Smart Glasses – China’s Advantage

Alibaba plays to its strengths.

• Hate them or love them, smart glasses are rapidly becoming a high-volume product where China has a key advantage, although to leverage that advantage, there remain dependencies on US technology.
• The advent of AI that the user can talk to and not get complete gibberish back has given smart glasses the easy user interface that they previously lacked.
• As a result, the utility of this category has increased meaningfully, and with it, demand has increased materially.
• With its early focus on the Metaverse, Meta was the first mover and remains far and away the market leader, but this category plays directly to China’s strengths in manufacturing and AI efficiency.
• What has impressed me most about China’s AI is not its frontier performance but how much capability Chinese companies can pack into tiny models.
• For example, Alibaba’s Qwen model comes in a 32bn parameter variant that competes effectively with models of its own size and can even give much larger models a run for their money.
• Smaller models have a much smaller memory and processing requirement, and so when one is using them to power a device with a battery, this becomes a key differentiator.
• This is where China’s difficulties play directly to its advantage, as it has been severely constrained on compute despite having plenty of electricity as a result of the export restrictions imposed upon it by the USA.
• As a result, China leads the world in terms of squeezing the maximum amount of performance from small models, which is ideal for a product category like smart glasses.
• Here, Alibaba has announced an update for its Qwen AI Glasses S1 that greatly expands their capabilities and makes the Meta Ray Ban Glasses look a little out of date.
• The biggest update is a proactive function which can surface reminders and relevant information based on one’s surroundings, calendar items and so on.
• This is interesting as it implies that Alibaba has found a way to have an AI running in the background without draining the battery so fast that the device becomes unusable.
• No one else has really managed to do this so far, as all of the demos from the Western players involve manually switching the AI on with a voice command, using it and then switching it off again.
• This will make the Qwen glasses more appealing than their local rivals, but they are unlikely to have any impact overseas.
• This is because the glasses are aimed at the Chinese users who spend their digital lives in the Chinese digital ecosystem, which has no appeal to non-Chinese users.
• However, start-ups may look at this and seek to adapt the Qwen models (which are available in open source) and use them in battery-powered devices aimed at non-Chinese markets.
• I think that this is part of China’s strategy to become a major player in AI, as it is unlikely to be able to catch up with the frontier labs due to its extreme compute constraints.
• However, even these models remain fairly dependent on US technology, as I suspect that Qwen is trained at least in part on US-sourced chips in data centres outside of the Chinese mainland.
• Furthermore, almost every smart glasses product uses Qualcomm’s smart glasses silicon, which makes a large contribution to both the size of the devices and their power efficiency.
• Hence, I think that Alibaba could easily score a large hit in the local market with this product, but whether it will stimulate a recovery in Alibaba’s e-commerce fortunes is another matter entirely.
• This, combined with the integration of AI into its shopping experience, could help it hold onto market share and fend off competition, but the real problem remains domestic consumption.
• This remains at a fairly depressed level, and until it recovers, I doubt whether there will be a meaningful recovery in the share price.
• Alibaba remains one of the cheapest e-commerce plays available, but it also remains a classic value trap.