Shanji Technology Closes Nearly ¥100M Series A+ To Scale Next-Gen AI Glasses
One year after launching its first AI glasses, Shanji Technology (Shenzhen) Co., Ltd. completed a Series A+ round totaling nearly 100 million yuan. The company is moving fast on manufacturing, optics, and an AI memory stack that aims to turn wearables into an active "external brain."
For product leaders, this is a clean look at how to ship a second-generation device: sharpen the core use case, fix the friction, and industrialize personalization.
Funding snapshot
- Round: Series A+
- Amount: ~¥100M
- New investors: Planck Venture Capital, Sky Workshop Venture Capital Fund, Wuhan Lingkonggang Economic & Technological Development Zone (with additional institutions closing)
- Use of funds: personalized flexible production line for AI glasses, computing-in-memory server cluster for active AI memory, global market expansion
Company at a glance
- Founded: February 2020
- HQ: Longhua District, Shenzhen
- Focus: AI glasses, charging and storage, data storage
- Core tech: AI hardware electronics/software, industrial design, and an active AI memory model
- Go-to-market: Dual business model across AI glasses, mobile power, and data solutions
- Next launch: New-generation AI glasses release on December 29, 2025
Product roadmap: where the value shifts
Shanji's second-gen device centers on two moves: all-day, active AI memory and hardware that wears like normal eyewear. The company partnered with memories.ai to bring a multimodal AI memory engine into smart glasses for the first time.
- Active AI memory: Real-time audio and video are converted into structured, retrievable memory. Paired with Loom OS 2.0, the system supports multi-device sync and orchestration across more than a dozen mainstream Chinese large models for flexible AI calls.
- From passive to proactive: Self-developed voice recognition and VAD automatically trigger recording for meaningful audio-environmental conversations, calls, and even Bluetooth audio like music and videos-building an all-day "life memory library."
- Practical outputs: Automatic diaries, weekly reports, personal biography drafts, past-event queries, and to-do reminders.
Optics, weight, and thermals
- Display engine: Co-developed with Fengmang Technology using its Dragonfly Optic Engine. It lowers full-screen power consumption and supports a flexible front frame with a binocular 3D display structure. The tech lineage is similar to what powers premium AI display glasses on the market.
- Silicon and radio: 4-nm main control chip plus a flagship-grade Bluetooth chip. Software-hardware co-optimization targets visual performance and battery life.
- Materials and fit: Polymerized low-density material machined via five-axis CNC fused with titanium alloy. Building on a first-gen baseline around 50 g, the new design aims even lighter while preserving structural strength for all-day comfort.
- Battery strategy: Optimized dual-chip architecture with a replaceable battery path designed for up to 15 hours of continuous wearing comfort.
Personalized manufacturing
To solve the "glasses must fit first" problem, Shanji is standing up a personalized, flexible line. Over the next year it plans to ship at least 10 frame types, each with three sizes. CNC machining plus titanium replaces traditional injection molding to improve texture and color options, with explicit attention to styles preferred by segmented groups such as female users.
Market reality and opportunity
Per IDC's smart glasses tracking, the first half of 2025 saw 4.065 million global units, up 64.2% year-over-year. China shipped over 1 million units (26.6%), with 2025 shipments projected at ~2.907 million and a strong multi-year growth path. Source: IDC.
That said, the domestic market underperformed expectations in 2025: fewer than 100 models launched in China, with only a dozen or so mainstream. Roughly one million units shipped across the year. By comparison, Meta shipped about 2 million units last year and is expected to pass 4 million this year.
Why many first-gen products missed
- Form factor mismatch: Sunglasses-style designs work for short outdoor sessions (2-3 hours). Most domestic buyers need optical glasses for myopia, presbyopia, and astigmatism-worn all day. Comfort becomes the gatekeeper.
- Low functional delta vs. phones: Navigation accuracy and latency trail phones, AI Q&A feels better on phones, and teleprompters are niche. If core value isn't better than a phone, repeat usage stalls.
Business momentum
- Shipments target: >100,000 units within one year after the new model's release
- Cash engine: Mature 3C energy storage (mobile power, storage devices) contributes hundreds of millions of yuan in revenue and tens of millions in net profit annually
- Runway: Company maintains hundreds of millions of yuan in available funds to support R&D and market expansion
Founder's product stance: glasses first, AI second
Zhang Bo frames the strategy as "external brain" plus "digital continuity," but with one non-negotiable: it must feel like normal glasses. Only then do features like photo, video, AI Q&A, and music make sense in daily life.
The lesson for product teams is simple: earn the right to add features by removing the reasons people take the device off their face-weight, heat, fit, and social acceptability. After that, focus on the few functions that a phone can't replace or handles poorly when heads-up and hands-free.
Practical takeaways for product leads
- Design for all-day wear: Weight, balance, and thermal management beat raw spec sheets.
- Prioritize value over novelty: Active memory, recall, and automated summaries map directly to real work and life tasks.
- Personalization drives adoption: Multiple frame families and sizes reduce try-and-fail returns and increase daily wear rates.
- Model orchestration over model worship: Make multi-model collaboration invisible to the user; optimize for outcome, not the badge on the model.
- Manufacture for variety: Flexible lines and CNC+metal hybrids let you ship aesthetics without sacrificing structure.
- Build trust by default: For always-on audio features, ship clear consent, recording indicators, and on-device controls. Friction here reduces bigger risks later.
What to watch next
- December 29, 2025 launch and first-quarter user retention on active memory features
- Fulfillment of the >100,000 unit shipment goal within 12 months
- Yield and lead times on the personalized frame line
- Real-world power savings from the Dragonfly Optic Engine paired with the 4-nm platform
- How Loom OS 2.0 balances multi-model routing with battery life and responsiveness
If your team is planning an AI wearables roadmap or upskilling for AI-driven product work, explore role-based learning paths here: Complete AI Training - Courses by Job.
Your membership also unlocks:
