Musk announced a great vision in 2006: to build increasingly cheaper electric vehicles to replace fuel vehicles and accelerate the world's transition to sustainable energy.

China is rapidly approaching this goal. The average selling price of new energy vehicles here has dropped from more than 200,000 yuan last year to less than 180,000 yuan, which is the same as that of fuel vehicles, and will continue to fall next year. New car-making forces such as Wenjie and other brands that originally targeted the 250,000-400,000 yuan market are also launching cheaper models. The Passenger Car Association estimates that for every 100 vehicles sold in China in 2024, 42 will be new energy vehicles.

However, Tesla is missing the change it helped initiate. Tesla will not have a new model in the Chinese market for at least the next 18 months. Its Model 3 and Model Y no longer have the price advantage they had when they were first launched. Tesla's only new model in the world, the Cybertruck, will produce only 125,000 units in 2024, and will mainly be delivered in the United States. The starting price of this future-looking pickup truck is about 430,000 yuan, 1.5 times the price when it was first announced in 2019.

Tesla’s previously announced new cars, Volkswagen models priced below 150,000 yuan, will have to wait until the second quarter of 2025 at the earliest to be mass-produced. By that time, Model 3 had been sold for 8 years and Model Y had been sold for 5 years, with only one round of facelifts that did not change the appearance.

A major sticking point that has put Tesla into a five-year model vacuum period is that the mass production of 4680 batteries has been delayed for many years.

Tesla releases 4680 battery in 2020. At the beginning of research and development, Musk believed that battery manufacturing efficiency was too low. Tesla could reinvent the battery manufacturing process, abandon conventional practices, and reduce costs by 50%.

The 4680 battery was originally planned to begin mass production in 2021, but it was not launched in small quantities until the middle of this year. Tesla's Texas factory in the United States has produced only 10 million 4680 batteries in the past four months, which is only enough to install 12,000 Cybertrucks.

It is understood that in the second half of this year, Tesla will start looking for Chinese battery companies to OEM battery pole pieces to meet production requirements. By the second quarter of next year, Panasonic will start supplying Tesla with 4680 batteries, but the production capacity will only be enough to load about 60,000 vehicles.

From Roadster to Model 3, and then to Model Y, Tesla has used "first principles" thinking on several generations of new models: that is, rethinking the conventions and stereotypes that most people are accustomed to, tracing back to the root cause, analyzing whether they are reasonable, and then Find new, simpler and cheaper solutions based on the principles of physics to accomplish goals that industry experts consider impossible. The 4680 battery is a continuation of this approach.

"The Biography of Musk" describes many times the process of capturing first principles. When making SpaceX rockets, Musk challenged authority and proposed replacing carbon fiber with cheaper stainless steel to make rockets. In the end, he only spent 2% of the cost of NASA's moon landing program to build Starship that can fly into space.

Tesla and other technology companies run by Musk always seem to be able to find new ways to rely on first principles, prove conventional wisdom wrong, and achieve technological leadership time and time again.

But on the 4680 battery, Tesla’s approach encountered obstacles. This battery determines the production capacity and pricing of Tesla's next-generation car, and its mass production time and performance have not reached the goals when it was initially released. This is the other side of the first principle that has been regarded as a treasure by many companies in recent years: when encountering complex innovations, starting from the principle and starting over is often just the starting point of a difficult journey.

Reinventing the battery and the battery factory

Musk proposed the “idiot index”: dividing the price of a part by the cost of the raw materials required for the part. The larger this number is, the "more idiotic" the component is. Either there are too many intermediate links or the manufacturing efficiency is too low.

Whenever it encounters a component with an idiotic index that is too high, Tesla will rethink the process and innovate manufacturing methods to reduce costs and bring the index back to "1" as much as possible. Musk seeks to make the manufacturing cost of a car infinitely close to the sum of the costs of steel, aluminum, silicon, lithium and other materials used in the car.

In 2007, after Musk checked the battery material prices on the London Metal Exchange, he calculated that the "idiot index" of the battery was 7: At that time, the cost of lithium, cobalt, nickel and other materials per watt-hour of the battery was only US$82, but the selling price of lithium batteries But it is more than 600 US dollars. This figure is the result of 20 years of hard work by battery companies such as Sony and Panasonic.

Tesla established a joint venture with Panasonic in 2014 to build a battery super factory, hoping to reduce battery costs. But by 2020, the "idiot index" of lithium batteries was still 2, and at that time the actual idiot index of the entire vehicle was less than 1.5. Tesla still has to sell the car for $40,000 (about 300,000 yuan, which is the selling price of the Model 3) to maintain gross profit, which Musk believes is not cheap enough.

Also in 2020, Tesla announced that it would develop and manufacture low-priced electric vehicles priced at about US$25,000 (approximately 150,000 yuan at the current exchange rate) to enter the more mainstream car market and directly compete with best-selling models such as Toyota Corolla.

Supporting this goal is a complete battery cost reduction plan. Since 2018, Tesla has established a project team code-named "Roadrunner" and began planning for self-research and production of batteries.

The original meaning of first-principles thinking is to question everything you can until only basic facts and principles remain. The first of the Tesla work laws is: "Question every request."

Looking at battery manufacturing from this perspective, Tesla changed the size of the cylindrical batteries around the goal of cost reduction, simplified the complicated process of wet production and drying that had lasted for decades, and designed new batteries and manufacturing processes.

Tesla chose a large cylindrical structure: increasing the size of the cylindrical battery from 21 mm in diameter and 70 mm in length to 46 mm in diameter and 80 mm in length. This is the origin of the name "4680" battery. A larger cylindrical structure can increase the proportion of energy material in a single battery, thereby increasing the battery energy density.

This design takes into account manufacturing efficiency and cost. Currently, the mainstream power batteries are divided into cylindrical and prismatic batteries. Cylindrical objects run faster in the assembly line than square objects. Square battery leader CATL can produce 25 cells per minute, while cylindrical battery leader Panasonic can produce 300 2170 cells per minute. The disadvantage of cylindrical batteries is that when packaged as a battery pack, there will be gaps between cylinders, and the space utilization rate is lower than that of square batteries.

Large cylinders can significantly reduce gaps. A R&D director of a battery company said that controlling the diameter of cylindrical batteries for vehicles to 45-50 mm can best balance battery capacity and space utilization. If the size is larger, the processing difficulty will increase, and the space utilization will also be improved. will become less.

"Increased size" may seem like a small change, but it will actually bring about a series of contradictory improvements. Just by making the battery bigger, the tabs, that is, the conductive parts that connect the internal and external circuits of the battery, will have to bear more current, which will make it easier for thermal runaway and increase safety risks.

"Since your ears are disobedient, throw it away." Musk said. Tesla immediately removed the battery tabs and instead used the entire battery bottom and casing to serve as tabs. This is the "full tab" (also known as "poleless tab") process, which can speed up battery charging and make the battery easier to charge. Heat dissipation - The battery casing is larger than the original small piece with protruding tabs, making it easier to dissipate heat.

Battery companies have been using wet processes for the past 30 years. They mix battery materials with toxic adhesives, liquid solvents and then apply them to thin foils. The processed pole pieces need to be baked in a 100-meter-long oven at a temperature of 90 degrees for 12 hours. During this process, toxic solvents and moisture are fully evaporated - the entire process greatly increases manufacturing costs.

Tesla believes that the wet process is inefficient: Since the pole pieces are to be made "dry", why do we need to wet the pole pieces first and then dry them? This is the idiotic part of battery manufacturing that Musk thinks. The equipment, labor, and factory building costs of wet coating account for 22.76% of the entire battery manufacturing.

In February 2019, Tesla spent US$219 million to buy Maxwell, a supercapacitor (electrical energy storage equipment used in camera flash and other fields) company, and switched the dry electrode process of supercapacitors to lithium batteries, directly making the pole pieces. Dry.

Dry electrodes do not use a liquid binder and therefore do not need to be baked, making them theoretically cheaper, faster and less damaging to the environment. Musk said that relying on this process alone, Tesla can reduce equipment expenditures per unit of production capacity by one-third and reduce the floor space and energy consumption of the electrode production workshop by 90%.

The current equipment expenditure for a prismatic battery production line is about 170 million yuan, while the equipment expenditure for the 4680 battery production line is only 50 million to 60 million yuan.

Tesla also hopes to speed up the battery assembly line to improve production efficiency. Musk envies the continuity and ultra-high efficiency of the beer and other beverage manufacturing industries and the paper industry: When making beer, there are no breakpoints in the production line. The beer bottles will not leave the production line until they are capped. However, after some parts of the battery are made, they are usually It is necessary to leave the production line temporarily, transport it to the next workshop by trolley, and then return to the assembly line. The fastest battery production lines now operate at 6 km/h, while the fastest beer production lines can reach 30 km/h.

To this end, Tesla has significantly merged its battery manufacturing processes and developed equipment that integrates multiple processes. For example, in 2021, Tesla launched a three-in-one machine for cutting, rolling and welding at its Berlin factory. These tasks originally required three different pieces of equipment to complete.

According to Musk in 2020, the entire 4680 battery solution can reduce battery manufacturing costs by approximately 20%, equipment investment costs by 35%, and factory floor space by 70%.

At that time, Tesla regarded the 4680 battery as the basis for large-scale expansion: using battery factories with less investment to produce energy storage and vehicle batteries, and then using cheaper batteries to manufacture cheap models priced at US$25,000 to stimulate sales and earn more The extra profits will then be invested in research and development and a new round of production capacity expansion, forming a growth flywheel to help Tesla achieve its ambitious goal of selling 20 million vehicles a year in 2030.

Perfect design, tough manufacturing

Make the battery bigger and get rid of the "idiot part" of getting the motor wet and then drying it. These obvious improvements have never been implemented in the battery industry over the past few decades because of difficulties.

The first problem is process transformation.

The automobile manufacturing process is relatively short and has low requirements for precision and environmental control. The main task of automobile manufacturing is the assembly of finished products. Improving one assembly process will rarely affect the previous and subsequent processes. Battery manufacturing is about turning materials into finished products, and the process is more closely related. The finished product of the previous process is the raw material for the next process. Improving a process means modifying the previous and subsequent processes as well.

In order for Tesla to use dry electrode technology and eliminate steps such as drying, it needs to make all links before drying "dry". This accounts for 50% of the entire battery manufacturing process, and the control of environment and precision is much higher than other processes.

The core of wet electrode is coating. Its task is like spreading butter, spreading the paste battery positive and negative materials with binder evenly on the metal foil, and the thickness of the application is generally only 30 microns. Battery companies such as CATL are committed to increasing coating speed and can now coat 100 meters per minute. The faster the coating speed, the more difficult it is to control quality. Behind the subtle improvements lies the experience gained from billions of attempts.

Dry electrodes also need to attach positive and negative electrode materials to the metal foil, but the positive and negative electrode materials are dry powder and have weak adhesion. It is not like spreading cream, but more like spreading sand, and it also pursues uniformity and speed.

In order to spread the sand evenly and stick it firmly, Tesla developed a new adhesive.

In 2020, Tesla applied for a patent for a dry electrode binder, which improved the PVDF binder originally used in lithium batteries. At a microscopic level, the new binder becomes fibrous when rolled, like a web. This makes sanding a flat surface “like sanding a marshmallow,” said one dry electrode process expert.

But at the time, Tesla didn’t know how much binder to mix into the positive and negative electrode materials: if the binder ratio was high, there would be less energy-carrying material in the battery, the energy density would be lower, and the binder would hinder The flow of lithium ions in the battery will shorten the battery cycle life; but the binder ratio is too low and the adhesion of the material is not enough.

An intuitive criterion for measuring the effectiveness of a binder is the first effect of the battery (the proportion of battery power in the design capacity when charged and discharged for the first time). An engineer who has disassembled the 4680 battery said that Tesla's sample in the middle of this year was able to achieve a first efficiency of 88%, while other battery companies participating in the project could only achieve 85%, but they have not yet reached the mass production of batteries. level. At present, the first efficiency of mainstream power batteries exceeds 92%.

"Based on this first effect, the Tesla 4680 battery can be cycled more than 1,000 times, but the current mainstream batteries can be cycled more than 2,500 times," he said.

Material research and development has only solved about 20% of the mass production problems. Next, there is equipment research and development.

Dry electrode equipment requires appropriate force to roll the binder in the material into a suitable fiberized state. In a laboratory setting, this is simple. But mass production requires equipment that can handle the entire task continuously and accurately.

Rolling needs to be carried out multiple times. If it is pressed only once, there will not be enough operating space to adjust the equipment parameters. Tesla's published patent uses three rollers to perform the rolling action in two steps. But a person close to the equipment supplier said Tesla later increased the number of rollers on the rolling equipment to seven.

More and more rollers have indeed raised the upper limit of accuracy, but they have also increased the difficulty of debugging. Every time the previous roller is adjusted, the parameters of all subsequent rollers will change.

Tesla, which often has bright ideas in the manufacturing field and dares to try, has to fall into this inefficient cycle. There are no shortcuts to debugging equipment and processes, it requires trying again and again. And it often affects the whole body, and changing one area requires adjusting multiple links. Removing the "extra link" of wetting the positive and negative electrode materials and then drying them is much more difficult than Musk initially imagined.

Tesla’s approach is to design its own equipment and then find a lithium battery equipment company for OEM production to help overcome some equipment problems.

An equipment supplier who has been in contact with Tesla said that Tesla will provide the core drawings of the equipment and restrict suppliers from modifying the design. "The end result is that the equipment engineers don't understand the process, Tesla people don't understand the equipment, and the equipment suppliers took a long time to build usable equipment, but it still doesn't meet Tesla's requirements," he said.

In 2021, Tesla found a number of battery equipment suppliers to manufacture equipment, and some of the equipment solutions have now been abandoned. It is understood that a leading lithium battery equipment supplier once provided Tesla with the entire 4680 production line equipment and sent an engineering team of nearly 50 people to Texas to assist Tesla in optimizing the production line. The team has withdrawn at the end of 2022.

Another difficulty with production equipment is the welding machine.

The original battery lug only has a small piece, but the 4680 battery adopts a full lug design, and the area of the lug that needs to be welded increases exponentially. The larger the welding area, the more likely it is to make mistakes. If the welding machine outputs too much energy, the welding will penetrate the pole lug. If the energy is not enough, the welding will not be strong. In either case, the battery will be scrapped.

"Tesla has not clearly defined what the welding effect should be until this year. Tesla is innovating from the source, but they are not sure about the details of implementation, so they spend money to find someone to help them realize their dreams." An equipment supplier According to sources, Tesla has not provided a better lug welding control solution. In addition, Tesla also encountered yield challenges in aspects such as laser sealing.

By the end of last year, Tesla's 4680 battery production yield was only 92%. According to battery industry calculations, the yield rate of 4680 batteries must exceed 95% to reduce costs and achieve commercial use.

The efficiency of Tesla's production lines is also far from industry expectations. An equipment manufacturer said that at the beginning of this year, the production efficiency of Tesla's 4680 battery was about 85 cells/minute. Previously, the industry believed that the upper limit of the efficiency of the 4680 battery was 350 cells/minute.

As Tesla's production lines run at faster speeds, quality control problems in the manufacturing process will continue to arise.

"Even if you miss even 0.001% of dust and debris, these things can cause a short circuit in the battery. The lab environment won't amplify the subtle possibilities, but the factory will. You will find new ways to fail all the time." Tesla battery project leader Drew Baglino said at the investor day in March this year.

Currently, Tesla’s 4680 battery manufacturing plan has not yet been finalized. "The problem is not only in production, but also in design modifications. Often the next version has to be started before a process is completed." said a Tesla engineer.

Only Tesla dares to build batteries like this

According to estimates by industry insiders, if you do not insist on dry electrodes, the Tesla 4680 battery can also reduce the cost of Model Y by about 8%, which means reducing the battery cost by 20%. Although this is less than half of Musk's goal, it is a remarkable achievement in the power battery industry. It will take Panasonic and CATL at least three years to achieve similar cost reduction results.

Tesla insists on developing the dry electrode process, even delaying the delivery of Cbyertruck and next-generation cars. In Musk's vision, the 4680 battery will not only serve Tesla's current sales of 2 million vehicles per year, but will also support the future sales of 20 million vehicles per year.

Only cheaper batteries can support Tesla's goal of building a $25,000 affordable model. In March this year, Tesla completed model development and broke through larger-scale one-piece die-casting technology. The 4680 battery was the final stumbling block.

In the automotive industry, only Tesla invests in technology ahead of time at all costs and binds its business expansion to technological breakthroughs. Both benefits and risks are magnified.

Tesla's previous successes and current encounters all stem from the methodology derived from first principles: abandoning industry conventions, questioning original requirements, and rethinking how to do something based on the most essential physics.

The situation where this methodology works is often when Tesla wants to challenge and subvert the old practices or habits produced in a certain technological stage, and now there are new technologies that can circumvent the previous constraints. Musk keenly saw that the limitations of the past no longer existed, breaking through the invisible "impossibility" in the minds of senior experts and reaching the solution.

This requires technical intuition and judgment. But it is difficult for Musk or any individual to objectively and comprehensively assess the overall technology and engineering level of a period.

In some cases, Musk's ideas just found technical and engineering support: when he thought of integrated die-casting from toy manufacturing, there happened to be a company on earth that could manufacture a 6,000-ton die-casting machine - Lijin.

In the automotive industry, only Tesla invests in technology ahead of time at all costs and binds its business expansion to technological breakthroughs. Both benefits and risks are magnified.

Tesla's previous successes and current encounters all stem from the methodology derived from first principles: abandoning industry conventions, questioning original requirements, and rethinking how to do something based on the most essential physics.

The situation where this methodology works is often when Tesla wants to challenge and subvert the old practices or habits produced in a certain technological stage, and now there are new technologies that can circumvent the previous constraints. Musk keenly saw that the limitations of the past no longer existed, breaking through the invisible "impossibility" in the minds of senior experts and reaching the solution.

This requires technical intuition and judgment. But it is difficult for Musk or any individual to objectively and comprehensively assess the overall technology and engineering level of a period.

In some cases, Musk's ideas just found technical and engineering support: when he thought of integrated die-casting from toy manufacturing, there happened to be a company on earth that could manufacture a 6,000-ton die-casting machine - Lijin.

The 33-engine parallel propulsion solution used by SpaceX is also an idea that former Soviet scientists tried but failed in the 1970s. Rocket engines will interact with each other when pushing flow, and engine settings need to be adjusted in real time. Computers at the time could not control so many engines simultaneously. By 2023, when SpaceX implements this idea, computer computing power and algorithms will be able to complete complex tasks—the computing power of mainstream AI chips is one billion times what it was 50 years ago.

In each of these examples, Musk went back to square one and rethought habits that had been accepted unthinkingly for years. In the long-established automobile and aerospace industries, large manufacturers pursue the ultimate in production efficiency along established production processes day after day, and it is difficult to have the awareness or room to think outside the box.

The limitation of first principles is that they cannot transcend the limitations of the times. When Tesla's goals cannot be achieved by current technology, it also has to invest a huge amount of time and money to make slow progress.

Tesla had a stumble when it mass-produced Model 3 in 2017: Musk believed at the time that manpower could be completely replaced with robotic arms, which was inevitable in the manufacturing industry. But he overestimated the level of the entire automation technology. The robotic arm could not even do simple combing of wires. Tesla immediately fell into production hell, and Musk finally got out of the crisis by calling workers back to the factory.

Battery manufacturing is also such a field: it involves multidisciplinary and complex tasks, and the processes are interlocked and affect the whole body. “Coating with wet positive and negative electrode materials and then drying them” was a step that Musk initially considered unnecessary. But deleting this link requires overturning the entire process from mixing, coating, die-cutting to drying.

Even if Tesla can complete the above process in the laboratory, it will still have to solve the two mutually reinforcing problems of mass production and commercialization. One million electric vehicles correspond to more than one billion cells, which requires repeating the production process billions of times on the same set of equipment, reaching a certain yield baseline, and controlling costs.

Many industry insiders predict that Tesla will mass-produce 4680 batteries in 2025, but the performance of the final mass-produced version will be significantly reduced compared to the version released in 2020. The energy density of the 4680 batteries currently produced by Tesla is only 265Wh/kg, which is nearly 20% lower than the industry forecast of 330Wh/kg. The Cybertruck equipped with this battery has a maximum range of only 547 kilometers, which is far lower than the previous target of 800 kilometers.

From the release of the 4680 battery to the present, Tesla has only achieved 1/3 of its original goal in 3 years. It has mass-produced dry-process graphite anodes with a simpler process, but there are still more difficult dry-process silicon anodes and cathodes.

From the perspective of cruel business competition: slow, sometimes unacceptable. Tesla's product rhythm has been disrupted. At a time when it is most important to pursue victory, Tesla’s new cars have been out of production for several years.

From the perspective of technological evolution, Tesla has explored the way for the entire industry: the 4680 battery has unified the size standards of large cylindrical power batteries; currently, battery giants such as CATL have also begun to explore the dry electrode process and expand this process concept to batteries Diaphragms and other components. The 4680 battery may be driving a technological revolution in power battery manufacturing.