2025: Half of the Tech Circle Bets on This Piece of "Glass" | 36Kr's Annual Perspective ③

AI-enabled smart glasses are emerging as a major new consumer hardware category, with 36Kr forecasting global shipments of 5.5 million units in 2025 (up 135% year-on-year) and a potential 90 million-unit market by 2030. Key enablers include domestic supply-chain maturity, sub-50g devices supporting prescription lenses at price points around one thousand yuan, and edge-capable chips (e.g., Qualcomm AR1 Gen1) combining multimodal vision and large models for real-time translation, transcription and scene recognition. Vendors across tech and auto (Huawei, Xiaomi, Baidu, Alibaba, Li Auto, Meta/Ray-Ban, Rokid, etc.) are positioning to capture a next-generation “super terminal” opportunity beyond smartphones, creating a competitive battleground with implications for hardware suppliers and AI software monetization.

Analysis

Market structure: The AI-glasses cycle reallocates value from standalone hardware ASPs to chipmakers, cloud AI providers and optics/sensor suppliers. 5.5M units in 2025 rising to 90M by 2030 implies roughly ~75% CAGR to 2030, favoring semiconductor and AI-service margins while compressing OEM hardware gross margins as prices move to ~1,000 CNY. Cross-asset: expect upward pressure on select semiconductor equities and capex-linked supply chains, modest supportive bias to CNY vs peers, and higher event-driven options vol ahead of major product launches.

Risk assessment: Key tail risks are privacy/regulatory constraints (camera bans, data-localization) and US export controls on AR-capable chips — either could reduce Chinese TAM by 20–40% in the near-term. Time horizons: immediate = headline/regulatory shocks (days), medium = shipment ramp and pilots (3–12 months), long = mainstream adoption and ecosystem lock-in (2026–2030). Hidden dependency: product success depends on on-device LLM latency/bandwidth economics and optical prescription partnerships, not just hardware specs.

Trade implications: Tilt toward chip and AI-service winners (QCOM, BIDU) while being selective on OEMs; expect winners to monetize via recurring services, not one-off hardware. Use relative-value shorts against consumer OEMs that cannot capture software monetization; harvest volatility with calendar/diagonal call spreads around product announcement windows (30–90 days).

Contrarian angles: Consensus prizes display/AR firms; however lightweight auditory and camera-less models could win mass adoption first, keeping AR-display specialists under pressure. The market may underweight privacy backlash and app-ecosystem risk — a failed “killer app” or restrictive rules could shave projected volumes by >30% and re-rate multiples across the device stack.

Key Decisions for Investors

• Establish a 2% long position in QCOM (6–12 month horizon); target +25% upside, stop-loss at -12%. Hedge financing: buy a 12-month LEAPS call ~20% OTM and sell 3-month calls to fund ~50% of premium to capture product-cycle upside while collecting short-term premium.
• Initiate a 1.5% long in BIDU (6–18 months) to capture AI-service monetization from multimodal glasses; increase to 3% if quarterly results show >15% YoY growth in cloud/AI revenue or a confirmed OEM partnership within 90 days. Cut to flat if next quarter misses consensus revenue by >5%.
• Pair trade: go long LI 1% / short BABA 1.5% (3–9 month horizon). Rationale: Li Auto can monetize in-vehicle AR integration; short BABA to hedge platform/retail exposure and margin risk. Exit LI if no product/integration announcement within 120 days; cover BABA if it posts >10% margin expansion.
• Take a small tactical short (0.5–1%) in KRKR over the next 3 months to capture hype decay around annual summaries and product narratives; take profit if the stock falls 20% or tighten stop-loss to +15% if it rallies on unexpected recurring-revenue guidance.
• Condition: pause new >2% long positions in China hardware OEMs until regulatory clarity — specifically, watch for PRC privacy rules or US Commerce announcements over the next 60 days. If rules would effectively restrict camera features or export AR chips and cut addressable China market >25%, reduce QCOM/hardware exposure by 50% within 5 trading days.