Key Takeaways: PCB Supply Chain
> Golden Week factory closures were cut to 1–3 days (vs. 5–6 days previously), with AI-tied production lines skipping holidays entirely to run at full capacity.> Capacity is fully booked through year-end, with mid-to-late October orders pushing production schedules out to January.> Non-AI PCB average selling prices (ASPs) surged ~10%+ MoM in October, led by automotive boards, driving net margins higher.> HC glass-cloth-free CCL solutions remain in early NVIDIA testing, while CSP supply chain focus shifts to 3.2T optical transceiver boards targeted for 2028 mass production using M9-grade CCL.
T-Rex 2X Inverse NVIDIA Daily Target ETF
NVDQ8.2502.36% ( +0.190 )
Share your thoughts
What's on your mind?
0 / 2,000
- S
$NVIDIA(NVDA.US) opened its Vera Rubin rack designs at the OCP Global Summit and lined up Meta, Google, and Cisco on the networking side
The company contributed Vera Rubin rack and compute tray designs through its MGX architecture. The updated MGX 1.2 spec supports racks up to 250kW with liquid cooling and fanless compute trays$Meta Platforms(META.US) is adopting NVIDIA Spectrum-6 Ethernet switches at 102.4Tb/s. With Spectrum-X Multiplane, the setup is meant to connect more than 100,000 GPUs across up to eight independent network pathsCisco $Cisco(CSCO.US) and $NVIDIA(NVDA.US) also announced USFP, a new pluggable optical module aimed at 400Gb/s per lane. They say that is 4x the bandwidth density and about 1.3x the power efficiency of today's 200Gb/s-per-lane optics, with capacity up to 12.8Tb/s in roughly the same footprintGoogle's Falcon Transport will ride on future ConnectX-10 SuperNICs and BlueField DPUs, and $NVIDIA(NVDA.US) also dropped open-source DSX tools for configuring networks and stress-testing GPU clustersNetworking is becoming as strategic as the GPU itself. Opening Vera Rubin while locking in $Meta Platforms(META.US), Google, and $Cisco(CSCO.US) is how $NVIDIA(NVDA.US) keeps the AI factory stack centered on its silicon even as hyperscalers push for more open infrastructureSource: StockMarket.News
- S
$AMD(AMD.US) unveiled Helion, an open server rack built for AI agents
The design is headed to the Open Compute Project so partners can build on it. One config is CPU-only with up to 32,768 EPYC cores for hyperscalers. Another pairs EPYC CPUs with Instinct MI350P GPUs for enterprise all-in-one deploymentsHelion comes in air-cooled and liquid-cooled versions, and AMD says a built-in router can send agent requests across frontier and open-source modelsAvailability is targeted for November 2026Agentic workloads need dense CPU orchestration plus optional GPU punch, not just more training clusters. Helion is $AMD(AMD.US) pitching that rack as open infrastructure while $NVIDIA(NVDA.US) still dominates the closed AI systems tradeSource: StockMarket.News
- S
$Microsoft(MSFT.US) just made its biggest bet yet on its own silicon
CEO Satya Nadella said Azure Maia will scale to multi-gigawatt deployments starting next year and become the primary inference system for $Microsoft(MSFT.US) products and servicesThe company also unveiled Maia 300, claiming about 40% more tokens per megawatt versus leading GPU systems and up to 7x the rack-scale token throughput of Maia 200Each rack packs 96 accelerators. Per chip that means 288GB of HBM4 across 8 stacks, 18 TB/s of memory bandwidth, and a 950W envelope on TSMC's N2P processMaia 200 already runs production workloads including Microsoft 365 Copilot, with GitHub Copilot and Foundry next. Training still leans on $NVIDIA(NVDA.US) and $AMD(AMD.US) partners, but inference is where $Microsoft(MSFT.US) wants its own chips to take overOwning the inference stack is how a hyperscaler cuts GPU spend and stretches power when electricity is the real bottleneck. Scale only starts in 2027 though, so $NVIDIA(NVDA.US) still owns the near-term AI cycle while $Microsoft(MSFT.US) builds the in-house escape hatchSource: StockMarket.News
- C
AT&T -9.8% and T-Mobile -13% after SpaceX $8B nationwide spectrum buy is market missing point - Starlink still needs ground game and $40B debt for Nvidia chips. US telco fear is overdone. At home, banks bouncing after last week’s STI -5% selloff is real - NIM holds at 5%+ yields, 6% div yield is floor.
Featured☕️ [Task Coins Giveaway] Daily Market Talk — SpaceX's $8B Spectrum Deal Sinks TelcosSpaceX is paying about $8B for nationwide spectrum, and US telcos got repriced in one session: AT&T -9.8%, T-Mobile -13%. At home, the banks are bouncing after last week's 5% STI slide. Let's dig in 👇...
- J
it will be $70B or more. so far it is right on the dot for those who has been following.
Featured🎁[Reward] OpenAI ARR: $50B or $70B? The Gross-to-Net QuestionOn Oct 8, 2026 SGT, the Financial Times dropped a story that spooked the entire AI complex: $OpenAI(OpenAI.NA) told investors its annualized revenue run-rate (ARR) was “approaching $50B” as of end-Sep...
- G
$Microsoft(MSFT.US)
Context: Nvidia is partnering with Microsoft to beat Apple in high end notebook space. Instead of GPU, Nvidia is building a CPU that directly competes with Intel and AMD. Microsoft claims that the Nvidia CPU that is installed in Microsoft branded laptops called$NVIDIA(NVDA.US)$Intel(INTC.US)$Netflix(NFLX.US) Surface Ultra can run twice as fast as Apple Macbook Pro M5 yet still cheaper by US$400. The Ultra cost US$2599.
My trade: Microsoft Surface laptop is only contributing to 2% of Microsoft revenue so I am neutral on it's stock price
Takeaway: Microsoft should consider competing with Apple Macbook Neo which recently was hiked to US$699. They should consider making AI notebooks affordable to the masses. The Ultra pricing is out of reach for many people especially the students.
- K
The headline turned out to be wrong and caused a sell-off even though they corrected it on Thursday. Would say opportunities are still there but have to be mindful of the upcoming week regarding CPI.
Featured🎁[Reward] OpenAI ARR: $50B or $70B? The Gross-to-Net QuestionOn Oct 8, 2026 SGT, the Financial Times dropped a story that spooked the entire AI complex: $OpenAI(OpenAI.NA) told investors its annualized revenue run-rate (ARR) was “approaching $50B” as of end-Sep...
- N
The OpenAI ARR controversy is not proof the AI boom is dead, but it is a very important warning shot: the market is no longer willing to accept un-audited, inflated revenue run-rate numbers without question.
For buyers: it is an opportunity only for patient investors who can tolerate volatility and pick quality companies; it is NOT an automatic buy signal for every AI stock.
Featured🎁[Reward] OpenAI ARR: $50B or $70B? The Gross-to-Net QuestionOn Oct 8, 2026 SGT, the Financial Times dropped a story that spooked the entire AI complex: $OpenAI(OpenAI.NA) told investors its annualized revenue run-rate (ARR) was “approaching $50B” as of end-Sep...
- 谋
Featured[Mou Weekly] Interest expenses exceed 1 trillion, can the US government no longer sustain it? (Week 41 of 2026 | Total Issue 291) In the absence of a recession, the deficit remains
Last Tuesday, the S&P 500 closed at 7,818.93, marking its first closing record high since August 13, with a weekly gain of 1.2%. On Wednesday, the 10-year U.S. Treasury auction cleared at 5.30%, a 24-year high. On Thursday, the Congressional Budget Office released its fiscal year 2026 ledger: revenue of $5.40 trillion, spending of $7.40 trillion, and a deficit of $1.99 trillion; net interest of $1.14 trillion, which is $227 billion more than defense spending. The deficit stands at 6.2% of GDP, unemployment is at 4.2%, with no wars on home soil and no recession...
+3
10Y Hits A 24-Year High. Can S-REITs Hold Up?
