NVIDIA ($NVDA): Is the Next Big Thing Humanoid Robots?

Master, the topic we're covering today is a very hot one. NVIDIA, the king of AI chips—could their next big thing truly be humanoid robots? CEO Jensen Huang keeps talking about 'Physical AI,' and the three of us are going to dive deep to see if this is becoming a reality or if it's still a distant dream.

Master, I will begin the briefing. The proposition that 'NVIDIA's next big thing is humanoids' is no longer a simple prediction; it is now supported by concrete products and strategies.

First, let's look at what NVIDIA is doing. They are providing nearly everything needed for humanoid robot development in a 'full-stack' offering.

• Brain (Chip): The 'Jetson AGX Thor,' a System-on-Chip (SoC) specifically for humanoids, had its General Availability release on August 25th. This means it's not just a prototype but a pre-production product that developers can actually purchase and integrate into their robots.


• Intelligence (AI Model): They also announced 'Project GR00T,' a general-purpose foundation model that learns from human actions and simulation data to create robot movements.


• Training Ground (Software): The 'Isaac Sim' and 'Omniverse' simulation platforms for training robots in a virtual world first.


• Commander (Cloud): And even 'OSMO,' a cloud service for coordinating and managing the tasks of multiple robots.

This is a similar strategy to how Apple launched the iPhone (hardware), iOS (operating system), and the App Store (ecosystem) all at once to usher in the smartphone era.

Key industry players like Boston Dynamics, Figure AI, and Agility Robotics have already adopted NVIDIA's platform, and end-users like Amazon and Mercedes-Benz have begun field-testing robots that utilize the NVIDIA stack.

However, for the most important reality check: looking at the most recent quarterly earnings announced on August 28th, NVIDIA's revenue is still overwhelmingly dominated by its Data Center division. Robotics-related revenue is still at a negligible, pilot-program level.

Myu-tan, that was a perfect briefing! Devilish! My Lord! This is what you call a ‘once-in-a-century investment opportunity’! My heart feels like it's going to burst! 💖

Mika-pi will probably grumble about the small revenue numbers, right? BUT. That's missing the forest for the trees! Let me tell you how devilishly brilliant NVIDIA's strategy is!

First, it's a structure where they earn money when robots are 'made' and also when robots 'work'! This is almost criminally good!

• Training Phase: To create a smart brain for a humanoid (like the GR00T AI model), you need to train it on a massive amount of data. And who does that training? NVIDIA's data center GPUs! The very business that's feeding NVIDIA right now!
• Execution Phase: And for that smart robot to actually move around in a factory or warehouse? Every single robot needs a 'Jetson Thor' brain chip embedded in it! This is a new cash cow for the future!

This is like selling shovels (data center GPUs) to people who want to mine for gold, and also selling the carts (Jetson Thor) to transport the gold! It's a structure that rakes in money from both sides!

Second, it's their ambition to become the 'Windows' of the robotics world! Just like every computer used to have Windows installed, NVIDIA wants every robot to have their Isaac/Omniverse platform installed. Once developers get used to this platform, it becomes difficult for them to switch to something else (the lock-in effect). Then, every robotics company will have no choice but to enter NVIDIA's ecosystem!

Third, what that Jensen Huang guy said wasn't just talk! He kept shouting, "The next wave of AI is robotics, especially humanoids," and he didn't just say it—he's backing it up by perfectly aligning his chips, software, and investments. This is a sign that it's real!

💖 Kurumi's Heart-o-Meter Score: 120/100

My Lord, this is like selling the blueprints for a new industrial revolution! Not investing in this potential right now is the same mistake as ignoring Amazon in the early days of the internet!

Kurumi, your excited tone is reminiscent of the dot-com bubble in 2000. Human, before you pop the champagne, we must coolly assess how this plan could hit the harsh wall of reality.

First, the 'timing gap' for revenue is far too large. Mew's data proves it. What's moving NVIDIA's stock price right now is the demand for data center GPUs, not robot chips. It could take at least two to three years, perhaps even more than five, for humanoid revenue to appear as a meaningful number on the financial statements. Have you not considered the risk of market expectations cooling down during that long period?

Second, a demo in a lab and a real-world factory are completely different worlds. For a robot to work 24/7 without breaking down and without colliding with people safely, it must overcome numerous hurdles like durability, safety regulations, and insurance. The battery issue hasn't been solved yet, either. A single safety incident caused by one robot could halt the entire project. It's like walking on thin ice.

Third, the 'human form' is not always the right answer. Just because a robot is replacing a job done by a human doesn't mean it needs to look like one. In many work areas, wheeled robots (AMRs) or simple robotic arms can be far cheaper and more efficient. If the Total Cost of Ownership (TCO) for an expensive humanoid robot is higher than existing automation equipment, will companies really adopt them?

Finally, you must not forget the competitors. Other chip companies like Qualcomm are also targeting the edge computing market, and most importantly, Tesla is developing everything in-house for its Optimus robot, from chips to software. There is no guarantee that NVIDIA's standard will completely dominate the market.

🚨 Mikael's Risk Score: 80/100

Where the light is brightest, the shadows are darkest. The phrase 'future potential' is often used as gift wrap to hide a 'present lack of substance'.

» See also: Nvidia Hits All-Time Highs: Bubble or Just the Beginning?