning.

Worse, new AI protocols allow agents to bypass traditional SaaS frontends entirely. Instead of logging into dashboards, AI can hit databases directly via APIs. The polished UI — once the moat — risks becoming irrelevant. If SaaS devolves into “just a database,” it stops deserving 40–50x multiples. Utilities don’t trade like luxury brands.

Meanwhile in China, capital is flowing toward where AI directly converts into revenue.

Enter ByteDance’s Seedance 2.0 (launched early Feb 2026). While most people focused on generation quality, the real breakthrough was cost compression. Upload product references, characters, camera instructions — lock in consistency for industrial-scale production. No more trial-and-error video shoots.

Reports suggest industrial video production costs can drop 60%+. For small Douyin sellers, API-generated high-quality promo clips can cost under RMB 5 ( SGD 91cts ). 

Compare that to traditional production — thousands of dollars and days of effort — versus seconds and pocket change now.

This isn’t just content tech. It’s a sales engine. Generate 100 ad variations, auto-test traffic, auto-optimize conversion — no editors, no media buyers. Closed-loop automation. From idea to monetization instantly.

So the divergence makes sense:

In the U.S., AI threatens SaaS margins by eliminating middle-layer labor.

In China, AI empowers apps that turn content directly into cash.

Chinese business culture doesn’t love recurring software subscriptions — it loves outcomes. If AI agents deliver compliance, accounting, sales — that’s tangible ROI. Capital is pricing in “digital employees” and automated revenue printers.

Zooming out, the value chain is reshaping into a dumbbell:

Left side: Infrastructure gods.

A joint study from Yale and a leading AI research group just dropped a bombshell concept they’re calling “Invisible Automation.” They estimate that right now, roughly $1.2 TRILLION in global wages is stuck in a Schrödinger’s-cat state. On paper, the jobs still exist.

HR is still posting openings.

Paychecks are still hitting bank accounts. But in reality? The actual work those people were doing has already been quietly dismantled by AI.Think about it:

A product manager making $500k a year?

Half of that value used to come from making PPTs, writing docs, pulling data, and sending emails.

Very soon, AI will handle all of that in seconds. Bosses haven’t fully woken up yet — they’re still “monitoring AI reliability” — but that $1.2 trillion premium is evaporating fast.The moment companies realize they can get the exact same output for $500 in API calls instead of $500k in salary…

that money doesn’t disappear. 

 It moves.Silently.

Invisibly.

From workers’ pockets straight into the pockets of the people who own the compute.

If you still want to make money in this market, you must understand the dumbbell.Because all the money is flowing to the two ends. Everything in the middle is getting crushed. 

Left end = The Shovel Sellers (the ultimate tax collectors)

No matter who wins — OpenAI, ByteDance, xAI, whoever — they all need compute. Chips. Memory bandwidth. That’s exactly why Montage Technology (6809.HK) exploded on debut and just ripped +14% today to close at HK$211 on pure momentum. To me, Montage isn’t just a “memory interface chip” company.

It’s the toll booth on the AI highway — same as Nvidia and TSMC. Every time someone generates a Seedance video, every time an AI agent reviews a contract, every time a data center spins up more Blackwell racks… you are quietly paying Montage (and its peers) a tax. They own the physical layer. Their monopoly is the hardest one to break. They collect war money.

Semiconductors, memory bandwidth, compute. 

Montage Technology, Micron, Sandisk, Seagate, Western Digital, Samsung, Hynixand Kioxia are rallying reflects one thing: every AI inference call needs bandwidth and Storage. Just like NVIDIA or TSMC, they collect the toll on every intelligence request.

Right side: Super apps.

Companies like ByteDance and Meta Platforms own traffic, models, and monetization loops. They convert displaced “human workflow value” directly into platform profit.

The middle?

Traditional SaaS — no compute moat, no traffic moat, UI-dependent — gets squeezed.

Everything now hinges on one catalyst: NVIDIA earnings on Feb 25, 2026.

The entire AI boom is built on one core assumption — relentless GPU demand. If guidance confirms sustained Blackwell momentum and aggressive capex from hyperscalers, the AI trade accelerates. SaaS weakens further. Infrastructure and super apps push higher.

But markets aren’t rational — they’re expectations games.

Even if NVIDIA beats earnings, that may not be enough. If results fail to exceed the highest analyst expectations, or if forward guidance shows even a hint of capex moderation, Wall Street could flip the script instantly. “Beat but not good enough” becomes the headline. And the sell-down begins.

In a positioning-heavy market, perception matters more than performance.

AI itself is ruthless. Jobs get automated. Legacy business models fracture. Entire layers of middle management and junior workflows compress. But capital doesn’t disappear.

The trillions being displaced don’t evaporate — they rotate.

From labor to compute.

From subscriptions to outcomes.

From human workflows to machine inference.

Disruption is not the debate anymore.

The real question is: who captures the redistributed value when the dust settles?

@TigerStars  @TigerObserver  @TigerPM  @TigerClub  @Tiger_comments