👀 Who to watch next? @baseapp

.@cobie was a plumber

Today I saw @UnitreeRobotics release the "Superman" robot: it can jump 2 meters high in place, with a top speed of 12.66 m/s, directly breaking the human record. At the same time, Unitree's IPO countdown has started. Global humanoid robot shipments reached nearly 19,100 units in the first half of the year, a year-on-year increase of 272%, with industrial and commercial scenarios accounting for over 70%. The lively action shows, record-breaking moments, and capital frenzy all remind us: the real moment to bet on robots is not when they "can walk," but when they start transforming from machines into labor. Recently, I re-examined the robotics track from scratch and found that most humanoid robots are currently undergoing a shift from "being able to perform actions" to "being able to work stably for long periods." After that comes dexterous operation, precise assembly, and eventually large-scale fleets. So, when judging the progress of this industry in the future, I will pay more attention to several data points: Robot active hours, number of tasks per unit, task complexity, autonomy level, continuous unattended operation time, repeat orders, and the customer's real ROI. Robot quantity × working hours × task quantity × complexity × autonomy may be closer to the true economic value created by this industry than pure "sales volume." Following this logic, the value of the robotics industry will also continuously shift: Motion control → continuous performance/thermal management → precision transmission/force control → lifespan and manufacturing consistency → fleet operation and management. So, I will focus on three companies next, but for completely different reasons. Inovance Technology: They have layouts in motors, drives, servos, encoders, control, and thermal management. If in the future many robot OEMs are unwilling to redevelop every component themselves, their most imaginative endgame is to become the universal Motion Platform for robots. Harmonic Drive Systems: Once robots enter dexterous operation and precise assembly, the truly critical joints will increasingly care about low backlash, miniaturization, accuracy retention, and long-term reliability. It’s more like betting on a few uncompromisable Critical Axes. Green Harmonic: What I care most about is whether they can transfer the precision manufacturing capabilities accumulated in harmonic reducers to more products like actuators, roller screws, and linear joints. If they can, their story will evolve from a "domestic reducer" to a broader precision execution platform; if not, they will remain a component company benefiting from robot volume growth. Here is another very important idea: The stronger the AI, the more hardware still matters. Software can compensate for friction, hysteresis, and some errors, but it cannot compensate for heat, fatigue, impact, wear, and material lifespan. The smarter the robot, the more complex the tasks it can perform, the longer the working hours, and the more stringent the physical world requirements. So I think the next phase is no longer the so-called "robot NVIDIA." Robot hardware is naturally more fragmented than computing architectures, and in the end, multiple Physical AI Toll Booths like ASML / TSMC / Keyence are more likely to emerge: each controlling key links such as precision transmission, motion control, perception, reliability, and operation & maintenance, then continuously charging as the entire robot economy expands. Risks will also be obvious: Success in the robotics industry does not mean that today’s favored suppliers will definitely succeed. OEMs can develop in-house, technology routes can change, and automotive and industrial giants will also enter. Parts that are scarce today may become dirt cheap in five years. So when judging moats, I now like to think in reverse: If a competitor has enough money and five years, can they copy you? What is truly hard to copy may not be a single parameter but scale manufacturing consistency, lifetime validation, co-development with customers, and reliability data from hundreds of thousands of devices. This is also the core of what I will continue to track in the robotics track. I will keep following: Which industrial stage are robots entering? How far are we from the economic singularity of Physical AI? When that singularity truly arrives, who controls the unavoidable bottlenecks? If these questions can be answered in advance, many companies’ revenues and profits will actually be the result that follows.

$BullCome I must be incredibly insensitive in the universe, but since the first time I noticed it, it has already exceeded 10x at its peak. The most worth reflecting on is that when I first saw it, I already thought "too expensive," so I never really got involved. I didn't understand the market, nor did I understand the trend. I always judge "whether this thing is worth so much money" based on my own aesthetic, but the market trades consensus, attention, and capital trends. Just because something has risen a lot doesn't mean the trend is over; you think it's expensive, but that doesn't mean the market thinks so. Every season is a process of continuous improvement. This season I can say I was really bad, basically losing 100 times and winning once, and that one win probably involved quite a bit of luck 😂 My biggest recent reflection is: give up personal aesthetics and respect the market. Study what the market likes; understand why capital flows where it does. The market is always right—first understand the trend, then discuss your own judgment. Also, I increasingly feel that large groups are actually useless. The more information, the more noise. What truly has value are 5-10 comrades who share the same goals, are willing to sincerely share, correct each other, and improve together. Each person contributes their strongest area, and over the long term, this creates an information advantage far more valuable than groups of hundreds or thousands.

We will seize all the hundredfold opportunities Here, on X, openly With no conspiracies We should all believe in ourselves!