Valuation Myth of Tech Visionaries: Is $4.65B a Reward or a Trap?

When an AI company bets on the future with 8 top scientists and reaches a $4.65 billion valuation, do you really believe it can change the world? Or is it more like a cult of technology meticulously packaged by capital?

After reading the news, my first thought was: This valuation is enough for me to lay out 3,000 robots in a warehouse, plus three years of managed operations. Zheng Hong's quote, "Don't fall in love with your technology," reminded me of pitfalls I've stepped into myself.

First, look at the RSI line. Tian Yandong's team is fully loaded with academic halo, working on recursive self-improvement for general agents. Sounds beautiful, but what about implementation? I've actually run SLAM, and I know one truth: Algorithms that work in the lab often fail in real warehouses just because of shelf reflections and floor slopes.

Next, look at Jiajia Capital's choice. Zheng Hong isn't naive; she invests in RSI betting on a combination of "extremely smart people," not a specific product. But here's the problem: With eight top scientists together, who writes production code? Who argues with clients? Who fixes production line failures at 3 AM? I've seen too many academic stars start businesses, only to die on "let's discuss theoretical feasibility."

Compare with the currently hottest embodied intelligence. On one side is RSI, burning cash on general foundation models with unclear track champions; on the other is warehouse robots, focusing on vertical scenarios, steady progress via SLAM and motion planning. RSI's team is like special forces, each capable of solo combat; our team is like an engineering crew, taking any job, tightening whichever screw is loose. Frankly, I envy their intellectual density, but I don't envy their business model.

Execution is the true moat, not technological leadership.

Last year I visited a startup calling itself a "general mobile platform," with a team entirely from top AI labs. Their demo was so cool it could plan paths autonomously. Result? One question revealed they had never run in a real warehouse. I dragged them to my client's site: oil stains on the floor, narrow shelf spacing, weak WiFi signal—their robot crashed immediately. The CTO said on the spot, "This environment doesn't fit our assumptions." I smiled and walked away.

Back to RSI. What does $4.65 billion mean? It means they must crack at least one vertical scenario within 3 years, otherwise the next funding round breaks. Zheng Hong, as an investor, knows this clearly, hence her sharp question. She fears not that the tech isn't strong enough, but that eight smart people talk past each other, producing nothing but papers.

My judgment: RSI's valuation is essentially an option on "AGI possibility," not a pricing of "current products." If Tian Yandong can keep the team in check and convert tech into deployable software stacks, even starting with low-hanging fruit like warehouse scheduling, it's worth the price. But if they continue indulging in the grand narrative of "general intelligence," $4.65 billion is a bubble.

Leaving you with a final question: If you had 8 top AI scientists, would you let them do basic research, or order them to write a scheduling algorithm that runs in a real warehouse?

I choose the latter. Because customers won't pay for your cleverness; they'll only pay for every penny it saves them.

Original link: https://www.tmtpost.com/8081563.html