news

Building batteries is harder than building rockets: Who is stopping Musk?

2024-08-21

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina


Tesla's only "Battery Day" so far was held in September 2020. Due to the epidemic, Battery Day was postponed by 5 months from the original date. The protective measures for participants were also novel-watching the event while sitting in electric cars in the parking lot.

During the five-month delay, the global new energy vehicle market exploded, with annual sales exceeding 3 million for the first time[1]. At the same time, the Shanghai Gigafactory began to increase its production capacity, breaking production records on a monthly basis, making this Battery Day a hot topic.


At the Battery Day, everyone was watching from the car

Not only is the way of participation ingenious, but Musk's choice of location is also very particular. The Fremont factory is Tesla's first vehicle assembly plant, and two years ago, Musk celebrated his 47th birthday here.

At that time, Model 3 was facing production hell. Musk personally led the team to "struggle for 21 days" and built a semi-automatic temporary assembly line in the parking lot to overcome the crisis with difficulty.

On the day of "Battery Day", Tesla executive Drew Baglino reviewed the days of struggling with his boss and summarized the theme of Battery Day in Musk's tone:

“To build the best cars in the world, we design the vehicles from the ground up in our factories. Now we’re going to do the same with the batteries.”

Afterwards, the real protagonist of this press conference, the "4680 battery", officially debuted.


As the name implies, 4680 batteries refer to cylindrical batteries with a diameter of 46mm and a height of 80mm. Compared with the previous generation 2170 batteries (i.e., a diameter of 21mm and a height of 70mm), the energy density is greatly improved, which can significantly reduce the production cost of vehicles with the same driving range.

Among Musk's many big plans, the 4680 battery is the one that is closest to realization. Today, almost all the black technologies on the PPT have been realized. At the financial report meeting in early 2023, Musk said that 2024 will be the "4680 year".

However, in April this year, Drew Baglino, a highly emotional executive who unveiled the 4680 battery at Battery Day, unexpectedly resigned and became one of the highest-ranking employees on the 10% optimization list in Tesla's global layoffs.

Drew Baglino has worked at Tesla for more than 18 years. After co-founder JB Straubel left in 2019, Drew Baglino became the de facto CTO and the leader of 4680 batteries and R&D and production.

So, what changed Musk?

Determination to break up

In June 2019, CNBC exposed a mysterious office building in Fremont, California, which is only a seven-minute drive from Tesla's Fremont factory. Except for the "Tesla" logo that can be seen everywhere, the outside world knows nothing about what happens in the building.


According to CNBC, this is Tesla's "secret battery lab," and it gave a very provocative title:Tesla is trying to produce its own batteries to break away from its reliance on Panasonic.


Panasonic was Tesla's most staunch supporter in its early days. Not only did it generously invest $30 million on the eve of the latter's listing, it has also been Tesla's most core battery supplier.

However, CNBC's report revealed one of the reasons why the 4680 battery project was launched:The alliance between Tesla and Panasonic has long been rifty.

Musk has been complaining about Panasonic’s battery production capacity since the Model S was put into mass production. When it came to the Model 3’s production capacity crisis, Musk publicly said that it was Panasonic’s battery production that was too slow. At the time, US media reported that it took seven hours to produce a set of 2170 battery packs, and the scrap and rework rate of the produced batteries was as high as 40%[2].

In addition to the issues of production capacity and yield, the substantive disagreement between the two sides is that the Shanghai Lingang Super Factory is about to be completed. Considering the battery transportation costs and the target production capacity plan of 500,000, Musk hopes that Panasonic will make additional investments and go to Shanghai together to build a battery factory and do a big job.

However, Panasonic's battery business has been losing money with Tesla for many years, and it hopes that Tesla will help Panasonic recover its investment as soon as possible. Therefore, Panasonic said it was willing to go to China with Tesla, but it has not taken any substantive action.

The widening rift with Panasonic has strengthened Tesla's determination to produce its own batteries.

Another big background of the 4680 battery project is another big pie that Musk painted on Battery Day:Within three years, a cheap car priced at US$25,000 (approximately RMB 178,000) will be launched.


In pure electric vehicles, power batteries account for about 40% of the total vehicle cost, and are seen by Musk as the biggest obstacle to the emergence of cheap models.

According to Musk's idea, a cheaper model than Model 3, the rumored Model 2/Q, is crucial. Cheap models can not only expand Tesla's car sales, but also be an important weight to support the growth of its software subscription revenue.

The first person to understand Musk was Cathie Wood, founder of Ark Fund. In an open letter in 2018, Cathie Wood disclosed her valuation logic for Tesla:Tesla will transform from a car manufacturer to a "Mobility as a Service (MaaS) company."

MaaS service refers to Tesla's software business built around autonomous driving algorithms, including subscriptions to the assisted driving function FSD, Robotaxi (autonomous driving taxi) business, and so on.

Among them, the most important FSD subscription revenue is combined with super charging station and other businesses as "services and other revenues" in Tesla's financial report.

For Tesla, even if only 1% of car owners purchase FSD software, as long as the car sales are high enough, the profit will be very considerable. The cheap car of $25,000 is the best weapon to expand sales. As the number of FSD subscriptions increases, the R&D costs will also be diluted, further increasing the profit margin.


So, on that Battery Day in 2020, whether it was getting rid of dependence on Panasonic, further reductions in battery costs, or new $25,000 models, they all depended on a clear goal:

Developed and mass-produced 4680 batteries.

Cost Game

The core indicator for measuring battery cost is the cost per unit of electricity, that is, the cost per kWh.

No one understands this better than the bargaining maniac Musk. According to a report released by UBS in 2018, the cost of Panasonic's 2170 batteries supplied to Tesla was only US$111/kWh, the lowest in the industry[4] and at least 25% lower than LG Chem and CATL.

Musk has an almost pathological obsession with controlling production costs. According to the Wall Street Journal, Musk is accustomed to calling Panasonic President Kazuhiro Tsuga directly to ask Panasonic to lower prices[9]. The latter complained about Musk[9]:

Elon asked for a price reduction many times, and I responded to him once that if he did this again, we would consider removing Panasonic employees and equipment from the Gigafactory.

On the day of "Battery Day", Tesla announced the goals of improving the capacity, range, power, etc. of the 4680 battery cells as usual, but Musk went a step further and openly disclosed how Tesla plans to "reduce the cost per kWh by 56%" and gave clear quantitative standards:

14% depends on design, 18% on production technology, 17% on positive and negative electrode materials (5% negative electrode, 12% positive electrode), and 7% on battery and automobile integration technology.


Design refers to the volume and structural changes of the battery itself.

Since the 4680 battery is larger than the 2170 battery, the capacity of a single battery has increased, so a battery pack (cell) with the same capacity requires fewer batteries.

According to estimates, the battery pack of the Model 3 long-range version requires 4,400 2170 battery cells, but if it is changed to 4680, only 960 cells are needed, which corresponds to a significant reduction in the number of structural parts and welds, thereby reducing costs.


2170 battery and 4680 battery

Battery and automotive integration technology,Refers to the integrated die-casting technology of the car body and battery cell.

The current mature solution is three-piece:The body is divided intoTwo front and rear parts and a middle part for placing the battery are cast separately and then assembled together.

Musk said why couldn't the three parts be cast together, allowing the battery to become part of the car's structure like an airplane fuel tank is welded to the wing?According to Musk, this solution reduces the number of parts by 370 and has the most direct impact on reducing the cost of the entire vehicle.

The adjustment of positive and negative electrode materials tests the laboratory level.

The positive electrode has a higher nickel content and a lower cobalt content because nickel metal (such as screws) is cheaper than cobalt metal; the negative electrode is doped with silicon, which not only increases the battery capacity, but the pure silicon negative electrode is also cheaper than the current mainstream graphite negative electrode (US$1/kWh).

The production process revolves around electrodes, which is also the biggest challenge in the mass production of 4680 batteries.

One is to remove the tab design and use the entire battery bottom and shell as the tab.In this way, there is no need to take the semi-finished battery products off the assembly line, weld the tabs and put them back. The assembly line can never stop, and there is no need to go off work as long as the power is not cut off.


Battery tab and tab-free design: Balder but stronger

The second is to develop dry electrode technology.The current mainstream wet electrode technology is to mix the positive and negative electrode materials with a variety of solvents and then coat them on the electrode. Since some solvents are highly toxic, they need to be evaporated by baking. This "oven" is 100 meters long and takes up to 12 hours to bake[2], which makes Musk, the efficiency freak, grit his teeth.

As the name suggests, dry electrode technology removes the thorn in the eye of "wetting and then drying" and "sticks" the dry positive and negative electrode materials on the electrode. It reduces the unit production capacity equipment expenditure by 1/3, and reduces the workshop floor space and energy loss by 90%[2], saving both time and rent.

Within the three-year deadline set by Musk, Tesla had solved almost all of its production challenges:

In September 2021, Musk announced that he had verified the performance and life of the 4680 battery[6]. This shows that Tesla has achieved its goals in both battery design and adjustment of positive and negative electrode materials.

In January 2022, Tesla produced 1 million 4680 batteries in its pilot factory, enough to supply 1,000 Model Ys. In October 2023, battery production capacity jumped to 20 million.

The only problem is the dry electrode technology.By the end of 2023, Tesla has actually released two generations of 4680 batteries, but neither has fully implemented dry electrode technology:Only the negative electrode is dry, while the positive electrode is still a traditional wet electrode.

Electrodes are the most expensive part of battery cell manufacturing, and they also receive the most KPIs in the cost reduction task. As long as dual-dry electrodes are not put into use, Tesla will be further away from its 56% cost reduction goal.

The biggest difficulty facing dry electrode technology is that there is no ready-made equipment, so one can only "learn by doing, and do by learning". If the effect is not good, the parameters must be adjusted. If the parameters are not right, the equipment must be changed. If the equipment is not changed well, the supplier must be changed. Even so, with the official rollout of the first CyberTruck equipped with dual dry electrode 4680, all doubts in the industry about 4680 batteries have disappeared.


Although there is still a distance from the cost reduction target of 56%, after the technological breakthrough, it is only a matter of time before the production capacity and yield rate increase.

After overcoming all the difficulties, Tesla is faced with a ridiculous problem:Is it still important for Tesla to produce its own 4680 batteries?

The East Asian model loses to tricks

One of the core reasons for launching the 4680 battery project was to get rid of dependence on Panasonic and reduce battery costs.

However, when Tesla completed all the OKRs for the project, it was embarrassed to find that all the problems it faced that year had been solved by someone else.

First, the supply chain structure: With the start of construction of the Shanghai Lingang factory, Tesla has introduced LG and CATL, and Panasonic only supplies batteries to the US factory. This supply chain structure is very stable and safe for Tesla.

Meanwhile, Panasonic and LG are also developing 4680 batteries, and are about to put them into mass production. Although CATL has not yet disclosed the progress of 4680 batteries, it launched the "hexagonal warrior" Kirin battery at the end of last year, which is said to have achieved considerable cost reduction.

More importantly, during the four years that Tesla has been working on the 4680 battery,Power batteryThe cost has dropped precipitously.

Although the price of lithium carbonate, a raw material for power batteries, soared in 2021, it has started to plummet since last year. In addition, the years of experience of Chinese battery factories, which have been frantically expanding production and account for 60% of the world's production capacity, have combined to push battery costs down.


Musk's idea for solving the cost problem is to "reduce costs using the most cost-effective method." Ten accountants would need a month to calculate the accounts, but Musk would have fifty programmers write a program to automate the accounting process and complete it in one day.

From the Model 3 super factory to SpaceX's starship, this idea has been successful. But when it comes to reducing battery costs, Musk has somewhat underestimated the diligence and bravery of the East Asian people.

Due to the huge downstream terminal demand, Chinese battery manufacturers have adopted the simplest cost-cutting idea in the manufacturing industry: spreading costs through scale.

At the beginning of this year, there were rumors that CATL was promoting lithium iron phosphate batteries with a price not exceeding 0.4 yuan/Wh to car companies. Based on the calculation that Tesla's 2170 battery cost (111 US dollars/kWh) dropped by 56% in 2018, Tesla's 4680 battery cost target is at least below 0.35 yuan/Wh. Even if Tesla achieves the goal, the gap between the two sides will be minimal.

At the Tesla shareholders meeting in June this year, after receiving a $46 billion salary package, Musk proposed a small goal for the 4680 battery team:Costs were on par with suppliers at the end of the year.

The battery cost cannot be significantly different from that of mainstream suppliers, which further affects another core reason for the establishment of the 4680 battery project:A cheap model at $25,000.

In Musk's conception, low-cost 4680 batteries are the key to bringing the price of new models to $25,000. However, times have changed, and Chinese counterparts can produce 100 cheap $25,000 models a year, and some Chinese automakers can iterate twice a year.

BYD, which was once mocked by Musk, has seen its gross profit margin increase to 28.1% while the average unit price of its cars has dropped to 140,000 yuan, surpassing Tesla.

Whether it is the battery factory that is struggling for 100 days, the fierce price war of electric vehicles, or the departure of Drew Baglino, the head of the 4680 battery core project, they all seem to convey a message:It is no longer that important to produce 4680 batteries in-house.

end

On Battery Day in 2020, Musk made no secret of his ambitions:Make Tesla the world's strongest manufacturing company.

However, this Battery Day has become Tesla's only Battery Day so far. In the following years, the annual "AI Day" took over the public opinion, witnessing the successive appearances of the FSD chip, Dojo training cluster, end-to-end autonomous driving algorithm and the humanoid robot Tesla Bot.


Tesla Bot makes its “real person” debut at AI Day 2022

In the first quarter of this year, Tesla's automotive business fell 13% year-on-year. Facing investors who felt they had made a mistake, Musk responded: "If you only see Tesla as a car company, then your understanding of Tesla is one-sided."

Compared with the casual updates of the "S3XY" series and the frequent delays of mysterious models, Musk's plans in recent years have clearly been moving closer to the field of artificial intelligence. From the live broadcast of the v12 algorithm to the rumors of FSD entering China, from the first release of robotaxi on Twitter to the robots entering the factory to work next year, Tesla is moving further and further away from the "manufacturing industry".

Wall Street's valuation logic is probably just superficial. The reason that is closer to the truth is probably that, whether it is the whole vehicle or key components such as batteries, production and manufacturing capabilities are no longer Tesla's advantage. Instead, Chinese suppliers have gradually taken away the voice in the supply chain by relying on their strong production capabilities.

In the field of autonomous driving and artificial intelligence built on chips, algorithms and software, Silicon Valley's talent supply and accumulation of software and hardware research and development are both in a clear leading position.

As the helmsman of an American company, Musk obviously knows better than anyone where Tesla's moat should be dug.


References

[1] The Global Market for Electric Vehicles in 2020,Canalys

[2] The difficult birth of 4680. Only Tesla dares to make batteries in this way. LatePost

[3] Three Questions and Answers about Tesla’s Fully Self-Driving (FSD) System, CICC

[4] Tesla batteries prove best value in class,UBS

[5] Tesla Giga Texas’ Model Y front megacast is as intimidating as it is revolutionary,Teslarati

[6] Is Elon Musk Back In “Production Hell” With Tesla’s 4680 Battery?IEEE Spectrum

[7] A $1 billion lawsuit reveals Tesla’s path to self-help with lithium batteries, Jianyue Auto Review

[8] Success with Tesla, failure with Toyota: Panasonic battery revelation, Yuanchuan Research Institute

[9] Tesla Needs Its Battery Maker. A Culture Clash Threatens Their Relationship, Wall Street Journal

Editor: Li Motian

Visual design: Shu Rui

Editor-in-charge: Li Motian

Cover image from ShotDeck