思维纪要社 HTML 来自本地 Obsidian vault 的静态发布
主题 2026-09-02/topics/356.md 原文

#外资研报

图片

无图片

正文

详情页仍未获取到全文。原始列表正文可能被截断。

总体总结

主题正文

  1. 详情页仍未获取到全文。
  2. 原始列表正文可能被截断。

附件总结 SemiAnalysis-大多数新云服务商安全状况堪忧-20260830.pdf

  1. is mixed, and ultimately not statistically significant. August 2026 will be its most prolific
  2. some cases, over 3 years old), and a healthy token budget for all the latest models (with all
    1. And, most critically, one scenario with cascading vulnerabilities (i.e. “all of the above”)
  4. industry leading Datacenter Model that tracks over 6,000 sites globally, including all the top
  5. total of 25 providers and 32 clusters where we did extensive testing for this round, and many
  6. cluster had a new partition key (0xa601), but the default partition key (0xffff) was still active.
  7. general availability in early 2023. The pitch is simple enough. The card offloads, accelerates,
  8. 5.6 Sol would judge it correct or not. After a point, though, 5.6 Sol would also start rejecting

UBS-5000+家A股公司告诉你什么-— Q226盈利同比增29%-20260901.pdf

  1. a 19.5% earnings gain for all A-shares in H126, estrisudnare ih A-sssro acthwro gsgnif earn oarymm Su well above the full-year 2026 rate of S1460524080001
  2. Figure 1: Revenue growth of all A-shares (YoY, YTD) Figure 2: Revenue growth of all A-shares (YoY, quarterly)
  3. Q112Q113Q114Q115Q116Q117Q118Q119Q120Q121Q122Q123Q124Q125Q126 Q112Q113Q114Q115Q116Q117Q118Q119Q120Q121Q122Q123Q124Q125Q126
  4. Figure 6: Earnings growth of all A-shares (YoY, YTD) Figure 7: Earnings growth of all A-shares (YoY, quarterly)
  5. Q112Q113Q114Q115Q116Q117Q118Q119Q120Q121Q122Q123Q124Q125Q126 Q112Q113Q114Q115Q116Q117Q118Q119Q120Q121Q122Q123Q124Q125Q126
  6. Figure 12: Gross margin of non-financial A-shares Figure 13: Net margin of non-financial A-shares (quarterly)
  7. Figure 15: CSI 300 earnings growth forecast adjustment Figure 16: CSI 300 earnings growth forecast adjustment,
  8. CSI 300 in April of the respective year. For example, T in 2026 refers to April 2026. CSI 300 in April of the respective year. For example, T in 2023 refers to April 2023.

UBS-爱博医疗-688050-德美并表拉动收入增长，盈利承压.pdf

  1. 0JQGWGm2pVdVaY7NdN9PoMpPtRoQeRmMyQjMsQsN9PrQnRuOsOxPuOoOtR

UBS-半导体速递：CBRS、英伟达、特朗普-习会与H1B数据-20260831.pdf

  1. 2018 2019 2020 2021 2022 2023 2024 2025 Total 2018 2019 2020 2021 2022 2023 2024 2025 Total 2018 2019 2020 2021 2022 2023 2024 2025
  2. Nvidia 10,570 8,714 11,201 15,667 17,699 20,617 33,868 38,268 156,604 461 330 529 699 874 1,028 1,405 1,895 7,221 $ 153k$ 167k$ 179k$ 186k$ 207k$ 205k$ 218k$ 222k
  3. Synopsys 7,316 7,619 6,828 7,712 6,273 5,203 8,285 6,181 55,417 177 208 188 221 198 150 216 194 1,552 $ 150k$ 154k$ 163k$ 163k$ 172k$ 182k$ 187k$ 194k
  4. Intel 4,962 5,154 3,706 4,050 6,508 3,286 5,651 4,634 37,951 1,530 1,581 1,484 4,003 4,809 3,286 3,553 2,625 22,871 $ 120k$ 121k$ 131k$ 128k$ 136k$ 140k$ 149k$ 158k
  5. Qualcomm* 31,879 34,699 67,509 66,227 81,119 73,043 2,337 33,154 389,967 648 687 1,806 2,449 3,448 3,164 2,337 2,167 16,706 $ 137k$ 143k$ 148k$ 153k$ 157k$ 165k$ 162k$ 172k
  6. Samsung 1,020 1,508 2,847 2,710 2,596 2,208 1,892 1,847 16,628 633 628 510 523 586 512 485 479 4,356 $ 129k$ 132k$ 139k$ 153k$ 153k$ 172k$ 184k$ 182k
  7. AMD 288 1,064 552 577 1,280 563 812 773 5,909 288 470 541 577 672 563 812 773 4,696 $ 140k$ 142k$ 146k$ 154k$ 161k$ 164k$ 175k$ 181k
  8. Micron 394 507 583 610 635 321 505 562 4,117 394 507 583 610 635 321 505 562 4,117 $ 109k$ 113k$ 112k$ 115k$ 122k$ 122k$ 133k$ 134k

UBS-北方华创-002371-快评：2Q26归母净利润符合预期，高端设备覆盖度进一步提升-20260825.pdf

  1. 5EPFVJjBsYcUeU9PcMaQpNnNtRsMlOnNxPlOmOoR6MqRoQvPoMyQuOnMxO

MS-中国股票策略：香港市场暂时退居次席-20260826_Laura Wang，Chloe Liu(原文).pdf

    • 2Q results season proving bottoming out of earnings, especially for the internet/e-commerce sectors, thanks to reining
  2. Footnotes Guideline Source: MSCI, FactSet, DataStream, Morgan Stanley Research. Data as of August 24, 2026. Past performance is no guarantee of future results.
  3. Footnotes Guideline Source: RIMES, FactSet, Morgan Stanley Research forecasts. Note: Weekly price index data as of August 24, 2026.
  4. 31-ceD 41-nuJ 41-ceD 51-nuJ 51-ceD 61-nuJ 61-ceD 71-nuJ 71-ceD 81-nuJ 81-ceD 91-nuJ 91-ceD 02-nuJ 02-ceD 12-nuJ 12-ceD 22-nuJ 22-ceD 32-nuJ 32-ceD 42-nuJ 42-ceD 52-nuJ 52-ceD 62-nuJ
  5. 31-ceD 41-nuJ 41-ceD 51-nuJ 51-ceD 61-nuJ 61-ceD 71-nuJ 71-ceD 81-nuJ 81-ceD 91-nuJ 91-ceD 02-nuJ 02-ceD 12-nuJ 12-ceD 22-nuJ 22-ceD 32-nuJ 32-ceD 42-nuJ 42-ceD 52-nuJ 52-ceD 62-nuJ
      • 3 2 0 0 % % -15% -19% -20% -25% -23% -20% -2 -1 2 6 % % -2 - 2 2 % 4% -24 - % 25% -1 -2 9 6 %- % 2 -2 1 4 %- % 1 -3 8 0 % % -2 -2 0 7 % - - % 2 1 4 7 % % -24% -20% -23% -20%
  7. Source and Source: MSCI, Bloomberg, RIMES, Factset, Morgan Stanley Research. Data as of August Source: Factset, Morgan Stanley Research. Note: Monthly data as of July 2026.
  8. Source: FactSet, Datastream, ModelWare, Morgan Stanley Research. Note: Data as of August 24, 2026. *Only index weight of Morgan Stanley Research-covered

MS-自动化——中国8月制造业PMI的启示-20260901.pdf

  1. A utomation – Read-across from Derrick.Yang@morganstanley.com +886 2 2730-2862
  2. In this report, we focus on China's manufacturing PMI, which we Greater China Technology Hardware
  3. • This is supported by the upturns of 0.5ppt and 2.1ppt i n the production index
  4. increasing policy fine-tuning this fall to facilitate rollout of the ~Rmb2trn in-
  5. Base case, 29x 2027e P/E. We believe this methodology better captures earnings growth
  6. with business cycles). As the industry enters an up-cycle, we think applying a near-peak cycle
  7. momentum, as its valuation tends to move with the business cycle. We apply a target multiple
  8. of 38x to our 2027 EPS estimate. We view this multiple as justified by the 47% operating profit

文档

SemiAnalysis-大多数新云服务商安全状况堪忧-20260830.pdf

总结

  1. is mixed, and ultimately not statistically significant. August 2026 will be its most prolific
  2. some cases, over 3 years old), and a healthy token budget for all the latest models (with all
    1. And, most critically, one scenario with cascading vulnerabilities (i.e. “all of the above”)
  4. industry leading Datacenter Model that tracks over 6,000 sites globally, including all the top
  5. total of 25 providers and 32 clusters where we did extensive testing for this round, and many
  6. cluster had a new partition key (0xa601), but the default partition key (0xffff) was still active.
  7. general availability in early 2023. The pitch is simple enough. The card offloads, accelerates,
  8. 5.6 Sol would judge it correct or not. After a point, though, 5.6 Sol would also start rejecting

预览

SemiAnalysis-大多数新云服务商安全状况堪忧-20260830.pdf

The Information-独家：SpaceX 自建涡轮叶片工厂 ，以解决数据中心电力短缺问题 Exclusive SpaceX Lays Groundwork For Turbine-Blade Factory to Solve.pdf

预览

The Information-独家：SpaceX 自建涡轮叶片工厂 ，以解决数据中心电力短缺问题 Exclusive SpaceX Lays Groundwork For Turbine-Blade Factory to Solve.pdf

UBS-5000+家A股公司告诉你什么-— Q226盈利同比增29%-20260901.pdf

总结

  1. a 19.5% earnings gain for all A-shares in H126, estrisudnare ih A-sssro acthwro gsgnif earn oarymm Su well above the full-year 2026 rate of S1460524080001
  2. Figure 1: Revenue growth of all A-shares (YoY, YTD) Figure 2: Revenue growth of all A-shares (YoY, quarterly)
  3. Q112Q113Q114Q115Q116Q117Q118Q119Q120Q121Q122Q123Q124Q125Q126 Q112Q113Q114Q115Q116Q117Q118Q119Q120Q121Q122Q123Q124Q125Q126
  4. Figure 6: Earnings growth of all A-shares (YoY, YTD) Figure 7: Earnings growth of all A-shares (YoY, quarterly)
  5. Q112Q113Q114Q115Q116Q117Q118Q119Q120Q121Q122Q123Q124Q125Q126 Q112Q113Q114Q115Q116Q117Q118Q119Q120Q121Q122Q123Q124Q125Q126
  6. Figure 12: Gross margin of non-financial A-shares Figure 13: Net margin of non-financial A-shares (quarterly)
  7. Figure 15: CSI 300 earnings growth forecast adjustment Figure 16: CSI 300 earnings growth forecast adjustment,
  8. CSI 300 in April of the respective year. For example, T in 2026 refers to April 2026. CSI 300 in April of the respective year. For example, T in 2023 refers to April 2023.

预览

UBS-5000+家A股公司告诉你什么-— Q226盈利同比增29%-20260901.pdf

UBS-爱博医疗-688050-德美并表拉动收入增长，盈利承压.pdf

总结

  1. 0JQGWGm2pVdVaY7NdN9PoMpPtRoQeRmMyQjMsQsN9PrQnRuOsOxPuOoOtR

预览

UBS-爱博医疗-688050-德美并表拉动收入增长，盈利承压.pdf

UBS-半导体速递：CBRS、英伟达、特朗普-习会与H1B数据-20260831.pdf

总结

  1. 2018 2019 2020 2021 2022 2023 2024 2025 Total 2018 2019 2020 2021 2022 2023 2024 2025 Total 2018 2019 2020 2021 2022 2023 2024 2025
  2. Nvidia 10,570 8,714 11,201 15,667 17,699 20,617 33,868 38,268 156,604 461 330 529 699 874 1,028 1,405 1,895 7,221 $ 153k$ 167k$ 179k$ 186k$ 207k$ 205k$ 218k$ 222k
  3. Synopsys 7,316 7,619 6,828 7,712 6,273 5,203 8,285 6,181 55,417 177 208 188 221 198 150 216 194 1,552 $ 150k$ 154k$ 163k$ 163k$ 172k$ 182k$ 187k$ 194k
  4. Intel 4,962 5,154 3,706 4,050 6,508 3,286 5,651 4,634 37,951 1,530 1,581 1,484 4,003 4,809 3,286 3,553 2,625 22,871 $ 120k$ 121k$ 131k$ 128k$ 136k$ 140k$ 149k$ 158k
  5. Qualcomm* 31,879 34,699 67,509 66,227 81,119 73,043 2,337 33,154 389,967 648 687 1,806 2,449 3,448 3,164 2,337 2,167 16,706 $ 137k$ 143k$ 148k$ 153k$ 157k$ 165k$ 162k$ 172k
  6. Samsung 1,020 1,508 2,847 2,710 2,596 2,208 1,892 1,847 16,628 633 628 510 523 586 512 485 479 4,356 $ 129k$ 132k$ 139k$ 153k$ 153k$ 172k$ 184k$ 182k
  7. AMD 288 1,064 552 577 1,280 563 812 773 5,909 288 470 541 577 672 563 812 773 4,696 $ 140k$ 142k$ 146k$ 154k$ 161k$ 164k$ 175k$ 181k
  8. Micron 394 507 583 610 635 321 505 562 4,117 394 507 583 610 635 321 505 562 4,117 $ 109k$ 113k$ 112k$ 115k$ 122k$ 122k$ 133k$ 134k

预览

UBS-半导体速递：CBRS、英伟达、特朗普-习会与H1B数据-20260831.pdf

UBS-北方华创-002371-快评：2Q26归母净利润符合预期，高端设备覆盖度进一步提升-20260825.pdf

总结

  1. 5EPFVJjBsYcUeU9PcMaQpNnNtRsMlOnNxPlOmOoR6MqRoQvPoMyQuOnMxO

预览

UBS-北方华创-002371-快评：2Q26归母净利润符合预期，高端设备覆盖度进一步提升-20260825.pdf

MS-中国股票策略：香港市场暂时退居次席-20260826_Laura Wang，Chloe Liu(原文).pdf

总结

    • 2Q results season proving bottoming out of earnings, especially for the internet/e-commerce sectors, thanks to reining
  2. Footnotes Guideline Source: MSCI, FactSet, DataStream, Morgan Stanley Research. Data as of August 24, 2026. Past performance is no guarantee of future results.
  3. Footnotes Guideline Source: RIMES, FactSet, Morgan Stanley Research forecasts. Note: Weekly price index data as of August 24, 2026.
  4. 31-ceD 41-nuJ 41-ceD 51-nuJ 51-ceD 61-nuJ 61-ceD 71-nuJ 71-ceD 81-nuJ 81-ceD 91-nuJ 91-ceD 02-nuJ 02-ceD 12-nuJ 12-ceD 22-nuJ 22-ceD 32-nuJ 32-ceD 42-nuJ 42-ceD 52-nuJ 52-ceD 62-nuJ
  5. 31-ceD 41-nuJ 41-ceD 51-nuJ 51-ceD 61-nuJ 61-ceD 71-nuJ 71-ceD 81-nuJ 81-ceD 91-nuJ 91-ceD 02-nuJ 02-ceD 12-nuJ 12-ceD 22-nuJ 22-ceD 32-nuJ 32-ceD 42-nuJ 42-ceD 52-nuJ 52-ceD 62-nuJ
      • 3 2 0 0 % % -15% -19% -20% -25% -23% -20% -2 -1 2 6 % % -2 - 2 2 % 4% -24 - % 25% -1 -2 9 6 %- % 2 -2 1 4 %- % 1 -3 8 0 % % -2 -2 0 7 % - - % 2 1 4 7 % % -24% -20% -23% -20%
  7. Source and Source: MSCI, Bloomberg, RIMES, Factset, Morgan Stanley Research. Data as of August Source: Factset, Morgan Stanley Research. Note: Monthly data as of July 2026.
  8. Source: FactSet, Datastream, ModelWare, Morgan Stanley Research. Note: Data as of August 24, 2026. *Only index weight of Morgan Stanley Research-covered

预览

MS-中国股票策略：香港市场暂时退居次席-20260826_Laura Wang，Chloe Liu(原文).pdf

MS-自动化——中国8月制造业PMI的启示-20260901.pdf

总结

  1. A utomation – Read-across from Derrick.Yang@morganstanley.com +886 2 2730-2862
  2. In this report, we focus on China's manufacturing PMI, which we Greater China Technology Hardware
  3. • This is supported by the upturns of 0.5ppt and 2.1ppt i n the production index
  4. increasing policy fine-tuning this fall to facilitate rollout of the ~Rmb2trn in-
  5. Base case, 29x 2027e P/E. We believe this methodology better captures earnings growth
  6. with business cycles). As the industry enters an up-cycle, we think applying a near-peak cycle
  7. momentum, as its valuation tends to move with the business cycle. We apply a target multiple
  8. of 38x to our 2027 EPS estimate. We view this multiple as justified by the 47% operating profit

预览

MS-自动化——中国8月制造业PMI的启示-20260901.pdf

Sam Altman：OpenAI 今年可能实现 AGI —— 作者 Alex Heath --- Sam Altman OpenAI may reach AGI this year - by Alex Heath.pdf

预览

Sam Altman：OpenAI 今年可能实现 AGI —— 作者 Alex Heath --- Sam Altman OpenAI may reach AGI this year - by Alex Heath.pdf

源文件：2026-09-02/topics/356.md
站内副本：2026-09-02/topics/356.md