Path1
There are two stages of investing in Tesla. The first stage is to make money from selling cars; the second stage is to make money from software.
The first stage is close to Apple, redefining products, vertical integration, self-developed core hardware, low SKU explosion models, much like Apple, with super high gross profit margins, but not yet Apple.
In the second stage, it becomes Apple and surpasses Apple, distributes a large number of goods, has the highest market share, has an ecology, and is inseparable from use, and achieves maximum marginal benefits through software. Then integrate upwards beyond the industry, expanding from mobility to energy. Beyond Apple lies in having greater scale and market space.
From this point of view, the first stage is the foundation of the second stage, and the second stage is the maximization of the first stage, so the connection between the two stages and the things that are maintained and transmitted are the focus of investment.
What is always passing is: data
Data is realized by: Tools
Implementation tools are: technological innovation
These three points should be bolded, which is the essential difference between Tesla and other car companies.
Path2
What is the data? It is a complex code network of user behavior + environmental perception + decision-making +…etc. In this network, the driver’s daily behavior and Corner cases, environmental variables, and decision-making methods are extracted, processed, and finally become the means of production.
So in the second stage, what Tesla earns is the realization of data.
It can drive autonomously, sell insurance, sell in-car entertainment subscriptions, provide robotic labor, and create a virtual grid.
On the basis of the data network, he replaces you with a driver with almost no cost, and you pay him a salary. And turn your idle car into a taxi, into an energy storage station, prolong the service life of the car, let you make money, and then he will take a commission. After the car’s life is over, the car’s body becomes garbage, but the soul is eternal and unique. In the long company, he has learned from you, mastered your habits and is better than you, when you have a new electric car. When the soul re-inhabits, it is as natural as a clone.
But the current FSD is not good enough, not smart enough, and the essence is that the amount of data is not enough. Tesla needs more than 10 million fleet running data. When it is smart enough, it can completely replace human drivers and expand from cars to robots, from travel to all aspects of human life.
The sign to observe whether the FSD is intelligent enough is: FSD follows a car or follows a person. Walking with a person means that the training level is high enough, and a personality ID is bound to a user. But the current level of intelligence is far from enough to generate user stickiness, so FSD is following the car.
Path3
The tools for collecting data are cameras and radar.
It is the processors and neural networks that process the data.
It is easy to collect and can be purchased externally; it is difficult to process data, so it must be in your own hands.
FSD chips, Dojo, and neural networks are Tesla’s core technologies and moats. Without these, high-quality raw materials cannot be processed. If you buy someone else’s, there is the risk of data loss, the risk of insufficient purification, and the risk of excessive cost. So every OEM aspiring to develop into a new era is secretly preparing for chip development (I won’t say who it is).
But self-research is extremely difficult. The most difficult thing is not money, but technology accumulation. You must have a very high strategic vision, make future investments when there is chaos, and build brick by brick over the years, and the top masters and the two masons are different.
The current situation is that in the entire industry, this stunt is unique to Tesla, and not even a mason with two knives can see it.
Therefore, the logic of Tesla’s insistence on core hardware self-development and the pure vision route is: the pursuit of data purity, the pursuit of reusability, and the cost of data.
The Tesla brand plays from high to low, and the logic of selling cheap cars is: to expand the data denominator; of course, this is also the reason why all L4 companies are bound to fail. Without data, nothing can be discussed.
Path4
Gold diggers can’t give control of selling shovels.
Blacksmithing also needs to be hard on its own. To obtain high-quality and cheapest data requires a series of new technologies and engineering.
I plan to talk about five underlying technologies and see how Tesla achieves tool breakthroughs through technological innovation.
The first technology: electrical and electronic architecture
One function is one ECU, and the more functions, the more ECUs. After a certain level, the wiring harness, computing power, and energy consumption will reach the limit and cannot be expanded. This is also the reason why traditional cars cannot be intelligent.
Traditional Volkswagen needs 70+ ECUs, Audi needs 100+, Model S needs 44, model3 needs 31, modelY only has 26, and the model2 under development may be less than 15.
12V fuse, Volkswagen’s latest MEB platform has 77, Ford Mach E has 88, and Model Y has 0.
The general consensus in the industry is that the future architecture is cloud supercomputing + central processing unit + sensor + actuator. Currently, only Tesla has cloud central supercomputing.
(to be continued)
$Tesla(TSLA)$ $Xpeng Motors(XPEV)$ $NIO(NIO)$
There are 2 discussions on this topic in Snowball, click to view.
Snowball is an investor’s social network, and smart investors are here.
Click to download Snowball mobile client http://xueqiu.com/xz ]]>
This article is reproduced from: http://xueqiu.com/4437116803/225038466
This site is for inclusion only, and the copyright belongs to the original author.