In China, companies are already shipping hardware with built-in artificial intelligence
Technology and artificial intelligence have been walking hand in hand for a while now, but what is happening in Hangzhou, China, is a whole different ball game.
While most of the world was still debating which language model was better or which cloud platform delivered the most, Chinese startups were already boxing up hardware and shipping it out the door.
No exaggeration.
This is a real turning point, and it is happening right now, blending AI with physical devices in a way that completely reshapes the global tech race.
From 43-dollar clip-on microphones with built-in AI agents to humanoid robots that take voice commands, China is answering in practice a question the rest of the world is still trying to answer on paper:
What can artificial intelligence do when it leaves the cloud and enters the real world?
And there is more, because it is not just startups in this race.
Giants like Alibaba and automakers like Volkswagen are also placing serious bets on hardware with AI, each with their own strategy and their own data as a competitive edge.
The landscape taking shape in cities like Hangzhou and Suzhou shows that the next battle in technology will not be won by whoever has the most powerful model, but by whoever manages to put that model inside every everyday device. 🤖
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
EinClaw and the 43-dollar microphone already running an AI agent
While major Chinese cloud companies were racing to promote OpenClaw in early March, a Hangzhou startup called EinClaw was already at a completely different stage: building physical devices. The company shipped its first 100 units of a 43-dollar clip-on microphone that lets users send voice commands to an OpenClaw AI agent. And the most impressive detail is that only two people developed and assembled the device, using parts sourced from suppliers scattered across China, as co-founder Arvin Chen explained.
This kind of product shows that the barrier to entry for creating hardware with embedded artificial intelligence is dropping fast. We are not talking about a mega-corporation with thousands of engineers and billion-dollar budgets. We are talking about two people in a WeWork-style office in Hangzhou who managed to design, assemble, and ship a functional device that puts advanced AI in anyone’s hands for under 50 dollars. That completely changes the math on who can compete in this market and how fast new products can emerge.
The concept behind the product is simple but effective: you clip the microphone to your clothing, say what you need, and the AI agent processes the command. Whether it is transcribing a meeting, pulling up information, or executing an automated task, the device works as a bridge between the human voice and the processing power of a language model. It is edge computing in action, with part of the processing happening right on the device, which cuts down the need for a constant connection to remote servers. For anyone working in environments where internet connectivity is spotty or where data privacy is a priority, this kind of solution makes all the difference. 💡
Humanoid robots with OpenClaw and the leap into the physical world
OpenClaw features are also making their way into robots. In Suzhou, a city near Hangzhou, the startup JoyIn says its Zeroth M1 humanoid is the first robot to integrate OpenClaw. Using Tencent Cloud tools, people can send commands to the robot and control it remotely. Pre-orders are expected to begin by July, according to the company.
This kind of development shows that Chinese robotics is not just keeping pace with the evolution of language models — it is absorbing that evolution directly into its products. The Zeroth M1 is not a robot that simply follows pre-programmed instructions. It receives commands in natural language, interprets what was asked, and acts in the physical world based on that interpretation. The difference between that and a traditional robot is massive, because it opens up use cases in contexts that would have been impractical before, such as warehouse logistics, service in public spaces, and home care assistance.
The role of AI in these robots goes far beyond recognizing a voice or following a pre-programmed path. The most advanced systems can already adapt their behavior based on the surrounding environment, make real-time decisions when facing unexpected obstacles, and even learn from mistakes made in previous operating cycles. This is possible because the models running on these devices were trained with massive volumes of physical data, from images captured by cameras at multiple angles to force and torque signals recorded during manual tasks performed by humans. All of that data becomes fuel for the robot to understand the world in a way that is more similar to how a human understands it, even though there are still important gaps the industry keeps working to close. 🏭
Is cloud-native outdated? What the OpenPie founder says
Together, all of these moves point to a shift that cuts across entire industries — from AI that only exists on the internet to AI that lives inside physical devices. And the person saying this is not an outside observer, but someone who built his career on the very model he now considers outdated.
Ray Von, founder, CEO, and chairman of OpenPie — a Tencent-backed startup created to build cloud data systems — was blunt: Cloud-native is a bit outdated. The technology is useful, but the business model is a bit outdated. Data sovereignty is now a concern.
That statement carries enormous weight. Coming from someone who founded a company specifically to work with cloud data, it signals that the market is concretely shifting direction. The question of data sovereignty is especially relevant in the context of China’s millions of factories. While these manufacturers are interested in unlocking the efficiencies of artificial intelligence, they are worried about sending proprietary information to the cloud, as Von explained.
The solution OpenPie is building tackles that problem head-on: devices that let you run AI tools locally, using low-cost Chinese chips. The goal is to ship 10,000 of these boxes by the end of the year, at a price of 100,000 yuan (roughly 14,627 dollars) per unit, before scaling up production. It is a clear bet on hardware as the vehicle for democratizing access to AI without giving up control over the data.
Style3D, texture data, and the bridge between AI and the real world
The expansion into the physical world is also transforming companies that were born software-first. Style3D, for example, started in 2015 using artificial intelligence to help apparel companies speed up the process from design to production. As CEO Eric Liu explained during a venture capital conference in Hangzhou, so many companies asked Style3D for access to its data on physical materials and textures that the company decided to enter that business directly, launching the SynReal robotics platform last fall.
The logic is fascinating: to work well in the real world, humanoid robots will need a massive amount of specialized information about textures, so they can grab items ranging from oranges to silk scarves. And that data is exactly what Style3D has accumulated over nearly a decade of working with the fashion industry. It is a perfect example of how specialized data becomes a strategic asset when AI moves out of software and into the world of physical objects.
Style3D’s transition illustrates a larger pattern spreading across the Chinese tech ecosystem. Companies that spent years collecting and refining digital data now realize that the same data is the key to building intelligent physical products. The accumulated knowledge about how fabrics behave, how materials react to touch, and how surfaces interact with light is not just useful for virtual simulations — it is essential for a robot to handle objects in the real world with the delicacy that everyday tasks demand.
Alibaba, Volkswagen, and the weight of data as a strategic asset
It is not just startups that have realized the value of uniting AI and hardware. Alibaba, which has been focused largely on AI tools within apps, revealed this month that its maps unit, Amap, is developing a quadruped robot. The initial goal is to assist blind people, given the shortage of guide dogs in China.
Alibaba’s strategy is to use specialized data from 20 years of digital cartography to gain an edge in robotics. Map data can help the robot’s sensors with navigation, while AI tools allow the robot to find nearby convenience stores, for example, based on commands like I am thirsty, as Mu Xu, head of embedded AI algorithms at Amap, explained.
However, Mu Xu raised an important warning: particularly for robotics, the ability to process powerful AI directly on devices becomes critical — and represents the biggest challenge. Once that constraint is solved, the question will no longer be about how capable AI models are in theory, but rather what the technology can do when it is inside every device.
Electric vehicle manufacturers are also in this race. Volkswagen, the German automaker, announced that it is rolling out in-vehicle AI tools that respond to driver voice commands. American, Korean, and German car companies rushed to present a new lineup of models for China during the Beijing Auto Show, using AI developed locally by ByteDance and other Chinese companies.
What Volkswagen understood is that competing in China’s electric vehicle market takes more than a good motor or an efficient battery. It takes a digital experience that makes sense for the local user, and that is only possible with hardware and AI developed together, with data that reflects the reality of that specific market.
This convergence between large corporations and the startup ecosystem creates an interesting dynamic. Startups bring agility, innovation, and a willingness to experiment. Big companies bring scale, infrastructure, and access to data volumes that no startup can generate on its own. When those two worlds meet, the result tends to be products that combine the best of both sides. 🚀
The geopolitical and economic context behind this race
All this movement in the hardware with AI sector is not happening in a vacuum. The geopolitical landscape is directly influencing decisions made by companies and governments. While foreign automakers bet on local technology to hold on to market share in the Chinese automotive sector, trade tensions between major powers are redrawing business routes.
Former Canadian Prime Minister Justin Trudeau recently stated that economic coercion from the United States and Europe almost pushed Canada into China’s arms, with Canadian companies turning to close deals with Chinese partners. And Singapore’s Foreign Minister Vivian Balakrishnan warned that if a war between China and the United States broke out in the Pacific, what is happening in the Strait of Hormuz would look like just a rehearsal.
These tensions reinforce the importance of technological sovereignty and explain why so many Chinese companies are prioritizing the development of their own hardware running on locally manufactured chips. The ability to operate independently of foreign infrastructure is not just a competitive advantage — in certain scenarios, it could be a matter of operational survival.
What this race means for the rest of the world
Keeping tabs on what is happening in Hangzhou, Suzhou, and other cities across China’s tech ecosystem is not just a geopolitical curiosity exercise. It is an important read on where the global technology market is heading. The integration of AI and hardware will redefine entire product categories, from wearables and medical devices to autonomous vehicles and industrial systems. And the countries and companies that arrive first with solutions that actually work in the physical world will hold a competitive advantage that is far harder to overcome than a lead in language model benchmarks. You can improve a benchmark with more data and more compute. A product in the market working with real users is a completely different story.
For tech companies outside China, the message is clear: the window to build leadership in this space is open, but it will not stay open forever. Chinese startups already have physical products with integrated AI running at scale, at prices that make adoption accessible to a much broader market than the premium segments many Western companies still target. That creates pricing pressure, feature pressure, and launch speed pressure that will force the entire industry to adapt.
At the end of the day, what China is showing is that artificial intelligence is most powerful when it touches the real world — when it is inside the device you use, the factory you run, the car you drive. And the startups that learned that lesson first are the ones setting the rules of the game right now. Smart hardware is not the future. For those in Hangzhou, it is already the present. ⚡
