Contents
Market Wrap
AI & Semiconductor Value Chain
AI Models, Applications & Capex
AI Cloud & Data Center Operators
GPU / CPU / ASIC
HBM / DRAM / NAND / SSD / HDD
Foundries
Semiconductor Equipment & Testing
Optical Communications & Photonics
High-Speed Interconnects, Connectors, Thermal & Power
Internet & Platforms
Software & SaaS
Consumer Electronics & Intelligent Vehicles
Huang's Curated Portfolio
Memory Index Trend
U.S. Memory Turnover Share
Semiconductors maintained relative strength while software and cloud stocks remained under pressure. AI demand is broadening from model training to persistent, always-on agents, shifting investor focus to tangible proof points: cloud contract pricing, long-term memory supply agreements, and optical interconnect capacity. Micron's upcoming earnings will test pricing power and cash returns, while the ability to convert aggressive capex into recurring revenue remains the pivotal question.
404K SEMI-AI | September 28, 2026
Market Wrap
During the latest U.S. trading session on September 25, the S&P; 500, Nasdaq 100, and Dow Jones ETFs rose 0.54%, 0.46%, and 0.94%, respectively. The S&P; 500 Equal Weight ETF gained 0.40%, while the Russell 2000 ETF added just 0.11%. While headline indices closed broadly higher, small-cap participation was muted. Investors must distinguish between consolidation in mega-cap tech leaders and broader sector rotation.
Divergence within tech became increasingly pronounced: semiconductor ETFs advanced 1.01%, whereas software ETFs fell 1.06%, cloud computing ETFs dropped 1.34%, and cybersecurity ETFs declined 2.17%. Over the past 20 trading sessions, semiconductor ETFs are up 5.86%, while software ETFs have lost 3.91%. Ongoing hardware infrastructure buildouts and rising competitive pressures on software business models are driving distinctly divergent valuations.
Going forward, the critical question is whether order intake, pricing realization, and paid adoption can justify massive capital investments. Nebius reportedly doubled pricing on new short-term contracts, Samsung Electronics locked in memory capacity via multi-year long-term agreements (LTAs), and Micron is set to report earnings. Concurrently, user growth in agentic AI applications still requires validation through retention and monetization. As tech continues to consolidate and capital rotates into previously oversold sectors, investors should treat this as selective rotation rather than extrapolating localized strength into a broad-based rally.
AI & Semiconductor Value Chain
AI Models, Applications & Capex
OpenAI:
Product direction for its personal AI assistant is becoming clearer. Reports indicate the company aims to make Ultrafast the default mode, though management emphasized this milestone has not yet been achieved. Integration rumors ahead of DevDay still await official product confirmation.
As real-time agents emerge as a dominant interaction paradigm, the demand for ultra-low-latency inference will surge. Cerebras could become a key infrastructure provider in this transition. For now, it should be viewed as an emerging post-launch opportunity, with tracking focused on default settings, adoption barriers, and actual deployment velocity.
Anthropic: Claude Cowork and domain-specific tools are extending competition into end-to-end enterprise workflows, reportedly executing tasks traditionally handled within legal, financial, and enterprise software platforms. The ultimate impact on incumbent software vendors hinges on how deeply agents can penetrate data access, permission controls, and workflow execution. Enterprises anchored by proprietary data, systems of record, or regulatory compliance may retain defensible moats. Key watchpoints include whether corporate customers curtail legacy SaaS subscriptions and whether agents can reliably execute end-to-end tasks.
