In today’s briefing:
- Trading Strategies on Samsung Electronics and SK Hynix Amid Heightened Competition from CXMT
- HSTECH Index: Market Consultation Conclusions, Updated Potential Adds & The Borderline Names
- TSMC (2330.TT;TSM.US): Preferred Scenario Is: Investment Slightly Exceeds Near-Term Demand by 2028
- Groundup.ai: Expanding from Pilot Projects to Enterprise Contracts (Initiation)
- Dongshan Precision (东山精密) A+H Pre-IPO: PHIP and 1H26 Results Update
- Nanya Q326 Earnings Review. Has The Cycle Peaked?
- Ardian & Oak Holdings / INWIT: A Towering Question
- Ai Robotics (TSE: 247A): A Japanese Beauty Company With a Plan to Grow Its Value 17x by 2029
- Jio Platforms Pre-IPO: Peer Comparison and Valuation
- Synopsys Investor Day Q&A Session. Key Takeaways & Full Transcript
Trading Strategies on Samsung Electronics and SK Hynix Amid Heightened Competition from CXMT
- In this insight, I provide a near-term trading strategy on Samsung Electronics and SK Hynix amid heightened competition from CXMT.
- Prior to its 3Q 2026 conference call, I expect Samsung Electronics to trade in a box trading range of about 210,000 won to 240,000 won per share.
- I also expect SK Hynix to trade in a box trading range of about 1,300,000 won to 1,520,000 won prior to the the 3Q 2026 conference call.
HSTECH Index: Market Consultation Conclusions, Updated Potential Adds & The Borderline Names
- On 7 October, Hang Seng announced the conclusions of the market consultation to increase the number of Hang Seng TECH Index (HSTECH INDEX) constituents from 30 to 50.
- While most of the proposals will be implemented as proposed, a liquidity requirement has been tightened, and revenue thresholds have been imposed for companies qualifying through the Sales Growth Stream.
- The changes will be implemented in a single step at the December rebalance and that creates a lot of flow for all the index members, current and new.
TSMC (2330.TT;TSM.US): Preferred Scenario Is: Investment Slightly Exceeds Near-Term Demand by 2028
- We estimate that TSMC’s revenue could grow by approximately 40%–43% YoY in 2026.
- 2027 growth remains dominated by AI and its the only Key Growth Driver
- The Key Question: What Will Happen in 2028? Investment Slightly Exceeds Near-Term Demand — Our Preferred Scenario
Groundup.ai: Expanding from Pilot Projects to Enterprise Contracts (Initiation)
- Proprietary Industrial AI Positioning: Groundup.ai pairs non-invasive sensors with agentic AI to diagnose equipment faults, operates in Singapore, Indonesia, Australia and the Gulf, and entered Vietnam and Japan post-Series A.
- Demonstrated Operating Leverage: Groundup.ai turned profitable in FY25 (year to March 2025), with a US$0.1m (S$115.2k) net profit on US$1.7m (S$2.3m) revenue, then secured an over-US$10.0m multi-site contract.
- Enterprise Customer Concentration Risk: Groundup.ai depends heavily on one US$10.0m+ contract whose hardware-heavy rollout could strain working capital, following negative operating cash flow of US$0.6m in FY25.
Dongshan Precision (东山精密) A+H Pre-IPO: PHIP and 1H26 Results Update
- Dongshan Precision, a maker of PCBs, precision components and optical transceivers, is seeking an A+H listing. UBS, Haitong, GF, and CITIC are the joint sponsors.
- While our previous note covers the fundamentals of the company, in this note we look at its business development, 1H26 results and A-H premium moves.
- We think the deal is worth following as the optical transceiver business acquired in 2025 raised 1H26 gross margin to its highest level since 2023.
Nanya Q326 Earnings Review. Has The Cycle Peaked?
- Nanya reported Q326 revenues of NT$133,649 million, up 61.9% QoQ and up 611.7% YoY.
- Gross profit was NT$114,482 million, with a gross margin of 85.7%, up 6.2 percentage points QoQ and up 67.2 percentage points YoY.
- ASPs rose more than 60% QoQ and more than 500% YoY, almost entirely reflecting the renewal of long-term agreements (LTAs) at materially higher prices.
Ardian & Oak Holdings / INWIT: A Towering Question
- Ardian and Oak Holdings, controlling 70.6% of INWIT, are reportedly arranging financing for a take-private. Mubadala could join, but no formal offer has emerged.
- TIM and Fastweb, accounting for 80% of revenues, are challenging their tower contracts. Our DCF yields €6.44/share with contracts preserved and €5.62 risk-weighted.
- At €7.15, a hypothetical €8.50 bid offers 18.9% upside, against 21.4% standalone downside. We would stay on the sidelines pending a formal proposal.
Ai Robotics (TSE: 247A): A Japanese Beauty Company With a Plan to Grow Its Value 17x by 2029
- Not my usual kind of company I often write about Japanese and Korean companies that nobody has covered in English before.
- Ai Robotics is a bit different. It publishes its results and IR materials in English, and Philipp Haas, a fund manager and Substack author, has already written it up and owns the shares in his fund.
- On top of that, I usually avoid AI, robotics and chip stocks, but this business is not what the name suggests.
Jio Platforms Pre-IPO: Peer Comparison and Valuation
- Jio Platforms, a digital connectivity provider, has filed its draft prospectus. Its 19 bookrunners include international banks MS, UBS, BOA, Citi, CLSA, GS, HSBC, Jefferies, JPM and BNP Paribas.
- Our previous note included the company’s business, financials and pre-IPO investor backing.
- In this note, we compare the company with its listed peers, forecast earnings and provide our thoughts on valuation.
Synopsys Investor Day Q&A Session. Key Takeaways & Full Transcript
- For “Factory 2” engagements, Synopsys collects upfront priority/customization fees and secures volume-based royalties once chips enter production. Royalties carry 100% gross margin and will begin ramping in FY28/FY29
- Outcome-Based Revenue Sharing (Lane 3): OpenAI hosts the model/compute, and Synopsys splits revenue on a service basis derived from the Power, Performance, and Area (PPA) improvements achieved by the model
- AI models cannot validate physics; they require Synopsys’ core software (and Ansys sign-off tools) for fidelity checkpoints, placement, routing, and GDSII sign-off. Consequently, AI increases rather than decreases EDA usage.

