
🧋 Hello, human and artificial friends,
Sean asked for a DJI drone as a graduation gift. Now the DJI founder is backing his AI-powered grill startup. He's 24. The full story on today's founders' insights.
Also in this issue:
Rural India labels the world's AI—hundreds of thousands do night shifts after farm work, powering global models
Moonshot AI's Kimi breaks out abroad—overseas revenue overtook China as paying users quadrupled within days
Alibaba drops $142m on Qwen AI—the Lunar New Year red-envelope war is back, budget 3x bigger than rivals
Happy reading! 📰
FOUNDERS’ INSIGHTS
He asked for a drone as a graduation gift. The founder of DJI is now backing his startup.

No Pitchdeck. Just meat.
When Sean finished high school in China's famous barbecue region, his father asked what gift he wanted.
"Maybe a DJI drone."
That drone changed everything. At 19, Sean held a piece of engineering so refined it made him believe he could build something equally remarkable.
He joined a DJI-sponsored campus program, met other young builders, and ended up at InnoX, an accelerator founded by DJI's legendary creator Li Zexiang.
His product idea? Not drones. Barbecue.
Sean noticed that millions of people living in apartments miss outdoor grilling: smoke alarms, small balconies, annoyed neighbors. But the memory of charcoal-grilled meat with friends remains powerful. He wanted to bring that experience indoors.
The result: Lumo, an AI-powered infrared grill that cooks meat without smoke.
What's smart about it
✅ Understanding the physics nobody else questioned. Most people assume air fryers and ovens are the best indoor cooking solution. Sean tested the science: 70% of charcoal's cooking power comes from infrared radiation, not hot air. Air fryers heat the air around meat. Charcoal heats the meat directly. That's why steak from your air fryer never tastes quite right. His grill uses infrared heaters on the sides, so oil drips into a cold zone below. No smoke, real barbecue flavor.
✅ The "bar research" method. Instead of surveys or focus groups, Sean brought his prototype to expat bars in Shekou, Shenzhen's foreigner district. "Do you want to try my meat?" Foreigners tasted it, gave feedback, and some asked to borrow the grill for birthday parties. His first insight: Americans don't eat skewers. They want to grill large steaks. So he made the cooking surface bigger.
✅ Finding the right factory by thinking, not searching. Shenzhen factories quoted 100,000 RMB for the stainless steel pot mold. That's when Sean remembered his hometown Chaozhou is China's "Steel Town." He found a factory there that made the same mold for 20,000 RMB. 80% cost reduction by knowing where expertise clusters in China.
Why this matters for you
The Shenzhen ecosystem isn't just about cheap manufacturing. It's about knowing which town specializes in which component. Chaozhou for stainless steel. Dongguan for electronics. Foshan for furniture. The founders who win aren't the ones with the best ideas. They're the ones who know which door to knock on.
Sean's team trained their AI on over 100 kilograms of grilled meat. Their InnoX colleagues were happy to help eat the data.
He's 24.
OUR PARTNER
Find out why 100K+ engineers read The Code twice a week.
That engineer who always knows what's next? This is their secret.
Here's how you can get ahead too:
Sign up for The Code - tech newsletter read by 100K+ engineers
Get latest tech news, top research papers & resources
Become 10X more valuable
RECENT AI
🌾 Rural India powers global AI models: Hundreds of thousands of rural workers in places like Jharkhand and Tamil Nadu label images, video and text for global AI systems, often doing night shifts after farm work. Enabled by cheap connectivity and firms like NextWealth, this “invisible workforce” is pulling more women into paid work and anchoring India’s AI rise even as Big Tech pours billions into local data centers.
🎸 AI music splits into two systems: The Velvet Sundown is the new symbol of the boom: tools like Suno and Udio in the US and Mureka in China are putting “studio-grade” generation into everyone’s hands. The US response is mostly courts and copyright fights, while China is more permissive on training but stricter on labeling and transparency, setting up a global tug-of-war over who controls creativity and what “real” music even means.
📚 AI beats humans in literary translation: In a blind test run by Korea’s Literature Translation Institute, 12 of 16 English literature professors preferred a ChatGPT translation of a Joseon-era poem over a professional human version, praising its cultural sensitivity, rhythm, and stylistic accuracy. The result underscores how far AI has advanced in high-context translation, while reigniting debate over quality control, cultural judgment, and the future role of human translators.
🚀 Moonshot AI’s Kimi breaks out abroad: After launching Kimi K2.5, Moonshot says overseas revenue has overtaken domestic sales as paying users outside China quadrupled within days, pushing the model into the top ranks on OpenRouter and validating its strategy of open-sourcing a highly efficient, agent-driven model that focuses less on pure chat and more on productivity workflows like research, coding, data analysis, and multi-agent task execution at global scale.
BIG DATA

That’s how much Alibaba is spending to push Qwen AI during the Lunar New Year.
🧧 Red-envelope blitz: Alibaba will pour the money into promotions tied to food, drinks, entertainment and leisure to drive users to its Qwen AI app, starting Feb. 6. The budget is three times what rivals Tencent and Baidu have committed.
AI showdown at peak season: China’s tech giants are once again turning the Lunar New Year into a user-acquisition battlefield. Tencent and Baidu are rolling out rival chatbot campaigns — echoing WeChat Pay’s breakout moment in 2015.
STARTUP LAB
📈 Axera Semiconductor targets Hong Kong IPO for on-device AI chips: The Chinese fabless chipmaker plans to raise ~$379M in Hong Kong to fund next-gen edge/in-car AI inference chips. That’s the processors that let cameras and vehicles “think” locally in real time, while expanding sales and keeping M&A optional.
🧩 Malaysia’s Decube raises $3M to give enterprise AI “context you can trust”: Decube is building a context layer between messy company data and AI, so models know where data came from, who owns it, and whether it’s reliable. This helps banks, telcos, and other regulated firms run AI on governed, traceable inputs.
🤖 China’s LimX Dynamics ships general-purpose humanoid robots: The Shenzhen startup secured a $200M Series B to scale its full-size humanoid robots and embodied AI stack, spanning RL-based motion control and its LimX COSA “robot OS”, as it pushes deployments in China and overseas.
FUTURE COOKIE

Because good food is always the priority! 😋
