chen

3.3K posts

chen

chen

@nikaidoula

Katılım Haziran 2014
1.9K Takip Edilen103 Takipçiler
chen retweetledi
chen
chen@nikaidoula·
@IamczcChen 伊斯蘭就這樣,還想影響別人,被打正常
中文
0
0
11
2K
村人C
村人C@IamczcChen·
埃及人去跟中國人坐一桌,謝謝。
村人C tweet media
中文
17
107
1.5K
53K
chen
chen@nikaidoula·
@dabowagaga 產與股市,然后看電視亂買,還能有百萬嗎😂
中文
0
0
0
180
風暴士兵 🪐🪬Taiwan only
2025 年台灣平均每戶家庭淨值創新高達 1,889 萬元,家庭中位數為950萬,而個人財富中位數為350萬元(台幣) 我國個人財富中位數高於南韓及新加坡(表示貧富差距沒有這兩國高)。 另,台灣股市參與人數為1410萬人,家戶資產股票佔比為34%,存款佔比35%,我國國民有股票超越存款的趨勢。
中文
31
7
132
6.1K
chen
chen@nikaidoula·
@dabowagaga 親中國先被加關稅 然后去中化轉單 在被中國內部卷價格,三件套他覺得還能四萬點🤣
中文
0
0
2
195
chen retweetledi
hsn.bot
hsn.bot@hsn8086xe·
hsn.bot tweet media
ZXX
45
71
3.5K
518.5K
大狗狗先生酒酒
大狗狗先生酒酒@DoppelDog·
家樂福被統一買了 LOPIA也被統一買了 家樂福的法國讚讚商品不見了,會不會Lopia 以後也不會是Low Price Utopia了😭😭😭
大狗狗先生酒酒 tweet media
中文
36
90
919
77K
chen retweetledi
周览资源
周览资源@grgerwcwetwet·
周览资源 tweet media
ZXX
24
297
2.3K
179.3K
chen retweetledi
駿HaYaO
駿HaYaO@QQ_Timmy·
$VIAV VIAVI SOLUTIONS Q3 FY26 財報電話會議 整體核心 - 最強正面:AI 資料中心需求已從「實驗室驗證」全面進入生產測試、現場部署、光學模組製造、先進封裝階段,動能預計延續至 2026 年全年。 - 最強負面:無線 RAN / 5G 測試需求依然疲弱,短期無復甦跡象。 - 受益趨勢:AI 網路晶片、光學模組、高速測試、先進封裝、記憶體、有線 DAA、國防自主。 - 受壓趨勢:無線設備、無線測試廠商、中小型光學模組廠。 五大重點 1. AI 資料中心測試已進入生產與現場階段 - NSE 部門年增 54.4%,超預期。 - 資料中心需求占現場光纖監控儀器已達 40-45%,很快逼近 50%。 - 最受益:Keysight(高)、Corning、Fabrinet、光纖廠(Prysmian、Sumitomo、Furukawa)。 2. 高速乙太網路強勁放量 - 800G 仍為高量主力 → 1.6T 實驗室爬坡 + 早期生產 → 3.2T 設計中。 最受益: Broadcom(高)/ Marvell(高)/ Arista(高) Credo、Astera Labs(中高) 3. 光學模組與矽光子需求強,但測試強度導致贏家通吃 - 800G 放量、1.6T 起飛,CPO 動能明顯。 - 複雜模組「一顆不良整顆報廢」,推升測試插入點與良率重要性。 - 最受益:Coherent(高)、Fabrinet(高)、Zhongji Innolight、Eoptolink。 - 相對受壓:規模小、自動化低的模組廠。 4. 共封裝光學(CPO)與先進封裝成為策略瓶頸 - 測試與封裝從「附屬」變成「核心策略資產」。 - 異質整合、玻璃基板、晶圓級封裝需求大增。 - 最受益:Advantest、Teradyne、FormFactor、Onto、Amkor、ASE、BESI 等先進封裝與測試廠商。 5. VIAVI + Spirent 整合提升競爭強度 - 3.2T 整合平台即將推出,對 Keysight、Anritsu 構成中幅度競爭壓力。 - 整體高速測試市場需求仍強,但市占競爭加劇。
TheValueist@TheValueist

$VIAV KEY READ-THROUGHS FROM VIAVI SOLUTIONS Q3 FY26 EARNINGS CALL VIAVI’s Q3 FY26 call was a high-signal upstream datapoint for AI infrastructure, high-speed optical/electrical interconnect, co-packaged optics, semiconductor test intensity, memory scarcity, cable access, and defense autonomy. The most important broader-market implication is that AI data center demand is no longer confined to accelerator procurement or early lab validation; it is now propagating into production test, field fiber monitoring, optical module manufacturing, semiconductor packaging, high-speed Ethernet validation, and deployment assurance. The call also provided a clear negative read-through for wireless RAN and 5G test recovery, with management explicitly stating that wireless demand remains weak and no near-term recovery is expected. The highest-conviction positive cross-sector signals are for AI networking silicon, optical modules, high-speed test and measurement, advanced packaging/test, memory suppliers, cable DAA vendors, and defense PNT/autonomy suppliers. The highest-conviction negative or mixed signals are for wireless RAN equipment, wireless test-exposed names, hyperscaler free cash flow intensity, and smaller optical/module manufacturers facing rising test complexity and yield risk. AI DATA CENTER TEST DEMAND HAS MOVED FROM LAB VALIDATION INTO PRODUCTION AND FIELD DEPLOYMENT (READ-THROUGH 1) Call evidence: VIAVI reported NSE revenue of $321.5M, above the $304M-$316M guide, up 54.4% year-over-year. Management attributed the growth to “strong demand for our lab and production and field products driven by the data center ecosystem.” Oleg Khaykin stated that “we are seeing strong demand across all data center segments, scale up, scale out, and scale across,” and added that “given strong and growing customer demand, we expect the data center ecosystem revenue momentum to continue through the calendar 2026.” The most differentiated comment was on field instruments: “I’ve never seen so much demand for our fiber monitoring solutions,” and data centers are “buying more equipment than regular service providers.” Management estimated that data center demand is approaching “40%, 45%, pretty soon, probably maybe 50%” of field instruments. Affected companies and impact: Keysight Technologies (KEYS: US): Positive, high magnitude. The transmission mechanism is higher demand for optical transport, protocol, Ethernet, PCIe, and high-speed network validation equipment across AI labs, production environments, and deployment monitoring. VIAVI’s commentary confirms that test demand is broadening structurally, which should support the broader high-speed test category even if VIAVI gains share in specific accounts. Anritsu (6754: Japan): Positive, medium magnitude. Anritsu should benefit from the same high-speed optical and network test cycle, but the positive data center read-through is partially offset by VIAVI’s explicit weakness in wireless test, where Anritsu has meaningful exposure. Corning (GLW: US): Positive, medium magnitude. The mechanism is stronger AI data center fiber deployment and monitoring intensity. VIAVI’s comment that hyperscale data centers are buying fiber monitoring equipment at levels comparable to or above service providers implies elevated demand for high-density fiber infrastructure, quality assurance, and optical network reliability. Prysmian (PRY: Italy), Sumitomo Electric Industries (5802: Japan), and Furukawa Electric (5801: Japan): Positive, medium magnitude. The mechanism is increased fiber intensity in AI data center campuses and interconnect architectures, with more monitoring, more deployment verification, and more fiber-related test insertion as bandwidth scales. Near-term trading catalyst: The Q4 guide implies continued sequential growth in NSE, and management explicitly guided data center momentum through calendar 2026. Longer-duration fundamental shift: Field instrumentation is being structurally repositioned away from a service-provider-only demand pool toward a hyperscaler and AI data center operating model. This is non-consensus relative to the historical view of field test as a telecom capex proxy. HIGH-SPEED ETHERNET, 800G VOLUME, 1.6T RAMP, AND 3.2T DESIGN ACTIVITY SUPPORT AI NETWORKING SILICON AND SWITCHING WINNERS (READ-THROUGH 2) Call evidence: Management stated that 800G remains “very much a high volume driver,” while 1.6T is “ramping in a lab” and entering “early production products to the module vendors.” VIAVI released 1.6T products approximately 1.5 years ago, and adoption has expanded from network equipment manufacturers to semiconductor vendors and co-packaged optics customers. On Spirent, management said the “first truly integrated product” between VIAVI and Spirent will be at “3.2 terabits.” VIAVI also said customers developing “AI chips for inference or the training” are buying optical transport and protocol test solutions. Affected companies and impact: Broadcom (AVGO: US): Positive, high magnitude. The mechanism is direct validation of sustained demand for AI networking silicon, switching ASICs, custom silicon, high-speed SerDes, and optical connectivity roadmaps. A test vendor seeing 800G volume, 1.6T lab ramp, and 3.2T development is an upstream confirmation that networking semiconductor cycles remain active. Marvell Technology (MRVL: US): Positive, high magnitude. The mechanism is accelerating demand for custom silicon, optical DSPs, high-speed connectivity, and data center interconnect components tied to AI fabrics. VIAVI’s commentary around chip-to-chip interconnect and co-packaged optics supports a higher-growth connectivity content cycle. Credo Technology Group (CRDO: US): Positive, medium-high magnitude. The mechanism is rising demand for high-speed connectivity, active electrical cables, SerDes-heavy AI fabric infrastructure, and Ethernet scaling. The call reinforces that AI networks are speed- and signal-integrity-constrained, which is favorable for high-speed connectivity specialists. Astera Labs (ALAB: US): Positive, medium magnitude. The mechanism is broader PCIe and high-speed connectivity complexity around AI systems. VIAVI specifically cited “PCIe Express test products” in the context of AI chip and system development, supporting the view that connectivity validation remains a critical bottleneck. Arista Networks (ANET: US): Positive, high magnitude. The mechanism is sustained AI Ethernet fabric expansion. VIAVI’s references to high-speed Ethernet, optical switches, NEM demand, and 800G/1.6T ramps are positive indicators for AI cluster switching demand and next-generation data center switching roadmaps. Cisco Systems (CSCO: US) and Hewlett Packard Enterprise (HPE: US): Positive, medium magnitude. The mechanism is incremental support for AI data center switching and high-speed networking opportunities, though each company’s broader enterprise and service-provider mix dilutes the magnitude relative to Arista. Near-term trading catalyst: 800G remains in volume, and 1.6T is already ramping in lab and early production. Longer-duration fundamental shift: The test cycle points to a multi-year AI networking roadmap extending from 800G to 1.6T and 3.2T, with increasingly complex electrical, optical, and protocol validation requirements. OPTICAL MODULE AND SILICON PHOTONICS DEMAND IS STRONG, BUT TEST INTENSITY CREATES A WINNER-TAKE-MORE MANUFACTURING DYNAMIC (READ-THROUGH 3) Call evidence: VIAVI said the data center ecosystem includes “high-performance semis, optical modules, NEMs, and hyperscalers.” Management described “strong demand” for optical transport, silicon photonics, communication protocol, and high-speed Ethernet test equipment. On production, Khaykin said VIAVI is seeing “a lot of momentum on this whole co-packaged optic area” and is selling equipment to semiconductor vendors developing “integrated packaged optic solution.” In Q&A, management added that every component in a complex optical module must be tested before final integration because “if you have one device is bad, you throw away the entire” sealed module. Affected companies and impact: Coherent (COHR: US): Positive, high magnitude, with medium execution risk. The mechanism is stronger demand for optical modules, silicon photonics, and high-speed optical components tied to AI data center scale-out. The risk is that rising test intensity and yield sensitivity increase manufacturing complexity and working-capital demands. Lumentum (LITE: US): Positive, medium-high magnitude. The mechanism is stronger AI optical demand and silicon photonics-related validation requirements. Magnitude is somewhat lower than Coherent and Fabrinet because the read-through depends on mix exposure to the highest-growth AI optical products. Fabrinet (FN: US): Positive, high magnitude. The mechanism is outsourcing demand for complex optical manufacturing, testing, and production execution. VIAVI’s commentary that optical components must be tested at multiple insertion points favors scaled contract manufacturers with strong process control and test infrastructure. Zhongji Innolight (300308: China): Positive, high magnitude. The mechanism is continued 800G volume and emerging 1.6T demand from hyperscalers and module customers. The call’s optical module commentary reinforces demand durability, although export controls and customer concentration remain separate risks not directly addressed in the call. Eoptolink Technology (300502: China): Positive, medium-high magnitude. The mechanism is similar to Zhongji Innolight, with AI data center optical module demand and higher-speed transition support. The magnitude is high for demand but moderated by competitive and geopolitical risk. Negative/mixed implication: Smaller or less automated optical module manufacturers face negative medium impact from the same trend. The mechanism is higher test capex, more insertion points, tighter yield requirements, and greater risk that defective components destroy value late in the production flow. CPO and silicon photonics reward scale, process control, and capital availability. Near-term trading catalyst: 800G remains a high-volume driver, while 1.6T is entering early production products for module vendors. Longer-duration fundamental shift: Optical module competition increasingly shifts from simple capacity and pricing toward yield management, test infrastructure, photonic integration, and manufacturing execution. CO-PACKAGED OPTICS AND ADVANCED PACKAGING ARE TURNING TEST AND PACKAGING INTO STRATEGIC BOTTLENECKS (READ-THROUGH 4) Call evidence: When asked whether testing is becoming a bottleneck in co-packaged optics, Khaykin responded, “You’re absolutely right.” He stated that “the whole test packaging used to be kind of a backhand afterthought. It is now the system. It’s now strategic assets.” He described glass substrates, photonic integrated circuits next to electronic integrated circuits, wafer-level packaging, heterogeneous integration packaging, and custom rack-mounted test systems. He also said VIAVI’s technologies are being pulled into the value chain from “individual optical components to wafer level packaging, to the heterogeneous integration packaging, all the way down to being integrated into major test platforms.” Affected companies and impact: Advantest (6857: Japan): Positive, medium-high magnitude. The mechanism is rising semiconductor test complexity as AI devices incorporate faster electrical interfaces, optical-adjacent integration, and more advanced packaging. Although VIAVI’s specific exposure is optical and communications test, the broader implication is more test value per AI package. Teradyne (TER: US): Positive, medium magnitude. The mechanism is greater electrical test complexity and higher-value package-level validation in AI and high-speed semiconductor devices. The read-through is positive, though less direct than for optical/network test vendors. FormFactor (FORM: US): Positive, medium magnitude. The mechanism is increased wafer-level and package-level probing requirements as photonic and electronic components converge in heterogeneous integration flows. Onto Innovation (ONTO: US) and Camtek (CAMT: Israel): Positive, medium magnitude. The mechanism is more inspection, metrology, and process-control demand in advanced packaging and heterogeneous integration environments. Amkor Technology (AMKR: US), ASE Technology (3711: Taiwan), and BE Semiconductor Industries (BESI: Netherlands): Positive, medium magnitude. The mechanism is higher advanced packaging intensity as CPO, photonic integration, glass substrates, and heterogeneous architectures move closer to production. Packaging capability becomes a strategic differentiator rather than a back-end commodity process. Near-term trading catalyst: Management said CPO/OCS-related sales are early and “not in the current numbers,” but could start ramping in “2 to 3 quarters.” Longer-duration fundamental shift: The call strongly supports the thesis that AI hardware value creation is shifting toward advanced packaging, photonic integration, and test/inspection intensity, not only front-end wafer fabrication. VIAVI-SPIRENT COMBINATION RAISES COMPETITIVE RISK IN HIGH-SPEED ETHERNET TEST DESPITE POSITIVE CATEGORY DEMAND (READ-THROUGH 5) Call evidence: Spirent revenue was $54.2M in Q3, in line with expectations and helped by orders pushed out from the prior quarter. Management expects roughly a $200M annual run-rate, with Q4 likely “just shy of the $50M, maybe $48M.” Strategically, Khaykin stated that Spirent brings “a pretty big established customer base” and “application-hardened software for all kinds of Ethernet traffic,” while VIAVI is “upgrading the hardware performance of their products.” He added that the 3.2T product will be the first truly integrated VIAVI-Spirent platform. Affected companies and impact: Keysight Technologies (KEYS: US): Mixed, medium magnitude. The category demand read-through is positive, but the competitive read-through is negative. The mechanism is that VIAVI now combines high-performance optical/network hardware with Spirent’s Ethernet traffic software and installed base, potentially increasing share pressure in high-speed Ethernet and protocol test. Anritsu (6754: Japan): Mixed to negative, medium magnitude. The mechanism is the same competitive pressure in high-speed network test, compounded by VIAVI’s separate comment that wireless test demand remains weak. VIAVI customers in semis, NEMs, and hyperscalers: Positive, medium magnitude. The mechanism is a potentially stronger integrated supplier offering optical transport, protocol, and Ethernet traffic validation at 3.2T, improving procurement efficiency and roadmap alignment. Near-term trading catalyst: Q3 Spirent contribution was strong, but Q4 normalization means near-term revenue support moderates. Longer-duration fundamental shift: The acquisition can evolve from revenue addition to product-cycle leverage if the 3.2T integrated platform is competitive. This raises share-risk questions for incumbent high-speed Ethernet test vendors.

中文
0
6
34
5K
chen retweetledi
Leonard
Leonard@Leoskie_L·
唷!台股市場打開天花板了嗎? 先是突破第七名 第六名, 個股法人可以持股20% .接下來IB盈透證卷也開放. 「門打開」!更多投資人可以進來 = 市場參與度變高,市場變大 = 流動性變好,台積電是最大受益者 相信更多小鬼股都可以被挖掘. 再來看看有什麼
Kay Capital@keyahayek

卧槽,IB 除了终于开了韩国交易权限以外,还能开台湾了?我这就来为 @woodycryptow 接盘🤣 以前要开韩国可太难了,要透过 Pershing 动辄几个月。开台湾股票交易权限对中国大陆出生的更是完全没听说过有路径。

中文
3
5
75
12.9K
chen retweetledi
老季
老季@XiaoJi0403·
给脱臼的动物们正骨的过程,超解压
中文
32
274
3.3K
112.8K
joe xu
joe xu@JoeXuY·
@kimm_wei 买台股是有门槛的,比如最少要买1万美金。很多人拿不出来的。
中文
6
0
2
621
kimm
kimm@kimm_wei·
我看推特上面很多台湾人有买山寨币被割的,我好奇为啥他们不买台股,台股几十倍的太多了,他们有这么多优势,便利,不像大陆人民想买优质股票很难,他们闭着眼买台股都稳赚啊?
中文
51
0
85
48.8K
chen
chen@nikaidoula·
@innerpeacejojo @kimm_wei 沾上台積電的小型股都能翻幾倍,台股2位數漲到4位數的股不少,選對了比中彩券簡單
中文
0
0
1
31
jolliebeejojo
jolliebeejojo@innerpeacejojo·
@kimm_wei 台股是不是也就TSMC一枝独秀?和韩国一样,全民三星海力士押注国运。
中文
4
0
0
444
chen retweetledi
Angus電科技
Angus電科技@wuanguscc·
⚡⚡⚡最新消息:義大利參議員正式要求政府加速核准Tesla FSD 👉 義大利參議員Carlo Calenda於4月28日正式向交通部長提出議會質詢🇮🇹,要求政府優先推進Tesla FSD(監督模式)的核准,並直接與Tesla及荷蘭車輛主管機關RDW技術接洽🇳🇱。 他指出:義大利路上已有數萬輛具備FSD硬體的Tesla,啟用將進一步降低因分心或人為失誤導致的事故風險。 RDW的FSD歐洲簡報已從20分鐘延長至1小時(5月5日),下週歐洲層級會議若順利,2026年夏季核准是相當現實的可能性。 ✍️ 這波歐洲FSD核准潮正在加速。Tesla預估德法義等國將在荷蘭核准後4-8週內跟進。歐洲的推進速度也是觀察台灣何時跟上的重要參考指標🇹🇼 ( ok 我知道台灣有台灣的玩法)
Angus電科技 tweet media
Andrea Stroppa 🐺 Claudius Nero's Legion 🐺@andst7

Il Senatore Carlo Calenda critica spesso Musk, ma questo suo atto al Senato è notevole: con l’interrogazione 4-02971 spinge per l’approvazione rapida della guida autonoma supervisionata di Tesla in Italia. Cita i benefici per la sicurezza stradale, l’approvazione già arrivata nei Paesi Bassi e l’esigenza di non restare indietro rispetto al resto d’Europa. Un’iniziativa interessante che arriva dall’opposizione. senato.it/show-doc?leg=1…

中文
3
4
32
1.7K
chen
chen@nikaidoula·
@KiwiFIRELife 覺得中國變的是看太多紅媒了🤣我從以前用吧貼用b站,到現在明顯感到更嚴格了
中文
1
0
10
1.4K
chen retweetledi
Fiona ❤️& ✌️
如果你想真的了解光通讯产业链,那以下的内容不容错过——一图了解全产业链。 文章稍长,但是把光通讯的每个环节都讲的清楚明白需要费一些笔墨。 光通讯相比较电信号本质上还是一种信息传输的方式。只是光通讯更加高效,可以传递的信息量/速度都远优于传统的电缆。现在数据中心不可逆的趋势,就是需要传递的数据量的指数级增长。这个背景下,光通讯的流行和应用是势之必然。 —--------------------------- 整条链我分成了6个板块,我们一个一个来看。 1️⃣原材料,基本中国是龙头 铟(Indium)是一种稀有金属,主要产在中国,全球产量里中国占60%以上。铟本身没什么特别用处,但和磷结合之后,就变成了磷化铟(InP)——这个组合有一个神奇的物理特性:通电之后会发出激光。 镓(Gallium)也是稀有金属,中国2023年宣布出口限制,和砷结合变成砷化镓(GaAs),也能发光,用在短距离的场景。 做一个类比:原材料就像是"面粉"。面粉本身不是食物,但没有面粉就做不出面包。铟和镓就是制造激光器的"面粉"。 2️⃣衬底层 把铟和磷在极高温度下熔融、结晶、切片,就得到一片圆形的薄晶圆,叫做磷化铟衬底(InP Substrate)。直径约4-6英寸,看起来就像一片黑色的圆玻璃。 它是所有后续加工的基础材料,就像做芯片要先有硅晶圆一样。没有这片衬底,后面的激光器根本没有东西可以"长"在上面。 衬底就是"白纸"。在上面可以画出激光器的电路结构。 $AXTI 是全球最大的白纸供应商,占全球60-70%的份额。另一家是日本住友。 3️⃣外延/芯片 拿到InP衬底之后,用一种叫外延生长(Epitaxy)的技术,在这片晶圆上一层一层地沉积几纳米厚的半导体薄膜,形成精密的量子结构。这个过程就像在白纸上精密印刷电路。 这步做完之后,晶圆上就有了激光器的"基因"——一个可以在通电后发光的微小结构。然后把晶圆切成一个个极小的芯片颗粒,每一颗就是一个激光芯片(Laser Die)。 外延相当于"在面团上发酵",让基础材料产生功能性结构。这一步技术门槛极高,良率只有15-50%,大量芯片在这步就报废了。 4️⃣激光器——让“信号”发光 激光器就是把上一步做好的激光芯片封装成一个可以使用的元件。通电之后,它会发出特定波长的红外激光——肉眼看不见,但光纤最爱传这个波段。 激光器是灯泡,光模块是台灯。没有灯泡,台灯就是一个空壳。 5️⃣光模块——光电世界里的翻译官 光模块(Optical Transceiver)是一个大约手指大小的小盒子,直接插进服务器或交换机的插槽里。它里面装着:激光器(发光)、光探测器(接收光)、驱动芯片、镜头和光纤接口。 它的作用是双向翻译:服务器发出电信号 → 光模块把它变成光信号打进光纤;光纤传来光信号 → 光模块把它变回电信号送给服务器。 光模块是连接两个世界的翻译官,他把光世界的语言和电世界的语言来回翻译。 6️⃣交换机层 ——指挥交通、分拣信息 交换机(Switch)是一台专门负责"分发数据"的设备,长得像一个扁平的大铁箱子,正面插满了几十上百个光模块插槽。 在AI数据中心里,成千上万台GPU服务器要互相通信——比如一个大模型训练任务,可能同时用到1万块GPU,它们之间每时每刻都在交换中间计算结果。交换机就是负责把这些数据包准确、快速地送到对的GPU。 一台顶级AI交换机(比如博通的Tomahawk系列)上面可以插128个800G光模块,每秒总吞吐量超过100Tb。 交换机很像我们现实世界的快递中转站。每个包裹(数据包)进来,它看一眼地址,立刻分拣到正确的出口。速度极快,几纳秒就完成一次分拣。 —--------------------------- 最后的最后,就是大客户们买单了。 看完这这链条,再炒CPO光通讯,心里是不是有谱多了?
Fiona ❤️& ✌️ tweet media
中文
5
75
243
40K
Yao Zhang 張堯 🇨🇦
引述:你無法在 2030 年之前購買新的燃氣輪機。GE、Siemens 和 Mitsubishi 的訂單已排到 2029 年。輪機價格自 2019 年以來幾乎翻了三倍。每個 AI 資料中心都需要電力,每個燃氣電廠都需要一台輪機。而每台輪機都有一個部件成為整個產業的瓶頸:葉片。 一組葉片造價 60 萬美元,並需 90 週的時間生產。同樣的金屬冶金技術驅動著現代噴射發動機。全球只有 3 家公司能夠製造它們。 中國花費了 420 億美元試圖迎頭趕上。他們購買了一台俄羅斯戰鬥機發動機,將其拆解,並複製了每個部件。他們的複製品大修間隔僅 30 小時,而原版則為 400 小時。現代西方發動機的運行時間則達 4,000 小時。 你可以逆向工程輪機葉片的形狀。但你無法逆向工程 60 年的冶金技術。
Gaurab Chakrabarti@Gaurab

You cannot buy a new gas turbine until 2030. Order books at GE, Siemens, and Mitsubishi stretch to 2029. Turbine prices have nearly tripled since 2019. Every AI data center needs power and every gas plant needs a turbine. And every turbine has one part that bottlenecks the entire industry: The blade. It has to survive in gas 500°C above the melting point of the metal it's made from and spin at up to 20,000 RPM under 10,000 g of centrifugal force. Each blade is grown as a single crystal of nickel superalloy, pulled through a vacuum furnace at 3 mm per minute. A set of blades costs $600,000 and takes 90 weeks to grow. The same metallurgy powers modern jet engines. Only 3 companies on Earth can build one. China spent $42 billion trying to catch up. They bought a Russian fighter engine, took it apart, and copied every part. Their copy ran 30 hours between overhauls versus 400 for the original. Modern Western engines run 4,000. You can reverse engineer the shape of a turbine blade. You cannot reverse engineer 60 years of metallurgy.

中文
4
13
72
3.8K
chen retweetledi
范士贵8964荷兰难民
范士贵8964荷兰难民@fanshigui8964·
2003年成都3岁的小女孩李思怡被饿死在家里。她的妈妈外出时把她反锁在家里。妈妈因为偷东西被抓。她多次跪求警察,说自己的孩子还反锁在家里。警察根本不理睬。可怜的孩子独自在家17天,活活饿死。 邻居听到过孩子的哭泣,但是他们没有管。直到闻到臭味才报警。 涉事警察只判了3年,但是没有关押。 老范当年多次为这个小女孩悲泣,现在再次看到这个信息依然泪流不禁。 希望这样的悲剧不要再发生在我们的同胞、我们的孩子身上。😭 Grok **李思怡事件* (也称“3岁幼女饿死案”)发生在2003年6月四川省成都市,是当时震动全国的一起悲剧案件,与你描述的基本一致:母亲李桂芳因吸毒和盗窃被警方带走,她多次跪求警察先安置家中被反锁的3岁女儿李思怡,但警方未采取有效措施,导致孩子被独自困在家中17天,最终活活饿死。涉事民警因玩忽职守被判刑,但量刑较轻,引发广泛社会讨论和对执法冷漠、社会保障缺位的质疑。 ### 事件详细经过 - **时间与地点**:2003年6月4日中午,家住成都市青白江区“九千小区”(或青江西路65号院)的李桂芳(当时39岁,长期吸毒,无固定工作,单亲母亲)将年仅2岁10个月左右的女儿李思怡(2000年10月出生)反锁在主卧室内,用绿色毛线绳拴住门把手,然后外出“找钱”。她先带孩子吃了午饭(这是李思怡最后一餐),孩子当时生病刚打过针,已睡着。 - **母亲被抓**:李桂芳到金堂县红旗商场(超市)偷了两瓶洗发水,被保安抓获并报案。金堂县公安局城郊派出所民警黄小兵、王华麟赶到,将她带回派出所。尿检显示她吸毒,派出所决定对其强制戒毒。 - **母亲的多次哀求**:在派出所、押送途中(警车由副所长王新驾驶),李桂芳反复告诉民警家中只有3岁女儿无人照顾。她拉住车门不肯上车、跪地哭求先回家安置孩子,或给姐姐打电话让亲属照顾,甚至在车过青白江(家附近)时用头猛撞车门。民警最终同意打了一个电话(给姐姐,无人接听)和团结村派出所的值班电话,但后续无人跟进,也未按规定在3日内送达《强制戒毒通知书》(通知书仍留在抽屉里)。警车两次路过李桂芳家门,均未停车。 - **孩子死亡**:从6月4日下午1点多到6月21日傍晚(共17天),李思怡独自被困。6月21日傍晚,邻居闻到腐臭味报警,青白江区团结村派出所民警从阳台进入,发现主卧门后有阻力,推开后见到高度腐败的幼童尸体(头发脱落、头部白骨化、满身蛆虫)。法医鉴定:非暴力或中毒致死,死于**饥渴**(胃、心脏等自溶,死亡时间距尸检10天以上)。现场显示孩子曾拼命求生:门上有抓痕、指甲损伤、柜子被翻找、屎尿用纸包裹、可能躲进衣柜过夜。 李桂芳此前因类似情况多次被抓后释放(考虑到孩子),但这次未及时处理孩子问题。团结村派出所(距离李家仅一个街区,姐姐家不足200米)两次接到相关电话,也未行动。 ### 法律处理与后续 - **涉事民警**:城郊派出所副所长王新(第一被告)、民警黄小兵(第二被告)被认定玩忽职守。2004年7月1日逮捕,8月20日成都市新都区法院一审判决:王新有期徒刑3年(后取保候审)、黄小兵有期徒刑2年。其他涉事人员(如金堂县公安局值班领导、团结村派出所所长等)受到行政处分或停职,但未全员刑事追究。 - **赔偿**:李思怡亲属获10万元国家赔偿。 - **社会反响**:事件曝光后引发巨大愤怒,网友自发建纪念网站、写诗、绝食悼念。中国科学院研究员康晓光实地调查后自费出版《起诉》一书,指出这不仅是民警个人失职,更是制度冷漠(民政、妇联、儿童福利院等均未有效救助李思怡,母亲吸毒背景导致其被排斥,无低保等支持)。成都市委政法委书记批示称“一个无辜的小生命就活活饿死在我们这些‘冷血者’手中”。 这个案件至今仍被视为中国社会保障与执法责任的典型悲剧案例,后续类似事件(如2013年南京幼女饿死案)常被拿来对比。信息主要来自当时法院审理、媒体报道(如新浪新闻、中国青年报)和维基百科总结,细节高度一致。如果你需要特定来源的原文链接或更多背景,我可以进一步说明。愿这样的悲剧不再重演。
范士贵8964荷兰难民 tweet media
中文
145
302
1.4K
257.6K