Summary: Analysts are doubting Tesla's ability to achieve industry standard benchmark on COGs based on the perception of chaos of recent activities
Critics claim Tesla has a split personality when it comes to production capability which they can't reconcile. Good at product development, horrible at general assembly.
These conclusions of chaos are premature if you don't understand the operating system.
Can Tesla Achieve Industry Standard COGS?
This is the thesis of this article. Tesla's profitability relies on competent manufacturing efficiency that can achieve standard industry cost benchmarks. It won't matter that Munro & Associates says the Model 3 should have an 18%-36% gross margin at 10K/wk, if Tesla can't reach industry standard costs and efficiencies to achieve those margins.
But what is causing doubt that they can achieve these COGs? Analysts have criticized the way Tesla has been operating its manufacturing facilities in Fremont and Nevada. From the bottle neck in battery packs at the Gigafactory to the general assembly bottlenecks and the Sprung structure at Fremont. Both bulls and bears are perplexed at how on one hand Tesla can produce state of the art, high density, integrated electronics and drive trains, yet, on the other hand, has not been able to handle the more basic aspects of production that have been around for decades.
Tesla seems to have some sort of a split personality when it comes to execution... or does it? This article is a direct response to the bear thesis that Tesla isn't capable of achieving the gross margins verified by 3rd party analysts. The media is reporting that cost overruns are due to incompetence, but my research has turned up something that explains it as part of an investment with an unfinished "work backlog". It also reconciles these seemingly contradicting personalities under a unifying strategy and contradicts the claim of incompetence.
Tesla Playing Catching Up On A 100-Year Lead
Tesla's mission is to accelerate the world toward sustainable energy. This is a daunting task that requires competing against legacy automakers, some of whom have a 100-year-old head start.
So how do you catch a 100-year lead? It appears that Tesla is using a development system known as Scrum. A concept co-founded by Jeff Sutherland and J.J. Sutherland specifically designed to accomplish "twice the work in half the time". When agile development teams use Scrum for hardware it is referred to as Extreme Manufacturing, according to its developer Joe Justice.
Clues That Led To The Scrum Hypothesis
Scrum is the most popular project management methodology, with 68% of agile software development teams using Scrum. I won't go into the specifics of how Scrum operates as this is not the focus of the article, but I will explain a bit about its characteristics and benefits. If you'd like more of a deep dive into it, you can learn more here.
In analyzing the outcomes of Tesla's production ramp progress, I recognized a few things that appeared similar to the characteristics of a Scrum project management methodology. I saw these as clues that would ultimately lead to my hypothesis. Clues included weekly production sprints that result in burst rates that are later sustained after several weeks. Other curiosities were the way Musk spoke about productizing the factory on the Q4 earnings call as well as having it move through iterations back in 2016.
""Our internal code name for the factory, the machine that builds the machine, is the alien dreadnought," Musk said on the call. "[When] our factory looks like an alien dreadnought, then we know it's probably right."
Musk said with Model 3 production, the factory will be at "alien dreadnought 0.5" in terms of advancements made to the production line. "And then it will take us another year or so, I don't know, summer 2018, to actually get to alien dreadnought version 1," Musk said."
There were also several media headlines that also provided clues. A Verge article titled, "Tesla keeps changing the way it builds its model 3". What kind of companies keep tinkering with the actual production method? There was also the NY Times article, "Can Elon Musk and Tesla Reinvent the way cars are made?" The common narrative was that there was only one way to build cars, but it appeared that Tesla was purposely ignoring it.
Scrum For Hardware Confirmed
I began researching to see if there was any merit to the hypothesis that they were using Scrum. That's when I came across a key note presentation by Joe Justice, the President of Hardware for Scrum Inc., entitled, "Scrum: Disrupting the Automotive Industry"
This is when I discovered that Scrum was being used successfully for hardware, and not just software. It also had a specific name. It was called Extreme Manufacturing. I highly recommend watching the entire video.
To my surprise, the presentation cited Tesla, starting at the 30:00 min mark of the video, to describe how Tesla's factory is, in fact, a "build server" that can produce iterations of a vehicle with each product development sprint. We saw this with the Model 3 getting improved suspension, wind noise, and back seats from the earliest versions.
Out of my general research on Scrum for Hardware, I learned that Tesla is in fact a pioneer of this application.
Applying Extreme Manufacturing to Tesla
So how does knowing that Tesla uses Extreme Manufacturing explain the contradictory behavior I introduced at the beginning of this article? Well, a basic Scrum workflow starts with a wish list. In the case of the model 3 design, the wish list included 25% profit margins at a specific performance level and battery range. The product development team is given the freedom to figure this out without being told specifically how to achieve it as mentioned in the video.
"requirements for innovation: specify why and for who but not how...
...this lets the build server attempt breakthrough methods and radical innovation, as they are not constrained by how to build but clearly measured on the compliance and the goals of the output."
This led to a task backlog that included a $150/KWh battery pack, centralized touch control interface and extremely high integrated circuit density to reduce component costs, and a 400HP motor to achieve performance, along with all the sub tasks to achieve those goals.
Ok, so this explains the technological proficiency, but what about the production ramp? If we apply the same Scrum workflow to the factory, we have a wish list that specifies a specific volumetric production density and output speed. That's pretty much it.
It was up to the designers to figure out how to cram as much vertically integrated production capacity as they can into the factory while at the same time having a goal of doubling the current assembly line speed of the best factories. The task backlog included such items as a certain percentage of automation and a reduced number of operations which links directly back to product designed for automation.
One of the key characteristics of Scrum are regular releases of functional iterations. We saw this when the line went through several shutdowns while process improvements were made. Successive burst rates were followed by sustained rates. Manual operations became automated and some automated operations became manual. Some robots were replaced, some were reprogrammed and all this happening because of a deliberate and strategic application of Scrum's backlog task burn down.
What seems to be chaos and disorder are actually weekly sprints with backlog grooming that tweak and fine tune the selected strategies. We saw that with Flufferbot and the GA4 tent.
The tent itself was also not the first iteration of this structure as there were already three other Sprung structures already in use at Fremont with the second largest one almost half the size of GA4.
We also saw real time updates to the production process itself, including streamlining welding and testing requirements as mentioned in the roadshow article, "Tesla changed its testing regimen for the Model 3, and that's OK - Roadshow"
In short, Agile Scrum sprints lead to iterations and versions whether it's the software that runs the car, the car components themselves, or the machine that builds the machines. They all have a wishlist, a backlog of tasks to achieve that list, production sprints producing functional iterations, and a sprint review to refine the task backlog for future iterations.
Deloitte reports that a network of teams is the future of organizations. The team of teams model already exists in a Scrum of Scrums, a.k.a. Scrum at Scale. I won't go into the details of this concept, but it basically scales the Scrum methodology to the organizational level which we see Tesla doing with its layoffs, and organizational restructuring as the company itself is going through iterations as it fine tunes its backlog of organizational tasks.
The narrative of Tesla's production incompetence and organizational chaos has led to speculation by analysts that 3rd party verified gross margins cannot be met since Tesla is incapable of achieving industry standard cost benchmarks. The onus is not on Tesla to prove that it can achieve industry standard production cost benchmarks because there is no data to support that they aren't achieving them now. If the only argument that they can't is anecdotal evidence from production delays, this is easily explained by Scrum for Hardware a.k.a. Extreme Manufacturing. This explains previously irreconcilable management strategies and decisions that, at first glance, look like chaos and incompetence. It's not called Scrum because it looks pretty, it's called Scrum because it gets the job done.
Disclosure: I am/we are long TSLA.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.