Why AI Companion Hardware Is Having a Moment
The pitch for most AI products sounds the same: be more productive, automate your workflow, save time. But a growing number of founders are betting on a completely different use case for AI companion hardware -- not efficiency, but emotion. Not tools, but toys. Not assistants, but friends.
The timing makes sense. Loneliness among young adults has reached levels that the U.S. Surgeon General has called an epidemic. Gen Z, the most digitally connected generation in history, is also the loneliest. Social media promised connection but delivered comparison. And while software-based AI companions like Character.ai have attracted tens of millions of users, they live on the same screens that cause the problem in the first place.
Physical AI companion devices offer something different. They exist in the real world. You can hold them, carry them, bump them against your friend's device. At CES 2026, multiple companies showed AI companion hardware that blurs the line between gadget and pet. The category is small but growing fast, fueled by better on-device AI, cheaper sensors, and a generation hungry for companionship without the baggage of another social feed.
And then there is the nostalgia factor. The Tamagotchi sold over 91 million units starting in 1997. Nearly 30 years later, the concept of a tiny digital creature you care for still resonates -- except now, AI can make that creature actually understand you.
Meet Sweekar: The AI Pet Device That Grows With You
Among the most talked-about products in this new wave is Sweekar (known as 话话糖 in Chinese), billed as the world's first physically growing AI pocket pet. Built by Shenzhen-based Takway AI, Sweekar is an egg-shaped device that weighs just 89 grams and fits in your palm.
What makes it unusual is that it evolves. Sweekar goes through four life phases -- Egg, Baby, Teen, and Adult -- and its personality develops based on how you interact with it. The device assigns itself an MBTI-type personality that shifts over time. Talk to it gently and it becomes warm. Challenge it and it gets feisty. Ignore it and, well, it notices.
Under the hood, Sweekar runs a hybrid AI model combining Gemini Flash with ChatGPT-class language processing. It has emotional memory, meaning it remembers not just what you said but how you felt when you said it. Context from previous conversations carries forward. Over weeks and months, the device builds a model of its owner that no generic chatbot can match.
One feature that stands out: NFC coupling. Bump two Sweekars together and the devices interact socially. Your AI pet meets someone else's AI pet. It is a small detail, but it turns a solo experience into a shared one -- and gives the product built-in virality.
CNET, reviewing the device after CES 2026, called Sweekar "a worthy successor to the iconic 90s Tamagotchi." Engadget and Cybernews covered it too, with Cybernews racking up 381,000 views on a single Instagram Reel about the device.
The Founder Who Bet $300,000 on Emotions Over Efficiency
Irving Gao is not your typical hardware founder. Born after 2000 (known in China as 00后), he is among the youngest founders we have featured on the Asiabits Podcast. He is currently a master's student at the University of Science and Technology of China (USTC), one of the country's top research universities.
Before starting Takway AI, Irving worked at three of China's most prominent AI and robotics organizations: Unitree Robotics (the quadruped robot maker), SenseTime (once China's most valuable AI unicorn), and the Shanghai AI Lab. He was on track for a PhD in robotics. The conventional path was clear.
Then, toward the end of 2023, he saw something that changed his mind. Multimodal large language models were getting good enough to run on small hardware. The gap between a cloud-based chatbot and a physical device with real personality was closing fast. Irving decided the moment to build was now -- not after a PhD.
The decision came with real risk. He took out a loan of 2 million RMB (roughly $300,000) to found Takway AI in Shenzhen in April 2024, with a second office in Las Vegas to be closer to the U.S. market. For a master's student in his early twenties, that is not a small bet.
The bet attracted attention. Irving secured investment from the Nanshan Science & Innovation Fund, connected to Li Zexiang -- the professor and investor behind DJI and other Shenzhen hardware success stories.
"Nintendo of AI, Not iPhone of AI"
Irving's philosophy breaks cleanly with the mainstream AI narrative. He calls it "Anti Tool-Rationalism" -- the idea that AI does not need to make you more productive to be valuable. Instead, AI can make you feel something. It can play with you. It can keep you company.
He frames Sweekar not as an "electronic pet" but as a "digital life." The difference matters. A pet is a feature set. A life has a personality, memories, and moods that develop independently. The AI gives the classic Tamagotchi format something it never had before: a soul.
The product design borrows heavily from game psychology. Irving studied gamification at InnoX Academy, and three principles show up clearly in Sweekar:
- The IKEA Effect: Every interaction shapes the pet's personality, which means every owner builds something unique. You cannot buy a pre-made version of your Sweekar. That creates emotional investment -- not because the hardware is expensive, but because you made it yours.
- The Skinner Box: Sweekar uses GPS and environment sensors to deliver unpredictable rewards. Walk past a park and your pet might react differently than in your apartment. The randomness keeps owners coming back.
- Social Currency: The NFC bump feature turns the device into a conversation starter. Showing someone your Sweekar and watching the two devices interact creates a moment you want to share.
Irving's shorthand for all of this: he wants to build the "Nintendo of AI, not the iPhone of AI." The iPhone is about productivity and utility. Nintendo is about play, surprise, and delight. In a world drowning in productivity tools, Irving is betting that play wins.
Why Launch in America First, Not China?
For a Shenzhen-based company, the obvious first market is China. Irving went the other direction. Takway AI is targeting the United States first, with a Kickstarter campaign planned for 2026 followed by direct-to-consumer sales and Amazon distribution.
The reasoning is practical. China's consumer electronics market is brutally competitive -- the Chinese term is 卷 (juan), meaning everyone is grinding on the same thing. For AI hardware, that competition mostly plays out as a spec war: more parameters, faster chips, lower prices. An emotional AI pet that charges a premium for personality does not fit that playbook.
The U.S. market offers two advantages. First, there is an established geek culture that embraces novel gadgets early. Kickstarter campaigns for quirky hardware regularly raise six and seven figures. Second, the mental wellness market in America is enormous and growing. Americans spend real money on products that promise emotional well-being -- from meditation apps to weighted blankets. An AI companion device sits at that intersection.
There is also a defensive reason. Launching in Shenzhen means immediate exposure to Huaqiangbei -- the world's largest electronics market and the capital of hardware clones. A new product shown at a local expo can be reverse-engineered and copied within weeks. By building the brand internationally first, Irving can establish pricing power and brand recognition before copycat products appear.
Irving describes the approach as "high dimension hitting low dimension" -- build a global brand with premium positioning first, then enter the Chinese market from a position of strength. It is the same playbook that worked for DJI, Anker, and a growing list of Shenzhen-born companies that went global before going domestic.
THE THREE-PHASE GO-TO-MARKET
Takway AI's launch plan follows three phases:
- Social media seeding: Building awareness through content, influencer demos, and viral moments (the 381K-view Cybernews Reel is early proof this works).
- Kickstarter 2026: A crowdfunding campaign to validate demand, generate pre-orders, and build a community of early adopters.
- DTC + Amazon: Scaling into mainstream retail through direct-to-consumer sales and Amazon, the two channels where novel gadgets find mass-market traction in the U.S.
What This Means for the AI Hardware Landscape
Sweekar is one product from one startup. But it represents a broader shift in how founders are thinking about AI hardware. The first wave of consumer AI devices -- smart speakers, AI pins, AI-powered glasses -- all tried to be productivity tools. Most of them struggled. Consumers already have a phone in their pocket that does productivity just fine.
The emotional AI companion device is a different bet. It does not compete with your phone. It fills a need your phone created: the desire for connection that does not come through a screen. For a generation that grew up with social media and came out the other side feeling lonelier, a pocket-sized companion with a real personality is not a gimmick. It is a product category waiting to happen.
Whether Sweekar specifically succeeds will depend on execution -- manufacturing at scale, battery life, AI latency, pricing, and a hundred other details that separate a CES demo from a real product in real pockets. But Irving Gao's $300,000 bet on emotions over efficiency is the kind of contrarian move that occasionally builds entire categories.
The Tamagotchi proved 30 years ago that people will care for a digital creature. The question now is what happens when that creature can actually care back.
This article is based on Asiabits Podcast Episode 7 with Irving Gao. For more on the Shenzhen hardware scene, explore our Insights or get in touch.
Get These Insights Every Morning
Join 18,000+ professionals who start their day with Asiabits. Free, every weekday, straight from Shanghai.
Subscribe Free →
