Risks, Falsification, and Actions

Warning Pool

• Any Serenity-mentioned A/H stock that gaps before primary evidence appears. 21jingji's 绿的谐波 report is a case study in social heat causing immediate price action.
• SIVE if the thesis is reduced to "Serenity likes it" or if market cap prices in 20-30x revenue growth before volume revenue is visible.
• Generic storage names after broker target hikes and media stories about sold-out HDDs. Real shortage, but discovery_gap is shrinking.
• AI data-center power names with only headline GW/MW claims and no interconnection, PPA, permitting, or active-load evidence.
• CXL names where the product is technically interesting but no design win, attach-rate, or ASP/margin evidence exists.

Watch Terms To Add Next Week

• "external light source" / "ELS"
• "CW DFB laser"
• "InP laser array"
• "1.6T LRO"
• "co-packaged optics" / "CPO"
• "linear pluggable optics" / "LPO"
• "NVLink Fusion"
• "UALink"
• "AI XPV"
• "contracted power"
• "active power"
• "time-to-power"
• "utility power cannot be sourced"
• "CXL KV cache"
• "KV cache server"
• "memory wall inference"
• "capacity access rights"
• "reference design"
• "design-in"

Watch Terms To De-Emphasize Or Tighten

• Generic "AI storage" unless paired with HDD exabyte demand, nearline lead times, controller attach, or explicit enterprise allocation.
• Generic "AI power" unless paired with interconnection delays, cancelled projects, onsite generation, or equipment backlog.
• Generic "AI photonics" unless paired with ELS, InP, DFB, CPO/LPO, customer design-in, or capacity allocation.

Scoring Guidance For Next Radar Run

• SIVE: bottleneck_score high, consensus_priced_score now at least 3.5-4, discovery_gap lower than early March. Label: watch_for_pullback or observation_high_heat, not core until product revenue and oligopoly capture are independently verified.
• STX/WDC: bottleneck real, but consensus score has risen. Label remains watch_for_pullback; do not upgrade to underdiscovered.
• SNDK/SMI: NAND/controller beneficiaries, but commodity-cycle and allocation dynamics matter. Label cycle_beta unless controller monopoly is proven.
• PENG/CXL: evidence velocity improving; monopoly score unclear. Label execution_option pending CXL ecosystem capture analysis.
• MRVL/AVGO: platform winners, increasingly consensus. Use them as demand/architecture signals; core inclusion requires valuation discipline.
• Power / electrical infra: DCD cancellation strengthens the time-to-power thesis. Raise evidence_velocity for onsite power and electrical equipment; add cancelled/paused project tracking as a primary signal.

Final reminder: Big-account commentary is only a strategy-sample generator. A stock can enter the main radar only after independent monopoly / oligopoly potential reasoning proves that the company captures an unavoidable bottleneck layer.