Jimmy Young

[Jimmy Young @jmieyng]

#台語 若 koh khah 好， sī LTA kap 2029 到 2030 ê visibility 若開始出現，m̄-koh chit-má iáu-bē 算入去。 市場 sī m̄ 知影這件代誌， á-sī kā『 会 tòng gōa 久』算 m̄-tio̍h， á-sī kā #力積電 wafer 起価 ê 効果估計傷高。

[Jimmy Young @jmieyng]

晶豪科 ê 産品 lóng sī 旧世代 ê 記憶体： SLC/SPI NAND、 NOR、 DDR2/DDR3、 MCP/eMCP。 kui-ê 市場 lóng teh 起価： SLC NAND 2026 後半年約 120%到 170%； DDR2 2026 Q3 約 35%到 40%； DDR4 Q3 起 30%到 40%， DDR3 猶原短欠； 高密度 NOR 後半年起 60%到 65%以上； MCP/eMCP 估季増 15%到 25%。

[Serenity @aleabitoreddit]

Yep, ESMT (3006) has exposure to multiple legacy memory segments.

Has a portfolio of: SLC/SPI NAND, NOR Flash, DDR2/DDR3, MCP/eMCP, so all the fun legacy stuff.

If you want to look at broader industry price hikes coming up that ESMT has exposure to:

1. SLC NAND: ~120-170% for H2 2026.
2. DDR2: ~35-40% for Q3 2026.
3. DDR4: +30–40% Q3 (with DDR3 tightening)
4. NOR +60–65%+ for H2 (high density)
5. MCP/eMCP: maybe ~+15–25% QoQ for ESMT from my own estimate

For official breakdown:

DDR2 + DDR3: ~45% of revenue
DDR4/other DRAM ~ 10% of revenue
eMCP/MCP ≈30%
Analog + other ≈15%

Regardless, duration/hikes is the key:

-> "none of the major global SLC NAND suppliers plan to add new production capacity in the near term."
-> TrendForce specifically expects depleted inventories to make SLC NAND even tighter in Q4.
-> Winbond withdrawing DDR2, would make inventory even tighter (and they're doing allocations into 2029-2030)
-> ESMT management said the market's main problem was supply, not demand; the company's foundry allocations covered only roughly 60–70% of customer orders
-> Nanya: "memory shortages could last through the end of 2027, with supply and demand still tight in 2028"
->Nanya explicitly said capacity constraints could cause shortages across DDR5, LPDDR5, DDR4, LPDDR4 and DDR3.
-> ESMT included eMCP among the products receiving order-transfer benefits as mature node resources were being pulled elsewhere
______

If I had to state my thoughts again:

I think ESMT's ~1.9x July annualized run-rate P/E could compress further, despite factoring in PSMC's 45% wafer hike due to further legacy memory price hikes.

Industry commentary supports durability throughout 2027.

Cherry on top is if we start seeing LTAs and visibility into 2029-2030 (but not factored in).

I think markets either don't know about this, are mispricing duration, or overestimating PSMC's wafer hike effects.

Maybe playing duoQ with Rank 1 Challenger "ESMT" in LoL classic... Is more fun than playing with hardstuck "Micron" in League?

[Jimmy Young @jmieyng]

#台語 照 #晶豪科 7 月 ê 利益年率換算 ê P/E sī 1.9 倍， 就算 #力積電 wafer 起価 45%， 猶原 ē-sái koh khah 低， in-ūi 伝統 ê memory ê ASP koh ē 起。 LTA 到 2029 到 2030 iáu-bē 算入去這 ê 数字。 業界 ê 評論支持講， 這会 tòng 到 2027。

[Jimmy Young @jmieyng]

#台語 $SKHY CEO： 短欠会維持到 2030 年底（HBM4E ê Indiana 量産排 2029Q3）。 $SNDK ： NAND ê 需要会一直穏定到 2030 年， 非 GAAP 粗利率大約 80%。 $NVDA ê 供給承諾対 1190 億起到 2790 億美金， 大部分 sī memory； 定価 í-keng 「極端」 ， 猶 leh 起。

[Jimmy Young @jmieyng]

#華邦電 講 tio̍h 2029 到 2030 ê 割当（確保高雄 Module B ê 産能）。 chit-má ê visibility 会使看到幾若冬以後， ùi 旧世代 DRAM 一直到 HBM， 所以 kui-ê 産業 ê forward P/E iáu-ū 上升余地。 注意点： 愛看 sī tó 一層， 真多伝統 DRAM/NAND/NOR ê 公司 forward P/E 依然 kan-na 1.8 到 3 倍。

[Serenity @aleabitoreddit]

所以… 回到存储这边我们现在有：

$SKHY CEO 表示: “我们预计[存储]短缺将持续到 2030 年底”

$SNDK 表示: “我们看到直到 2030 年，NAND 都存在结构性的巨大需求”非 GAAP 毛利率将维持在约 80%.

$NVDA 表示 供应承诺从 1,190 亿美元增加到 2,790 亿美元，主要由存储采购推动…与此同时, 存储定价处于“极端”水平 而且还在继续上涨.

Winbond 正在与多家客户就延伸至 2029-2030 年的配额展开讨论.

我之前就说过, 存储需求看起来是结构性的, 而现在从传统 DRAM 到 HBM，需求可见度已经延伸到了未来数年..

所以我确实认为, 整个板块的远期 p/e 倍数都有进一步上升的机会.

[Jimmy Young @jmieyng]

#台語 #力積電 sī 台湾 ê 製造業者。 若 #晶豪科 無 koh 供給 #DDR2， 後手会受害真大； 若 #華邦電 退出， #晶豪科 真 ū 可能 sī 上大 ê 公開市場供給企業。 Wafer 割当減少， ASP 上升， 利益会維持足強。

[Jimmy Young @jmieyng]

#華邦電 退出 #DDR2 去做 #DDR4 / #DDR3 ， 2027 会 koh khah 緊， í-keng leh 講 2029 到 2030 koh khah 高階 ê 産品， 無理由 koh 倒転去做旧--ê。 認証時間長； 誘因真低。

[Serenity @aleabitoreddit]

Winbond is actually withdrawing from DDR2 to focus on DDR4/DDR3 and said they expect 2027 to be even tighter.

They're beginning capacity discussions for 2029-2030 + moving toward higher end, so don't quite see them having much to allocate toward older lines again.

As for other players moving toward older lines, qualification takes a lot of time... and there's just very low incentive to.

I looked for foundry dependence, PSMC is Taiwanese, so all good geopolitically.

As for wafer allocations, if I had to put my supply chain hat resiliency hat on, if ESMT stopped supplying DDR2, it would just be absolute destruction downstream, so don't see that scenario (since they're probably the world's largest merchant supplier for this line if Winbond exits)

So don't quite think it would happen... but if they get less wafer allocations, the ASP just gets hiked even more, so earnings would likely remain strong.

[Jimmy Young @jmieyng]

#台語 CW 会 sī 後一 ê $NVDA Nvidia 主導 ê 光学 chokepoint。 LTA lóng í-keng 簽好到 2030， 所以市場供給減少。 $SIVE kap #穩懋 sī 少数猶 ū CW 生産能力 ê 供給業者（koh ū CPO 等級 ê laser）。 m̄-koh 執行力 sī 問題。

[Jimmy Young @jmieyng]

市場 leh 看 $NVDA ê 時陣， Landmark í-keng kap 美国顧客簽 4 年 ê CW 契約， 為著確保「充分供給」 。 現況整理： $LITE kap $COHR ê 生産能力 í-keng 用了了， Landmark í-keng 簽約確定供給， $AAOI 主要用 tī transceiver， $MTSI iáu-bē 稼動， $SMTC ê 生産能力有限。 #台語

[Serenity @aleabitoreddit]

I know everyone’s watching $NVDA earnings right now…

But did you know Landmark signed a 4 year CW agreement with US customer to ensure “sufficient supply”?

So now you have:

- $LITE (capacity gone)
- $COHR (capacity gone)
- Landmark (capacity committed)
- $AAOI (capacity for transceivers)
- $MTSI (not online)
- $SMTC (limited)

I remember saying earlier this year CW lasers would be the next optical shift and heavily bottlenecked by Nvidia?

So fun watching this play out, with players signing LTAs already to 2030 (signaling less availability for other players).

Implications for $SIVE / Win Semi are very material as one of the few remaining CW merchant suppliers with capacity (+ CPO-grade lasers).

Let’s see how they execute.

[Jimmy Young @jmieyng]

#台語 進前 $NVDA 対 $NBIS Nebius 投資 20 億美金了後 ê 発表資料， 内底 ū 講会支持「khah 早採用 Nvidia 上新一代」 ê architecture。

[Jimmy Young @jmieyng]

接続進前 $20B 交易了後， $NVDA 宣布 Groq3 LPX í-keng leh 量産。 $NBIS （透過 Nebius token factory） sī 第一 ê 採用企業， Artificial Analysis benchmarks 性能評測 ê 成績真好： tī Gemma 4 31B 上 ū 3,400 tok/s， 100k context， 比上倚近 ê 対手緊約 4 倍。

[Serenity @aleabitoreddit]

$NVDA, following their $20B deal... Announces Groq3 LPX now in full production.

With $NBIS the first to adopt through Nebius token factory.

- 3,400 tok/s in Artificial Analysis benchmarking running Gemma 4 31B, with 100k token context
- 4x faster responsiveness than the nearest alternative

Thing that caught my attention.. was a former press release after Nvidia's $2B investment into Nebius.

That it would support their "early adoption of Nvidia's latest generation" architectures.

Guess we're seeing that special treatment here.

[Jimmy Young @jmieyng]

#台語 概算今年大約 2 万台 Spectrum-X 系統， 需要 60 万到 70 万粒 FAU； 2027 年需要会到幾百万粒， 既存生産能力目前無法満足。 量産開始維持 tī 2027 後半年。

[Jimmy Young @jmieyng]

FAU ū 可能 sī 台積電 kap NVDA ê CPO roadmap 未来 ê bottleneck。 媒体（UDN）評論講， 大立光、 波若威 kap 中国 ê 天孚通信 í-keng 加入 NVDA FAU ê supply chain。

[Serenity @aleabitoreddit]

I know everyone's focused on Nancy Pelosi copytraders sending $BE up 11.83%. Or Trump...

But you know what else is cool?

FAU bottlenecks for CPO.

I remember saying earlier this year FAU/Foci might be a future bottleneck around $TSM / $NVDA CPO roadmaps...

This was some media commentary supporting that:

"It is understood that Largan Precision (3008), FOCI (3363), and China’s TFC have entered $NVDA FAU supply chain"

"Based on an estimated 20,000 Spectrum-X systems shipped this year, the optical engines installed in those systems would require approximately 600k-700k FAUs"

In 2027, demand is expected to reach several million units, and existing supply chain capacity is currently unable to meet demand." (UDN, mandarin -> english translation is kinda off)

Cough cough, "existing supply chain capacity is currently unable to meet demand"

Tad early right now since mass production is expected H2 2027, but cool to read updates around this.

[Jimmy Young @jmieyng]

Lú-lâi lú 無法支持 $AAOI ， in-ūi i 連続推出 500M 到 600M 規模 ê ATM 股份発行。 到某一点， 営業展望 ū 可能好転， m̄-koh 股份結構 kap 融資愈来愈対股東不利。 真正無愛 ATM kap 不断 leh 資本募集， 就算 sī leh 拡大生産能力。 #台語

[Serenity @aleabitoreddit]

It's getting much harder to support $AAOI when they keep dropping $500m or $600M ATMs left and right.

At a certain point, the operating outlook could be positive.

But the share structure/financing becomes increasingly shareholder unfriendly.

Really, really, dislike ATMs and incessant capital raises, even if they're building up capacity.

[Jimmy Young @jmieyng]

#台語 $TSLA 時価総額 1 兆美金以上， m̄-koh Optimus 只 sī khah 大間公司 ê 一部分。 #宇樹 科技証明公開市場対 #humanoid 専門企業 ê 需要比逐家想--ê khah 大。

[Jimmy Young @jmieyng]

宇樹科技（688836） í-keng 上市， 起 492%， 時価総額大約 53.3 億美金。 現在 humanoid 総算 ū 一 ê 主要、 ū 上市 ê 専門企業做基準来参照。 做参考来講， Agility（ ū NVDA / $AMZN 出資， 透過 CCXI）估計第四季会用大約 2.5 億美金 ê 融資前評価額上市。

[Serenity @aleabitoreddit]

Unitree (688836) is now public.

And is valued at ~$53.3B, after rising +492.18%

We finally have a major public company benchmark for humanoids.

For reference, Agility (backed by $NVDA, $AMZN) via $CCXI is expected to go public at $2.5B premoney valuation Q4.

$TSLA is $1T+, but Optimus is wrapped inside a much larger company.

But I think Unitree showed that the demand for pure play humanoid players is much larger… than people expected in public markets.

[Jimmy Young @jmieyng]

@allen1392 @julian202188653 Indonesia 原住民選擇用韓國諺文來寫 in ê 語言，tú-hó mā ē-tàng 證明文字 ê 選擇 kap 語言無一定有關係。

https://t.co/BSB7nUvymY

[Jimmy Young @jmieyng]

@allen1392 @julian202188653 Anglo-Saxon 本來是用 Runes 字母，後來改換用拉丁字母。若是照你 án-ni 講，法語 kap 德語 ài 改轉去用拉丁字母寫，英文ài改轉去用 Runes 字母寫。

[Jimmy Young @jmieyng]

$MRVL 収買 Celestial 了後， koh kap $AMZN ū lēng-gōa ê warrant（為著 photonic fabric ê 購買）。 Jensen 進前講過， Marvell 会使変做下一 ê 1 兆美金以上 ê 公司； 看起来 Nvidia 早就知影， Marvell teh 透過 warrants 用 networking 事業入去 hyperscaler。

[Jimmy Young @jmieyng]

$MRVL hō͘ $GOOGL warrant， 上多 ōe-tàng 買到公司 12.2 億美金（大約 5900 万股， 一股 206.58 美金）ê 股票。 這 ê 交易包含 TPU 用 ê 広泛技術。 若達成購買目標， 到 2033 年可能会転換為大約 1200 億美金 ê 収入。

[Serenity @aleabitoreddit]

$MRVL gives $GOOGL options to buy $12.2B of the company.

"The Marvell-Google deal covers a broad range of technologies used ​with TPUs, including processors that run AI models, manage data storage ​and move information ⁠across networks."

Which could translate into ~$120 billion ​in revenue through 2033.

Marvell also has separate warrants with $AMZN after their Celestial acquisition (for purchases of photonic fabric).

Remember the good times when Jensen said Marvell would be the next $1T+ company?

Seems Nvidia knew Marvell was "networking" its way into all the hyperscalers with warrants.