Exclusive interview with Dong Hongguang of Guangfan (光帆): AI hardware's universal solution can't rely on startups acting alone

Startups first, but not alone

Dong Hongguang (董红光), founder and former early employee of Xiaomi (小米), tells TMTPost that the next wave of AI hardware cannot be won by startups acting in isolation. Guangfan (光帆科技), barely more than a year old and reportedly sitting on nearly RMB 300 million in seed funding, is pitching an audacious thesis: the ideal carrier for persistent, agent‑style AI won’t be the smartphone but a multi‑device wearable system. The company launched an earbud-plus-watch pair in mid‑May and has built a native AI OS to coordinate cloud models with always‑on edge sensing. Why earbuds and watches? Because they can listen, see and stay with the user — something phones, by design, rarely do.

Product logic and system bets

Guangfan defines its kit as an “AI full‑sensing wearable” or a “second main device”: the earbud includes a camera and microphone, the watch displays UI and shares sensing, and the charger case holds an eSIM so both items can work independently of a phone. The startup says its OS is purpose‑built for multimodal, natural‑language interaction using a cloud+edge architecture — cloud models as the brain, the devices as perceptual limbs. Dong argues the category decision was driven backward from the assistant concept, not forward from product templates: “We aren’t an earbud company; we are building an AI assistant and then choosing the hardware to support it.”

Ecosystem limits and geopolitical backdrop

Dong is frank about the constraints: sensors, chips and power limits force painful engineering tradeoffs, and the supply chain is long. It has been reported that Guangfan’s seed round includes industry funds such as CATL (宁德时代) and Shokz (韶音), underscoring his point that startups need industrial partners. Counterpoint data cited by the company projects wearable devices as a trillion‑dollar cumulative opportunity from 2026–2032, with edge‑AI wearables accounting for roughly 75% of that value. Yet access to advanced semiconductors and sensors is increasingly shaped by global export controls and geopolitical friction — a reality that could bottleneck ambitious hardware plays regardless of product design.

A cautious sprint toward scale

Price, retention and software economics matter more than headline sales, Dong says. The earbud+watch launched at RMB 1,899 — reportedly close to cost — and Guangfan is focussing on user retention and iterative UX rather than subsidised volume. Big tech moving into the space is not necessarily a death knell; it validates the category and can accelerate supply‑chain readiness. But who will define the eventual form factor — Apple (苹果), a major OEM, or a cluster of niche innovators like Guangfan? Dong’s bet is that the answer will be plural: the path to a universal AI hardware solution will require startups, industrial partners and larger players working in concert, not a lone heroic founder.