This guy built a $300M a year business selling açaí bowls. 🫐 @oakberryacai now has 1,000+ locations across 50 countries but the bowl itself is simple: frozen açaí, banana, granola, honey. The real business is what sits behind it. They built a factory in the Amazon to process and freeze the fruit, then franchise the stores so other operators can open locations under the OAKBERRY brand. And to make sure people recognise that brand before a store even opens nearby, they sponsor Formula One and tennis. That’s how a fruit bowl becomes a $300M business.

It costs $20,000 to build a single robot that does your dishes. The company making it is worth $1.15 billion right now. The company is Sunday, founded by Tony Zhao and Cheng Chi, both Stanford-trained roboticists. Zhao had built ALOHA, a low-cost imitation-learning system, and a technique called Action Chunking with Transformers, which lets a robot predict a short sequence of moves at once instead of guessing one twitchy motion at a time. Chi had built Diffusion Policy, a different approach better suited to the messy, unpredictable nature of real-world manipulation. Neither breakthrough was complete without the other. The two had never met before their papers put them on each other's radar. Their shared insight was that the actual bottleneck in robotics was never the algorithm. It was data. A factory robot can be teleoperated by a trained specialist for months, repeating one motion until it's flawless. A home robot has no such luxury. Every kitchen is different. Every grip on a wine glass needs a different amount of force than a grip on a coffee mug. That kind of chaos can't be simulated cheaply, and it can't be taught by one engineer in a lab. So instead of teleoperation, Sunday built something they call the Skill Capture Glove, a roughly $400 wearable that records a person's own hand movements while they do ordinary chores, folding socks, clearing a table, loading a dishwasher. They call the people wearing them Memory Developers. More than 500 households are already doing this, generating what the company says is over 10 million recorded episodes of real household routines, none of it from a lab. That data trains ACT-1, Sunday's foundation model, and it feeds their first product: Memo, a wheeled, two-armed home robot. Today it can clear a table of plates and delicate glassware, load and run a dishwasher, pull an espresso shot with real crema, and fold a pile of socks. In one internal test, it succeeded at folding laundry 99.1% of the time across 785 attempts in homes it had never seen before. Sunday emerged from stealth on November 19, 2025. Four months later, on March 12, 2026, they closed an oversubscribed $165 million Series B, led by Coatue Management, with Tiger Global, Benchmark, and Bain Capital Ventures participating, at a $1.15 billion valuation. Total raised: $200 million. "We raised our Series B to stop giving demos," CEO Tony Zhao said at the time. "Now we're focusing entirely on deployment." Sunday expects mass manufacturing to cut that $20,000 build cost by at least half before it ever reaches a customer's home. Beta deliveries are targeted for late 2026, with thousands of applicants already waiting.