AI 眼镜终于卖动了,但不是因为 AIAI Glasses are Finally Selling, But Not Because of AI


AI free token/AI免费积分

本文来源:https://mp.weixin.qq.com/s/Ddxu4p-NBxq_zfXVVbmrsQ
若有版权问题,请来信告知:13951431913#139.com 或在右方提交评论。

2026 上半年全球出货量同比涨了 4 倍,但拆开销量榜你会发现一个尴尬的事实:买的人,多数不是为了 AI。

销量榜上的"反常识"

过去半年,AI 眼镜是消费电子最热的赛道。Meta、雷鸟、Rokid、XREAL、小米、雷朋轮番上新,连带供应链厂商股价翻倍。但如果你去看电商评论区和退货率,会发现一个反常识的现象:被夸最多的功能是"听歌不漏音"和"骨传导通话";被用得最多的 AI 功能是"翻译"和"提词器",不是各家宣传的"视觉 Agent";退货主因是"太重""续航太短""戴着怪",没有一条和 AI 能力有关。也就是说:AI 眼镜的销量,主要由硬件成熟度驱动,而不是 AI 能力驱动。这跟厂商发布会讲的完全是两个故事。图片

真正的拐点是这三件事

销量爆发,靠的不是 GPT-5,而是三个不起眼的工程突破:衍射光波导良率过了 70%——镜片终于可以做轻、做亮、做便宜;硅碳电池进入量产——同样体积电量提升约 30%,让"全天佩戴"成立;端侧推理芯片功耗压到 1W 以内——离线翻译、提词、降噪不再烫脸。这三件事没有一件是 AI 算法突破,全是硬件供应链的活。换句话说:这一波 AI 眼镜红利,本质是光学和电池红利。图片

谁在真正赚到钱

赚到钱的,不是吆喝最响的品牌方,而是三家:光波导模组厂:单模组价格从去年 800 元降到 300 元,但出货量涨了 6 倍;硅碳电池厂:产能被预定到 2027 年;代工厂的"眼镜事业部":传统眼镜厂 + 电子厂的复合团队,今年招聘涨薪 40%。相反,几家主打"AI 原生眼镜"的创业公司,融了不少钱,但都在为毛利率发愁——因为用户不愿意为"AI 多出来的能力"额外付钱。

一个不那么乐观的判断

下半年如果"视觉 Agent 杀手级应用"还出不来,这一波 AI 眼镜会重蹈 2012 年 Google Glass 的覆辙——硬件卖动了,但戴出门的人越来越少,最终沦为"高级蓝牙耳机"。现在最让厂商焦虑的不是竞品,而是:用户戴上之后,每天打开 AI 功能的次数不到 2 次。这个数字如果不能涨到 8 次以上,复购率撑不住。图片

普通人该不该买

三个判断标准:你每天开超过 5 个小时会议,提词和降噪真的能让你效率提升——可以买;你是外语环境工作者,实时翻译能救命——必须买;你只是觉得"AI 眼镜很酷"——等下一代,下半年还会有三次大幅降价。

写在最后

2026 年的 AI 眼镜,像 2010 年的智能手机前夜——硬件准备好了,应用还没准备好。卖动的是眼镜,不是 AI。

 

Source: https://mp.weixin.qq.com/s/Ddxu4p-NBxq_zfXVVbmrsQ. For copyright inquiries, please contact: 13951431913#139.com or submit a comment on the right.

Global shipments in the first half of 2026 increased 4-fold year-over-year, but looking at the sales rankings reveals an embarrassing truth: the people buying them are mostly not doing so for the AI.

Anomaly in the sales rankings

Over the past six months, AI glasses have been the hottest track in consumer electronics. Meta, LeTV, Rokid, XREAL, Xiaomi, and Ray-Ban have been releasing new models one after another, causing supplier stock prices to double. However, if you look at the e-commerce review sections and return rates, you'll find a counter-intuitive phenomenon: the most praised features are "leak-proof listening to music" and "bone conduction calls"; the most used AI features are "translation" and "teleprompter," not the "visual agents" that each company advertises. The main reasons for returns are "too heavy," "too short battery life," and "weird to wear," none of which are related to AI capabilities. In other words, the sales of AI glasses are primarily driven by hardware maturity, not by AI capabilities. This is completely contrary to what the manufacturers are saying at their press conferences.图片

The true turning point lies in these three things

The sales explosion is due not to GPT-5, but to three unpretentious engineering breakthroughs: the yield rate of diffraction waveguides has exceeded 70%—lenses can finally be made lighter, brighter, and cheaper; silicon-carbon batteries have entered mass production—energy capacity has increased by about 30% for the same volume, making "all-day wear" a reality; the power consumption of edge inference chips has been reduced to under 1W—offline translation, prompting, and noise reduction are no longer causing facial heat. None of these three things are AI algorithmic breakthroughs; they are all hardware supply chain advances. In other words, this wave of AI glasses dividends is fundamentally an optical and battery dividend.图片

Who is actually making money?

The ones making money are not the brands with the loudest promotions, but rather three types of companies: waveguide module manufacturers—single module prices have dropped from 800 yuan to 300 yuan, but shipments have increased 6-fold; silicon-carbon battery manufacturers—capacity is booked until 2027; the "eyewear division" of contract manufacturers—composite teams of traditional eyewear and electronics factories have seen wages increase by 40% this year. Conversely, the few startups that tout "AI-native glasses" have raised significant funding but are struggling with profit margins because users are not willing to pay extra for "AI-added capabilities."

A less optimistic judgment

If "visual agent killer apps" do not emerge by the second half of the year, this wave of AI glasses will repeat the fate of Google Glass in 2012—hardware sells, but fewer people wear them out, eventually relegating them to "high-end Bluetooth headsets." The most anxiety-inducing factor for manufacturers is not competition, but the fact that users open AI features less than twice a day. If this frequency cannot increase to above eight times, repeat purchase rates will not be sustainable.图片

Should ordinary people buy?

Three criteria: If you attend more than 5 hours of meetings daily, prompting and noise reduction can genuinely improve your efficiency—buy it; if you work in a foreign language environment and real-time translation is a necessity—definitely buy it; if you just think "AI glasses are cool"—wait for the next generation; there will be significant price drops later this year.

Final thoughts

The AI glasses of 2026 are like the smartphones of 2010—the hardware is ready, but the applications are not. What's selling is the glasses, not the AI.

暂无评论,快来发表第一条评论吧!No comments yet. Be the first!

📮 需求咨询📮 Inquiry