While everyone is focused on AI chips for smartphones and PCs, Ankai Micro has quietly made a significant move in the realm of “glasses”. On September 22, this chip company revealed in its investor relations record that the AI glasses chips that have been taped out can cover three major categories: audio, photography, and display. They are now being promoted on a large scale, with some products already having prototypes! Even more impressively, from last November to now, they have taped out over six chips, with plans for new products using a 12nm process in the future.
As an engineer who deals with chip design daily, let’s delve into this matter: Are AI glasses chips merely a “gimmick” or a “real demand”? Can Ankai Micro carve out a share in the edge AI race? What changes can the 12nm process bring to AI glasses?
1. Understanding First: Why Do AI Glasses Need “Dedicated Chips”? Can’t We Just Use Smartphone Chips?
Many people might ask: Since smartphone chips can run AI, can’t we just install a smartphone chip in AI glasses? The answer is no, because the “pain points” of AI glasses are too specific for smartphone chips to handle.
Let’s look at the three core functions of AI glasses, each of which has “customization requirements” for the chips:
- AI Audio Glasses: For example, smart glasses that can provide real-time translation and noise reduction require chips with “low power consumption + high computing power”—translation needs to run voice recognition AI models, and noise reduction needs to process audio signals. However, the battery life of glasses can only rely on small batteries; if a smartphone chip is used, it might run out of power in half a day, so a chip optimized for power consumption is necessary.
- AI Photography Glasses: Glasses equipped with cameras that can record videos require chips with “fast response + strong image processing”—for instance, if a photo is taken, it needs to be edited in real-time and objects recognized, which requires the chip’s ISP (Image Signal Processor) and AI computing power to work in tandem. Although smartphone chips are powerful, they are too large and consume too much power to fit into lightweight glasses.
- AI Display Glasses: Glasses that can project screens and display navigation require chips with “low latency + good display adaptation”—navigation information needs to be synchronized in real-time without lag, and the resolution and refresh rate of the glasses’ display differ from those of smartphones, necessitating chips that specifically adapt to display protocols. Ordinary chips are prone to “screen tearing”.
In simple terms, AI glasses chips are like “custom suits” that need to fit the demands of “small size, low power consumption, and strong adaptability”; whereas smartphone chips are like “ready-to-wear” clothing that, while universal, are either “too large” or “too power-hungry” when used in glasses. Ankai Micro’s development of three types of chips precisely addresses this “customization pain point”.
2. Six Chips Taped Out + 12nm on the Way: Ankai Micro’s Ambition Goes Beyond Just Numbers
From last November to now, Ankai Micro has taped out over six chips, which is considered “quite fast” in the chip design industry—it’s important to note that it typically takes at least six months to a year for a chip to go from design to tape-out, with costs ranging from hundreds of thousands to millions. Ankai Micro’s rapid tape-out clearly indicates a well-defined “product matrix planning” rather than just “testing the waters”.
What’s even more noteworthy is the “12nm process chip”—currently, Ankai Micro’s AI glasses chips are likely using 28nm or 16nm processes, while 12nm is considered a mid-to-high-end process that can bring two key improvements:
- Power Consumption Halved, Battery Life Doubled: The more advanced the process, the lower the chip’s power consumption. For example, a 28nm chip running an AI model consumes 1W, while a 12nm chip might only need 0.5W. This means that for AI glasses powered by small batteries, battery life could increase from 4 hours to 8 hours, directly addressing “battery anxiety”.
- Increased Computing Power, Capable of Running More Complex AI Models: The 12nm process can integrate more AI computing units (such as NPUs). Previously, a 28nm chip could only run simple voice recognition models, but a 12nm chip might be able to run small-sized image generation models, allowing the glasses to take a photo and generate different styles of edits in real-time.
This reflects Ankai Micro’s “strategic judgment”: edge AI devices are spreading from “smartphones and PCs” to “wearable devices”, with AI glasses being the product with the “most potential to replace some smartphone functions”—for instance, using glasses to view navigation while commuting or answer calls without pulling out a phone; or using glasses for real-time translation and recognizing tourist attractions while traveling, which is more convenient than using a smartphone. Ankai Micro’s intensive tape-out and advanced processes aim to position themselves well before the “explosion of AI glasses”.
3. The AI Glasses Chip Race: Is It a “Blue Ocean” or a “False Demand”? Three Questions Need to Be Clarified
Although Ankai Micro is moving quickly, many still doubt: Can AI glasses really take off? Will the chips be in vain? The key to judging this market lies in three questions:
The first question: Do users really need AI glasses? The answer is “yes, but it must solve ‘scene pain points'”. For example, many people want navigation while driving, and looking down at a phone is unsafe. If AI glasses can project navigation information in front of them, it would be very practical. Similarly, for those who struggle with foreign languages when traveling abroad, wearing AI glasses that can provide real-time translation is much more convenient than holding up a phone. However, if AI glasses only offer features like “playing music and taking photos” without addressing tasks that smartphones cannot do or are inconvenient for, they are unlikely to succeed. The three functions covered by Ankai Micro—”audio, photography, and display”—correspond precisely to practical scenarios like “translation, recording, and navigation”, hitting the pain points.
The second question: Is there a “technical barrier” for AI glasses chips? Will everyone rush to make them? The answer is “there are barriers, and they are not low”. AI glasses chips cannot simply be about “stacking computing power”; they must find a balance between “computing power, power consumption, and size”—for instance, the chip must be small enough to fit into the arms of the glasses, have low enough power consumption for adequate battery life, and still possess enough computing power to run AI models. This requires deep optimization of the chip architecture, not just modifying a generic chip. Ankai Micro has prior technical accumulation in the fields of “IoT chips and automotive-grade chips”, such as low-power design and embedded system optimization, which can be reused in AI glasses chips, forming a barrier.
The third question: When will the industry explode? Is it too early to make chips now? The answer is “soon, 2025-2026 may be a critical juncture”. According to IDC data, in 2024, smart glasses will account for 8% of global wearable device shipments, and this figure is growing at a rate of 20%. More importantly, major tech companies are making moves—Apple’s Vision Pro, while a VR headset, is already exploring “AI interaction in glasses form”; Huawei and Xiaomi have also launched smart glasses products, although the current AI features are still relatively simple. As AI models become “lightweight” (for example, small models that can run on the edge) and chip power consumption decreases, we may see AI glasses that are “practical and affordable” by 2025-2026, leading to an explosion in chip demand. Ankai Micro’s current layout positions them well to catch this wave.
4. Ankai Micro’s “Risks” and “Opportunities”: Don’t Just Focus on Fast Tape-Outs, Avoid These Pitfalls
Ankai Micro has clear opportunities in the AI glasses chip market, but there are several “pits” to avoid:
First, let’s look at the opportunities:
- First-Mover Advantage: Currently, there are not many companies in China specifically making “AI glasses dedicated chips”; most chip companies are still focused on smartphones and PCs. Ankai Micro’s early entry allows them to capture “customer mindshare”—for example, collaborating with glasses manufacturers to develop products, making it harder for other chip companies to enter later and compete for customers.
- Deep Scene Binding: Ankai Micro’s chips cover three core scenarios, allowing for “deep binding” with customers—if a customer wants to create AI translation glasses, Ankai Micro can provide a complete chip solution for “audio processing + AI translation”, eliminating the need for customers to find other chip manufacturers to piece together solutions, which enhances customer loyalty.
Now, let’s look at the risks:
- Many Tape-Outs Do Not Mean “Can Mass Produce, Can Make Money”: Tape-out is just the “first step in chip design”; subsequent steps require “testing, optimization, and customer validation” before mass production. Many chips find that “performance does not meet standards” or “costs are too high” after tape-out, leading to abandonment. While Ankai Micro has taped out six chips, whether 2-3 can successfully reach mass production remains uncertain.
- Market Demand “Uncertainty”: If AI glasses ultimately do not take off—perhaps due to high prices or users feeling they are “unnecessary”—then the chips will not sell, and the initial tape-out investments will be wasted. This is a risk all chip companies face when “laying out ahead of the market”.
- Competition Pressure from Giants: If large companies like Apple and Huawei start developing their own AI glasses chips or seek custom chips from Qualcomm or MediaTek, Ankai Micro, as a small to medium-sized chip company, may find itself at a disadvantage in terms of “computing power, brand, and cost”.
5. In Conclusion: AI Glasses Chips Are the “Next Card for Edge AI”, But Don’t Be Blindly Optimistic
Ankai Micro’s layout for AI glasses chips essentially bets on the “sub-segment scenarios of edge AI”—everyone is talking about “AI everywhere”, but AI will not just stay on smartphones and PCs; it will permeate devices like “glasses, watches, and headphones” that are closer to daily life. AI glasses chips represent the “next card for edge AI”; whoever can balance “power consumption, computing power, and cost” effectively will gain an advantage in this market.
For the average person, it’s not about focusing on the “number of tape-outs”, but rather on the “final product experience”—for instance, can Ankai Micro’s chips enable AI glasses to last over 6 hours? Can they smoothly run real-time translation and image recognition? If these can be achieved, then AI glasses may indeed become the “next commonly used device”; if they can only “run AI but provide a poor experience”, then no matter how advanced the chips are, they will be of no use.
For the industry, Ankai Micro’s actions signal that edge AI chips are evolving from “general-purpose” to “scene-specific”—no longer is it about “one chip fits all devices”, but rather “what device needs what chip, we make that chip”. This kind of “customization for segmented scenarios” may become the “breakthrough path” for small to medium-sized chip companies—after all, they cannot compete with Qualcomm and MediaTek in general-purpose chips, but there are still opportunities in segmented scenarios like “AI glasses and smartwatches”.