Summary

夏勇峰 is not betting on “a cooler pair of glasses,” but on the next personal terminal that technology spillover will seek once smartphones mature. The smartphone replacement cycle has stretched from an average of roughly 8 months at its peak to nearly 36 months, leaving upstream technology with a shrinking “pond” and an inevitable need for new hardware hosts; his view is that head-worn devices will sit closer to people, with human sensors and the “computing center” concentrated in the head, making head-worn wearables the eventual opportunity.
蜂巢’s most consequential strategic choice was to abandon the camera after completing a first-person camera glasses product, and first solve the basics of looking good, feeling comfortable and being wearable all day. 夏勇峰 believes the industry cannot tackle the technical challenge and educate the market at the same time: sunglasses must weigh no more than roughly 50 grams, while optical glasses should be above 20 grams but lighter is always better; until smart functions can “make him an extra RMB10,000 a month,” consumers will not pay for ugly, heavy glasses that hurt their noses. “First-person shooting has nothing to do with AI,” and Ray-Ban Meta’s success has not changed his view.
界环 is explicitly targeting people who already wear glasses, using observed behavior from more than 150,000 users to test the narrow path of “glasses replacing part of the earbud.” 夏勇峰 even suggests that people with no habit of wearing glasses or sunglasses “can probably skip this product”; the host focused on battery anxiety caused by 12-hour endurance, and he replied that running out before getting home represents heavy usage that more than 90%, even 95%, of users will never encounter, while offering the classic answer half-seriously: “Buy another pair.”
The more meaningful commercialization signal today is channel penetration, not feature stacking: 界环 entered 宝岛 at the end of September 2024, then expanded into 博士眼镜, Sam’s Club and other systems, reaching more than 800 stores in roughly half a year. The pace is more than 100 stores a month, supported by the traditional eyewear industry’s strategic backing for upgrading and by channel economics somewhat better than those of ordinary consumer electronics—but still “far below” the margins of traditional eyewear. 夏勇峰 is not worried about channels developing their own products: while the industry is still unprofitable, he believes a startup focused on one thing can “outlast them.”
蜂巢 is trying to turn its first-mover awareness into full-stack R&D, patents and a frames ecosystem, rather than merely leading by one product cycle. 夏勇峰 estimates the company is 3–6 months ahead, perhaps 12 months ahead of most others; it uses no off-the-shelf solutions, develops its hardware and software in-house, has filed more than 300 patents—including 11 for the replaceable hinge alone—and makes roughly 100 third-party frames compatible with the same electronics module. He puts the current maturity of the audio-glasses industry at 20–30; if the EV supply chain was 100 several years ago, later entrants cannot simply skip the trial and error.
The so-called “battle of 100 glasses” currently looks more like more than 100 companies copying Ray-Ban Meta than an effective competitive market. The program opened by noting that Ray-Ban Meta sales approached 2M pairs last year, but 夏勇峰 says capital has not been as generous as expected, and many products originally planned for launch in the first half of 2025 will be pushed into the second half; at the peak, there could be at least 3 launches a week, followed by “100 companies becoming 50 after a month, then 20 after another month.” He remains convinced China is not a sunglasses market, while allowing for internet memes to trigger herd buying: “Then it’s no longer about product logic.”
夏勇峰’s defining measure of success is earning a sustainable 1 yuan through normal growth. He says 100% of smart-glasses projects over the past 20 years have lost money; if 蜂巢 becomes sustainably profitable, “if I make 1 yuan, I’ll be the first in the world,” but he will not manufacture profits by “laying off half the staff first.” The longer route runs from AI audio glasses to every pair of glasses carrying AI by default, then back to AR and digital layers over the real world; the next step could be brain-computer interfaces replacing glasses—his “all in for the second half of my life.”

Deep dive

1. Smartphone replacement cycles stretch from 8 months to 36, forcing technology to find a new host

夏勇峰 moved from the media and Geek Park into Xiaomi, first working on routers and then helping found the ecosystem chain. He transferred to the smartphone division in 2018; the last product he completed in depth was the Mi 10 Ultra. By the time he left Xiaomi to found 蜂巢, he had already decided to focus on smart glasses.
His starting point was the smartphone industry’s shift from growth to maturity: at the peak, the average person replaced a phone roughly every 8 months; later, the cycle approached 36 months. As performance, quality and durability stabilized, a massive upstream technology base faced a gradually shrinking pond.
夏勇峰 sees the resulting “technology spillover” as a structural trend: every technology and the companies behind it need to keep reproducing and expanding, and the next host may ultimately be head-worn hardware. “Head-worn devices will be closer to people,” while all of the body’s sensors and its computing center are concentrated in the head.

2. Smart glasses must first clear the pass mark for ordinary glasses

蜂巢 did not choose earbuds; when asked about earbuds, hats or masks, 夏勇峰 said he had “never considered earbuds.” He wanted the new terminal to interact with the eyes from day one, because the eyes connect the physical and digital worlds. The initial answer was first-person camera glasses capable of showing a live view in front of the wearer.
Only after completing the 米家眼镜相机 did he realize that the industry could not solve two challenges simultaneously: fitting technology into glasses that were light and attractive enough, while also teaching the public to understand a new function for which it had no existing concept. A product could be well finished and still, “apart from being of little use to most people,” serve only vertical users.
The physical thresholds are concrete: total sunglasses weight must stay below roughly 50 grams; optical glasses should be above 20 grams but lighter is always better. Consumers anchor on ordinary frames, and even the coolest metaverse concept can struggle to offset bulk, nose pain, ear pain, eye strain and headaches.
夏勇峰 uses an intentionally exaggerated benchmark to show how little value the product currently offers: “For example, make him an extra RMB10,000 a month.” Only that might persuade users to tolerate ugliness and discomfort. Since smart capabilities are nowhere near that level, he decided to clear the market and wearability hurdles first, then layer in technology step by step.

3. Ray-Ban Meta’s success did not shake the decision to leave out the camera

The host pressed the obvious counterexample: after seeing Ray-Ban Meta succeed, did he ever doubt the audio route? 夏勇峰’s answer was “no,” because China is not a sunglasses market; the choice before him was between Western smart sunglasses and Chinese smart optical glasses.
The judgment also came from firsthand failure. Before the first-generation Ray-Ban Stories met expectations, 蜂巢 had already built camera glasses and understood how first-person shooting consumed power, weight, system resources and R&D capital.
夏勇峰 draws a sharp line between the concepts: “First-person shooting has nothing to do with AI.” First-person shooting is Ray-Ban Meta’s core function today, but he believes future AI glasses may involve a camera without fully overlapping with this kind of first-person capture.
His conclusion is categorical: “Others may build them, but I won’t build shooting glasses.” 蜂巢 would use audio to carry AI first, while building its self-described Voice UI interaction framework underneath, rather than changing course to match the industry’s most visible benchmark.

4. 界环 serves existing glasses users rather than creating a wearing habit

夏勇峰 offers an unusually direct deterrent to prospective users: “If you don’t normally wear glasses, I think you can probably skip this product.” The current feature set cannot turn someone who never wears glasses or sunglasses into an all-day wearer.
For myopic users, the value is avoiding another electronic device: the temples provide earbud functionality at any time, maintain a persistent connection, and are easier to wear while driving or running outdoors. The product changes the device mix within an existing wearing habit rather than creating demand from scratch.
The new generation of 界环 weighs roughly 27 grams, already close to traditional glasses weight; 夏勇峰 says the company has spent “tens of millions of yuan” on appearance and comfort. Good looks and comfort are T0-level features, because no powerful capability can save a product people do not want to wear outside or that starts hurting after 2 hours.
He himself has not used earbuds for a long time: frequent online meetings once gave him headaches and forced him to stop, and after looking into it he believed he might already be showing signs of mild hearing decline. That experience explains why the team puts open-ear audio and long-term wearability ahead of sound-quality specifications.

5. Is 12-hour battery life enough? The host and 150,000 users give different answers

The host’s experience is that once the glasses connect automatically at 7 a.m., they begin generating anxiety: advertised battery life of roughly 12 hours means the sound may disappear at 7 p.m.; AirPods go back into their case to charge when unused, while all-day glasses have no equally natural replenishment action.
Citing usage observations from more than 150,000 users, 夏勇峰 responds that if the glasses are dead before the wearer gets home, that generally represents heavy use encountered by no more than 90%, even 95%, of users; most people use them for at most around 10 hours a day, and actual endurance is longer when they mainly listen to audio and use the microphone less.
For genuine heavy users, his “classic advice” is: “Buy another pair, and you won’t have battery anxiety.” Wearing one pair of glasses and carrying another glasses case is perfectly normal; the answer acknowledges that seamless all-day endurance has not been fully solved and treats two-device rotation as the current product boundary.

6. The audio route is not simple subtraction, but a reprioritization of technology

Moving from camera glasses to audio glasses was not as simple as deleting the camera: the former had no speaker, while the latter added one, requiring resources to be reallocated around audio. What was truly redefined was what those resources should serve first.
Once appearance and comfort were satisfied, the team believed only two technical routes had a chance: Micro LED with waveguides for near-eye display, or an audio solution based on speakers that could replace part of the earbud. Internally, people repeatedly defended the display route because it was “definitely cooler,” but user research kept pushing the answer back toward audio.
蜂巢 also did not adopt the left-right wired architecture used by companies such as Bose and Amazon. Instead, it developed hardware with independent left and right sides based on mature TWS designs, aiming for longer battery life, replaceable frames and more frame shapes at once.
R&D spending went into unglamorous details that determine the experience: batteries cost more than RMB10M; hinge prototypes were made more than 10 times, with every iteration requiring repeated insertion and removal followed by roughly 1–2 weeks of review. 夏勇峰 says the company has spent heavily on batteries, hinges and other product details.

7. The hinge turns an aesthetic debate into a scalable interface ecosystem

夏勇峰 does not try to define beauty through personal taste. He first studied the eyewear industry’s existing categories and history—why aviator, Boston and browline frames emerged, what trends they corresponded to, and which styles endured—then looked for details that could be improved across industries.
The team brings technologies used in Apple products, Xiaomi phones and other smart products into the frames, then uses extensive user testing to calibrate them. He does not think 蜂巢 is making fashion, but believes trends may emerge naturally after “working hard to make it beautiful.”
Diversity requires decoupling the electronics from the frame; otherwise every frame shape would require a new hardware development cycle, which for a consumer-electronics product is “a dead end.” The replaceable hinge therefore serves as a standard interface: the electronic temples remain standardized, while frame manufacturers and designers only need to design around the interface.
The hinge alone has more than 11 patents, and 夏勇峰 says some peers produced very poor products while trying to work around them. In addition to official frames, at least roughly 100 frame styles that are not 蜂巢-branded are currently compatible with the system. Beauty comes from the number of choices, not the endorsement of a single star designer.

8. More than 800 stores show that traditional eyewear is willing to embrace an upgrade

界环 began with 宝岛 at the end of September 2024, then entered 博士眼镜, Sam’s Club and several major chains. In roughly half a year, offline distribution exceeded 800 stores—more than 100 new stores a month, faster than 夏勇峰 expected.
Before the AI-glasses concept took off, he relied on friends’ introductions, WeChat messages and in-person visits, often receiving no reply. It felt like reliving the rejection he encountered as a reporter more than a decade ago. “History repeating itself” may be too dramatic, but he did have to rebuild his mental readiness.
The turning point was traditional eyewear’s awareness that the industry was upgrading: strong chains did not want to miss a structural shift in an industry they had operated for decades, so they sought products and teams that were mature, clear in their thinking and capable of long-term cooperation with traditional eyewear businesses. Effective support had to start with owners inspecting and trusting the product, then pushing it from the top down.
Commercial “friendliness” means leaving channels somewhat more room than ordinary consumer electronics, but still far less than the margins of traditional eyewear itself. 蜂巢 has no investment from traditional eyewear giants and does not accept the traditional eyewear profit-sharing model; the partnership rests on strategic interests rather than existing capital ties.

9. If channels enter the market themselves, 蜂巢 is betting it can endure longer

The host worries that eyewear chains are treating 界环 as a “guinea pig”: use its product and pricing to validate demand, then build their own product if sales are good. 夏勇峰 is completely unconcerned: “If everyone wants to do it, go for it. I support everyone doing it themselves, and then let the products speak.”
His confidence does not come from claiming traditional companies cannot build the product, but from believing profitability will not come quickly. If both sides lose money in the first and second years, a startup betting on one thing is more likely to make it to year 3; a multi-business channel operator may be less willing to absorb losses for that long.
This also explains why 蜂巢 must invest before the market matures. 夏勇峰 believes he is on the right path and that mass-market smart glasses will begin with products like this. “When nobody understands it, you’ve already run 1,000 meters ahead”—investment can then become a time-based moat.

10. A 3–12-month lead in awareness must be converted through full-stack R&D

夏勇峰 describes the company’s cognitive lead cautiously: he estimates it at 3–6 months, perhaps 12 months ahead of more people, not an absolute lead. The value is that the team has already validated tradeoffs between sound quality and comfort, and between call noise cancellation and size, while new entrants still have to reopen every debate.
蜂巢 did not source generic hardware from a solution provider and wrap it in a new ID layer. Apart from the hinge and battery, the overall architecture, including hardware, and the software are fully self-developed, with more than 300 patent applications filed. 夏勇峰 believes that without a mature supply chain, a small startup that cuts this R&D spending will struggle to reach the same level of finish.
His analogy is that if the EV industry’s maturity was 100 several years ago, audio glasses or AI audio glasses today are probably only at 20–30. The supply chain, products, patents and channels are not yet standardized; later entrants cannot obtain a mature product simply by “making some software and putting a skin on it.”
蜂巢’s moat is therefore not a single piece of black-box technology, but a chain of decisions that cannot easily be skipped: frames, hinges, batteries, noise cancellation, wearability, software and eyewear channels. Only once 界环’s route is widely recognized as successful can copycats reduce the upfront cost of understanding the problem.

11. Eyewear, consumer electronics and the internet all see this as their battlefield

After launching 界环, 夏勇峰 spoke with at least 50% more people than he had when working on other products. He has exchanged views with almost every internet company, because 3 major industries claim the product at once: eyewear says “these are our glasses,” consumer electronics says they are smart hardware, and the internet says “this is AI.”
His view of control depends on the time horizon: the initial phase must begin with eyewear; “at the very end,” if a successful Top 1 emerges, he believes it will definitely be a consumer-electronics company. The 50-year figure is only a proxy, since he cannot even tell how many years the initial phase will last.
The host questioned why phones and computers did not take 50 years to complete an industry transition. 夏勇峰 pointed to 毛源昌眼镜, a Chinese time-honored eyewear brand with roughly 160 years of history: eyewear is inherently a long-cycle industry, and its evolution cannot be mapped directly onto new electronic categories.
After seeing an unreleased 蜂巢 product, a veteran of the traditional eyewear industry said it was “the first smart glasses among so many that made me feel a little scared,” because its iteration speed might accelerate the replacement of traditional optical glasses. 夏勇峰 admitted the comment exceeded his expectations, but said the happiness was “roughly like a child scoring 90 in all 3 subjects.”

12. The uniformity of the “100-glasses war” exposes a lack of original thinking

The market signal given at the start of the program was that Ray-Ban Meta smart-glasses sales had approached 2M pairs the previous year. 夏勇峰’s “more than 100 companies” is a narrower count: more than 100 projects copying the Ray-Ban Meta form.
The template connects several narratives at once: it is eyewear, consumer electronics with a camera, a vehicle for first-person videos on TikTok and Instagram in the US, and an AI product. Teams of every kind can therefore easily feel that it belongs to them.
夏勇峰 believes most projects have not thought through the user, market and product. They first use Ray-Ban Meta’s success to validate the direction, then use elimination to discard other eyewear forms that failed, ultimately forcing everyone onto the same route.
For China, he remains convinced that sunglasses-wearing habits are insufficient and that copying overseas benchmarks has a fundamental problem. But he explicitly leaves room for being wrong: an internetized society may create herd effects around a meme and suddenly make a product popular. “That’s no longer product logic.”

13. Capital did not take over according to script, pushing the real shakeout into the second half

The original financing narrative was that capital would move from large models to agents and applications, then to AI hardware; Ray-Ban Meta was theorized as the best hardware product, so AI glasses would naturally be next. But according to 夏勇峰, fewer than 80 of the 100 projects successfully raised enough money.
Investors gradually realized that founders might not have thought the product through and that glasses cannot be built as a casual side project. Both the pace and size of financing are weaker than expected 6 months earlier. 夏勇峰 believes these products are generally not cheap to build, because insufficient investment produces an insufficient product.
At the same time, many teams “fell into the glasses trap as expected,” discovering that the difficulty was far greater than that of ordinary hardware. A group of products originally planned for launch in the first half of 2025 will probably move to the second half, so the market has not yet reached its genuinely noisy phase.
夏勇峰 expects the peak could bring at least 3 launch events a week, followed by rapid elimination: “100 companies could become 50 after a month, then 20 after another month, and after another month the game is over.” Failure could mean no one buys the product, a broken funding chain, or even some founders becoming defaulters.

14. From reporter to Xiaomi to his own startup, he has always chosen to reduce the next decision

Before joining Xiaomi, 夏勇峰 had been invited every Lunar New Year for 2–3 consecutive years by another internet tycoon to build hardware. After 阿黎 learned about it, he spoke with 雷总 for roughly 8 hours and received 3 options: invest in TMT through 顺为, become a product manager for Xiaomi TV, or become a product manager for Xiaomi routers. He chose routers.
The router team initially had no other product manager, leaving him to coordinate with roughly 8 technical teams. Before the first product was finished, he became the ecosystem chain’s “No. 1 employee.” He joined in 2013 as a product manager and became a department vice president by 2016, attributing the speed to “the times and luck,” as well as to a rapidly expanding company that kept creating startup-like openings.
He does not view a possible 蜂巢 failure as unacceptable. A first startup has a high probability of failure; if a wrong judgment closes the company, he would suffer but could bear it. He would then correct course along a similar direction, because he believes the odds of success rise on a second attempt.
His decision model comes from Texas thinking: “Don’t do things that make the situation more complicated.” Saying no eliminates 80%–90% of the possibilities. When he left Xiaomi, he rejected becoming a partner at a friend’s company because unknown information and uncertain expectations about his capabilities would make the next step more complicated; he chose to start from zero.

15. All in for the second half of his life; the first milestone is simply earning 1 yuan healthily

夏勇峰 believes “you can’t do many things in a year, and you can’t do many things in a lifetime,” so he sets very few goals for work and life. The smart-glasses cycle is so long that it cannot be completed without going all in, but if he finishes earlier than expected, he will not “cling on desperately and keep going.”
The route begins with AI: in the future, “AI glasses” may no longer exist as a separate category because every pair of glasses will carry AI by default. The path then returns to AR, seamless switching and digital layers over the user’s field of attention, moving toward Iron Man and Ready Player One-style experiences. The next step could be brain-computer interfaces replacing glasses, but that is not what he plans to build.
If he had more time, he would like to return to journalism. When he interviewed companies in the past, he felt that the subjects were “standing on bricks” while the reporter was “standing on cotton.” After gaining experience in products, investment and supply chains, he hopes to be able to say, “Don’t even think about fooling me,” while keeping material security separate from his ideals as a journalist.
His honest answer on startup success is that “desire is hard to satisfy.” He says 100% of smart-glasses projects over the past 20 years have lost money; the program also repeated a claim that the money spent on smart glasses in the US would have been enough for Oppenheimer to build the atomic bomb 10 times. If 蜂巢 becomes profitable while developing healthily, “if I make 1 yuan, I’ll be the first in the world,” because that would mean the model had finally turned positive.
That 1 yuan cannot come from short-term financial engineering: “I won’t lay off half the staff first just to become profitable.” What he wants is a sustainable positive feedback loop during expansion. Faced with an extremely high ceiling and an extremely low historical win rate, his method is to reduce anxiety by “comparing myself with people who are worse than me,” then continue to “follow the wind.”

Energy Singularity’s Yang Zhao on Taming Fusion: A 3-Hour Interview Fiverr Goes All-In on AI: Empowering Creators, Not Replacing Them, with Micha Kaufman, CEO of Fiverr