Most people in China know Justin Sun from his bizarre stories.

Spending 30 million to have dinner with Warren Buffett, only to cancel at the last minute due to a kidney stone; buying a duct-taped banana for 6.2 million dollars at an auction and eating it in front of everyone at a press conference; investing 75 million dollars to become the largest backer of the Trump family's crypto project, securing a seat at a White House dinner; and at the age of 35, flying past the Kármán line, declaring himself the youngest Chinese commercial astronaut.

There are also many negatives. In 2023, he was sued by the SEC for market manipulation, with accusations including over 600,000 wash trades to inflate the price of TRX and hiring celebrities for promotion without disclosing payments. Currently, he is in a lawsuit with the Trump family-related project WLFI.

These stories spread so widely that they almost overshadowed a serious matter. This man, over the past decade in the secondary capital markets, has almost never missed betting on any trend.

From buying BTC starting in late 2013, to 2016, advising post-90s generations not to buy houses, but to buy:

Bitcoin, NVIDIA, Tesla, Tencent.

Ten years have passed. As of May 2026, Tesla's total return is approximately 2683%, and NVIDIA's total return is nearly 24000%.

If you had listened to Brother Sun back then and invested 10,000 yuan in NVIDIA, it would be 2.4 million today; 10,000 yuan in Tesla would be 278,000 today. A listener who bet 200,000 yuan on each item on that list in 2016 would have seen just the NVIDIA bet turn into approximately 48 million, and the Tesla bet into approximately 5.4 million, totaling over 53 million.

And this man continues to take shots today. On November 6, 2025, Justin Sun dropped a statement:

The capital market's reaction to this statement didn't reach fever pitch until 2026. SanDisk (SNDK), spun off from Western Digital, rose from a low of around $35 to $1,439 in a year, a maximum increase of nearly 50 times.

HBM memory capacity from Samsung, SK Hynix, and Micron was fully booked for 2026, already sold out, with orders scheduled through 2027-2028.

While everyone is still chasing the memory concept frenzy, in early 2026, Justin Sun changed his tune again in a video.

That video was originally about the outlook for 2026. Besides topics leaning towards wellness like putting health on the new year's list, he dedicated a segment to advise young people: embodied AI, drones, spatial computing, space exploration.

The author collected Justin Sun's public statements over the past two years regarding these four directions. Stringing them together, preliminary capital leaders have also emerged on each path.

Who Is the Next Storage Stock?

The first area Justin Sun pointed to was embodied AI.

The concept of robots has been discussed by humans for at least a hundred years. The Czech playwright Čapek coined the word "Robot" in 1920; industrial robotic arms have been used since the 1980s; Honda's ASIMO could climb stairs over twenty years ago. But the real bottleneck has always been the brain.

In the last two years, the entire industry has shifted towards VLA models—Vision-Language-Action. In plain terms, robots used to act based on code; now robots are starting to act based on seeing the world.

Unitree shipped over 5,500 humanoid robots in 2025, ranking first globally, and applied for a Science and Technology Innovation Board IPO in March 2026. Galaxy General raised a new round of $300 million (approx. 2.1 billion RMB) in December 2025, with cumulative funding of about $800 million and a valuation of $3 billion (approx. 21.1 billion RMB), setting new records for both single-round and total funding in the embodied AI sector.

Justin Sun said he likely won't personally venture into building robots, but he has a nose for narratives and the direction of capital flow. He said in a Bloomberg interview, "In a market where 99% of people don't know what a wallet is, the cost of education must be factored into the business model."

This statement, used to explain stablecoins in 2018, applies equally well in 2026. 99% of Chinese haven't used an embodied AI robot yet, but as long as that robot can cook, move boxes, or care for the elderly, the remaining 1% is the next opportunity.

The second sector he pointed to is drones.

While humanoid robots are still ramping up production, drones have already taken a step ahead into commercial deployment. They are naturally suited for what AI does best, from autonomous navigation to swarm coordination to data collection. They don't need to walk; flying is simpler than bipedal locomotion.

On the Russia-Ukraine battlefield, AI drone swarms have already taken over a large part of the role of tank units; Ukraine's annual drone production target has climbed to several million units. Over rice fields in rural China, DJI's agricultural drones fly, one replacing ten farm workers. In Shenzhen, Meituan has operationalized drone delivery, with order-to-delivery times not exceeding 15 minutes.

Drones are ahead of humanoid robots. They are the first form where AI has achieved a commercial closed-loop in the physical world.

The third area Justin Sun mentioned is spatial computing. This is the least mainstream among the directions he pointed out.

When Apple released the Vision Pro in 2024, most people saw it as a VR headset that cost several times more. This might be a misreading.

The Vision Pro's ambition has little to do with VR. It's Apple's first attempt to make AI understand space—how big your living room is, how far the table is from you, whether the coffee cup is to the left or right of the sofa, if you can reach it when you stretch your hand. This sounds simple but is ten times harder than training ChatGPT. Large language models only need to understand language; spatial computing needs to understand physics.

This is precisely the common prerequisite for robots, drones, and autonomous driving—they all require a form of spatial intelligence. NVIDIA's Cosmos platform, Google's Genie 3 world model, Tesla's FSD, are all doing the same thing: transitioning AI from understanding text to understanding the world.

ChatGPT only needs to understand language, but the next generation of AI needs to understand the world itself.

For the first three sectors, Justin Sun only verbally pointed them out. When it comes to space, he actually went there in person.

On August 3, 2025, he boarded Blue Origin's "New Shepard" NS-34 capsule and flew past the Kármán line.

After returning to Earth, he expressed an ambition: to make his company not just a "cryptocurrency exchange" but a "service provider for space economy infrastructure," using blockchain to solve space asset ownership, satellite data trading, and interplanetary payments. It sounds like science fiction. But look back at how he evangelized USDT ten years ago—people thought that was science fiction too back then.

Back on Earth, his message to young people was more direct: "Space exploration is a common mission for all humanity. I hope this flight inspires more young people to devote themselves to technology and innovation, collectively shaping humanity's interstellar future."

Brother Sun's Investment Logic

Justin Sun's publicly expressed investment logic is: find sectors with a clear direction, place bets on both ends, and do not gamble on the execution of a single company.

On the robotics line, his framework is to bet separately on the body and the brain.

Bet on Tesla for the body. In early 2026, Tesla announced discontinuing Model S and Model X, converting the Fremont factory into an Optimus production line, targeting an annual output of one million units with a mass-production unit price of about $20,000 to $25,000; current version Optimus robots are already handling parts transport and sorting at Austin and Fremont factories; the Gen 3 production line is scheduled to start in summer 2026.

Bet on NVIDIA for the brain. Jetson Thor puts server-grade AI inference into the robot body, Isaac GR00T has almost become the industry's common base, and Jensen Huang declared at GTC that there will be 1 billion humanoid robots globally by 2035.

Whether Optimus delivers on schedule is Elon Musk's problem, not NVIDIA's. As long as the sector takes off, the toll will be collected.

On the drone line, the core judgment is the irreversibility of Physical AI in military scenarios.

AeroVironment's Switchblade loitering munition became an iconic weapon in Ukraine, with monthly production capacity increased from 40 to 500 units, targeting 1,200 units; a $3.9 billion order locks in revenue for the next three years in advance. Kratos' XQ-58 Valkyrie is the "loyal wingman" to the F-35—manned aircraft on missions, drones on the flanks—with a unit price less than a fraction of a fifth-generation fighter jet, rising 280% in 2025 and another 72% in 2026.

One makes tanks uneconomical, the other makes manned fighter jets redundant—the logic on both ends is complementary.

On the space line, Justin Sun bid $28 million for a Blue Origin flight seat in 2021. This money was donated to the Blue Origin STEM charity fund, distributed among 19 non-profit organizations. On August 3, 2025, he boarded the New Shepard NS-34 mission, completing a suborbital flight.

In public markets, SpaceX filed a confidential IPO draft with the SEC in April 2026, targeting a valuation of $1.75 trillion, poised to become the largest IPO in human history; Rocket Lab's Q1 2026 revenue exceeded $200 million, making it the most direct alternative investment when SpaceX isn't available.

Once SpaceX goes public, the entire space sector's valuation framework will be rewritten.

Listen to Brother Sun

Stringing together Justin Sun's statements over the past two years, "AI, robotics, and blockchain have reached their iPhone moment" is his judgment on embodied AI. "Robot armies, robot police" is his prophecy for autonomous weaponized AI. "The integration of AI, robotics, and spatial computing" is his bet on the next generation of human-machine interfaces. "Earth is too small; it's our home" is his perspective shift after flying past the Kármán line.

These four things pieced together form the complete picture of Physical AI.

Over the past two decades, the internet changed the way information flows. WeChat replaced letters, Taobao replaced markets, Douyin replaced TV.

But the underlying rules of the physical world didn't change. Workers are still workers, factories are still factories.

In the next two decades, AI might change the very way the real world operates. Factories will have tireless humanoid robots, roads will have autonomous vehicle traffic, battlefields will roar with drone swarms, and the first "residents" on the Moon and Mars are likely to be pioneering AI robots.

That young man who in 2016 told everyone not to buy houses has now flown past the Kármán line.

And most of us are probably still waiting for the next Yanjiao.