As Elon Musk's business expands, what is truly valuable?
As Elon Musk's business empire continues to expand, the most common mistake outsiders make is to focus on his individual products.
Tesla, SpaceX, and xAI seem to have nothing in common. However, investors can see clearly: after the merger of SpaceX and xAI, the expected valuation is $1.25 trillion. OpenAI just completed a $110 billion financing round, and its post - investment valuation is about $840 billion. The former's valuation is nearly $400 billion higher.
Where does this $400 billion difference come from? Look at what Musk has been up to recently. In an interview in Berlin, he mentioned that Full Self - Driving (FSD) will be approved in Europe in March, Cyber Cab will start mass production in April, and Tesla will expand five production lines. More vehicles mean more data. Meanwhile, SpaceX has applied to the FCC to launch one million satellites to build an orbital data center to provide computing power for AI training. Tesla's road condition data will be fed back to xAI, and the capabilities of FSD will support the mass production of Cyber Cab.
Rockets, satellites, AI, manufacturing, and scenarios are all part of a closed - loop system.
Section 1 | Cars, Robots, and Satellites are All Collecting Data
What makes this closed - loop system work? Data. Cars, robots, and satellites are all data entry points.
Tesla's vehicles drive by "observing like humans" rather than relying on detailed maps. The FSD system uses cameras to observe road conditions, pedestrians, and traffic lights. Every drive is an opportunity for it to learn and accumulate experience. Musk confirmed that the approval process in Europe will start as early as March. Every Tesla on the road is a mobile data collection node, continuously transmitting road conditions, traffic, and driving scenarios back to the system.
Cyber Cab will start mass production in Texas in April and large - scale production by the end of the year. It doesn't need a driver and can operate 24/7 on urban road networks. This means that AI can continuously learn in scenarios that human drivers can't cover, such as early mornings, rainy days, and peak traffic hours.
The humanoid robot Optimus collects a different type of data. Musk said that its initial tasks will be simple, but as it accumulates capabilities, it will enter households, hospitals, and nursing homes. He predicts that in a few years, robots may even perform surgeries. The robot's indoor movement, object grasping, and human interaction are all data sources for training AI to understand the human world.
The Berlin Gigafactory is preparing for this. Musk clearly stated that the factory has started producing batteries and may produce humanoid robots in the future. Every additional device produced by the factory means an additional data collection point deployed in the real world. This is the ground - based layout. In space, SpaceX is taking more aggressive actions.
SpaceX has applied to the FCC to launch one million satellites. What does this number mean? Currently, there are more than 5,500 Starlink satellites in orbit, which already constitutes the world's largest satellite network. One million satellites means expanding this scale by 180 times, achieving an unprecedented level of coverage density. These satellites can observe the Earth's surface in real - time and collect global environmental data such as meteorology, geography, traffic, and agriculture.
When we put all these together, we can see a data network.
Cars collect road traffic data.
Taxis collect urban operation data.
Robots collect indoor interaction data.
Factories collect manufacturing process data.
Satellites collect global environmental data.
Each entry point collects different types of data. The data is fed back to train AI. As AI becomes more powerful, it can enter more scenarios and collect more data. This is a positive cycle.
Although Tesla, SpaceX, and xAI are engaged in different businesses, they are weaving the same net: enabling AI to understand and learn about the real world from more dimensions.
The denser this net is, the faster AI can learn. The larger this net is, the harder it is for others to catch up.
Section 2 | Rockets, AI, and Manufacturing Capabilities Others Don't Have
For the data closed - loop system to function, it also needs computing power support.
Autonomous driving needs to learn from a vast amount of road conditions, and robots need to understand complex movements, all of which consume computing resources. Ground - based computing power is reaching its limit: there is a shortage of GPUs, electricity is expensive, and the cost of heat dissipation is soaring. The solution is to build data centers in space. The advantages of space are obvious: the sun - synchronous orbit provides 24/7 solar power, the vacuum environment offers natural heat dissipation, and it is not restricted by the ground power grid and land.
Sam Altman of OpenAI once publicly questioned:
"The idea of building data centers in space at present is absurd. A rough calculation of the launch cost compared to the electricity cost on Earth shows that we are not there yet."
However, Musk has a combination that others can't replicate.
SpaceX has rockets that can send equipment to any orbit, and the launch cost has been reduced to a commercially viable level. Starlink is currently the only large - scale operational satellite network, providing ready - made space infrastructure. xAI, established more than two years ago, is now capable of training top - level models and knows what kind of computing power and data AI needs.
All of this requires hardware, and manufacturing capabilities determine whether it can be scaled up. Tesla's Shanghai factory can produce a car in less than 40 seconds. This production efficiency means that it can quickly mass - produce the hardware required for space data centers. The Berlin Gigafactory is about to start battery production, and lithium and nickel cathode refineries have been launched in Texas. These are the industrial chain foundations that support hardware manufacturing. Musk said that five main production lines will start mass production this year.
Rockets solve the transportation problem, the satellite network provides infrastructure, AI clarifies the demand direction, and factories provide large - scale production capabilities. Only when these four elements are in place can the space data center turn from a concept into a feasible solution.
Other American players only have one or two of these elements at most.
Google has AI and will cooperate with Planet to launch equipment next year, but it doesn't have rockets and has to rely on SpaceX.
Jeff Bezos has the rocket company Blue Origin, but he doesn't have an AI laboratory or large - scale manufacturing capabilities.
Eric Schmidt, the former CEO of Google, bought the rocket company Relativity Space because he believes that orbital data centers will become a big thing. But he only has rockets and no satellite network or AI.
The startup StarCloud has launched an experimental satellite to test the performance of GPUs in space. But they lack SpaceX's launch capabilities and scale.
What most competitors lack is not ideas, but the underlying capability foundation.
In an interview, the host asked Musk: "What can European car manufacturers learn from Tesla?" His answer was a bit helpless: ideas can be shared, but without the corresponding capabilities, learning is useless.
Section 3 | The Most Valuable Thing is the Right to Define the Future
The combination of capabilities is just a means. The key is what this combination can be used for.
After the merger of SpaceX and xAI, the valuation reaches $1.25 trillion. In the same month, OpenAI just completed a $110 billion financing round, and its post - investment valuation is about $840 billion. The answer to the nearly $400 billion premium mentioned earlier lies here.
Companies like OpenAI are fiercely competing in the application layer: pursuing better conversations, stronger reasoning, and more accurate generation. On the other hand, SpaceX and xAI are working on the infrastructure layer: they are solving how to provide computing power, energy, data, and real - world physical scenarios for AI.
The competition in the application layer is about "whose model is smarter", while the competition in the infrastructure layer is about "who can enable AI to continuously evolve". The ceiling of the former is the model's capabilities, while the ceiling of the latter is the growth space of the entire AI industry. This is the logic behind the valuation gap.
Most AI companies or technology giants essentially sell single - point products or services. However, Musk's model is to build the infrastructure required for the future world in advance and then charge for using this infrastructure.
SpaceX controls the access to space, Starlink dominates the global satellite network, Tesla controls ground - based mobile data, and finally xAI processes and manages this vast amount of data. This is not just a product matrix; it is an all - around infrastructure monopoly.
While Sam Altman of OpenAI is focused on building the most powerful model, Google is using its cash flow and technological accumulation to advance its AI layout, and Anthropic is committed to the safety route... They are all struggling to adapt to the AI era, while Musk is defining how the AI era should operate.
While others are competing for market share in a certain field, he is designing how these fields are connected, who provides the power, and how the data flows.
When a person controls the infrastructure leading to the future, he has the right to define the future.
Conclusion | What's Truly Valuable
The combination of rockets, satellites, AI, manufacturing, and scenarios cannot be replicated by others.
Data is collected from cars, robots, and satellites. Computing power is solved in space, and hardware is mass - produced in factories. All these elements operate in a closed - loop system.
While others are making better AI products, Musk is building the infrastructure for AI operation.
The former sells services, while the latter charges tolls.
The market premium is for the right to define the AI era.
This is the gap.
📮 Original links:
https://www.youtube.com/watch?v=VvKV_ulseKc&t=7s
https://www.youtube.com/watch?v=XOBy9lOx2Do
Source: Official media/Online news
Typesetting: Atlas
Editor: Deep Thought
Chief Editor: Turing
--END--
This article is from the WeChat official account "AI Deep Researcher". Author: AI Deep Researcher. Republished with permission from 36Kr.