[0]: https://thedriven.io/2021/01/12/the-factory-is-the-product-m...
One thing I've noticed that the owners' or CEO's personality tends to permeate throughout the organization, for better or worse.
0. https://www.uthsc.edu/its/business-productivity-solutions/le...
Recommended podcast episodes about Deming: https://smallbatches.fm/search?query=deming
Can't argue with SpaceX, though - an engine a day is an insane pace.
Talk to an actual industrial engineer and they can tell you how laughably bad some of Tesla's manufacturing practices. You're never going to catch up to Toyota or Honda in pure speed or quality.
But Tesla has figured out a ton of clever ways to cut corners (mostly by having a customer base that accepts a very tiny static lineup of car models) and have gone all out on economies of scale.
> "What's driving Mr Terashi's effort is the EV's faster-than-anticipated takeoff and rapid-fire adoptions of cutting-edge innovations by Tesla and others," one of the people said.
[0] - https://www.reuters.com/business/autos-transportation/exclus...
Go and actually watch the disassembly and analysis done by Munro engineers. They do detailed analysis and costing. And Tesla is routinely doing things nobody else is doing that is highly effective.
Their use of aluminum casting is industry leading. Their electronics are industry leading by a wide margin. They are great at working with large glass panels. Their battery packs are industry leading. Their heating-cooling loop is fantastic. I could go on.
Just go and look a Tesla looks under the frunk and then look at how a VW ID.4 looks.
That doesn't mean they are the best at everything. Their suspensions are not as good. Their fit and finish is not as good. Again, there is more one could mention.
To just flat say that Tesla it worse at manufacturing the their competition is simply not true. If it were true they could never achieve anywhere close the margin they got. And this margin is achieved with pure EV, something most companies barley make profit on.
Tesla is also amazing at iteration. They are iterating faster then Toyota. In 2017 Model 3 had a lot of issues, if you look at Model 3 in 2021 the internals are very different.
> But Tesla has figured out a ton of clever ways to cut corners (mostly by having a customer base that accepts a very tiny static lineup of car models)
A the old 'all their costumers are idiots' defense that can't be backed up by anything substantial.
> and have gone all out on economies of scale.
They are pretty small car company still in terms of volume. And centralizing and vertically integrating production is a valid and effective industrial practice.
They had better margin then the competition while having 1/2 or less production on a architecture so just saying 'economics of scale' is just nonsense.
> They are iterating faster then Toyota. In 2017 Model 3 had a lot of issues, if you look at Model 3 in 2021 the internals are very different.
To put it in perspective in these 4 years, Toyota iterated for every single model year across 14 different model. And on top of that released all new platforms for the Yaris, the Camry, the Rav4, the Corolla, the Land Cruiser, the Tundra, and the Avalon.
This actually undermines your point to a certain extent. Tesla is great at iterating but that often comes at a cost to quality. It is very hard to manage reliability when parts and suppliers are constantly churning.
"Older models typically fare better in reliability, as companies tend to make tweaks and redesigns to solve known problems, while sticking with the same parts and suppliers.
But Tesla deviates from this approach, Fisher explained. “At almost random times during the year Tesla will switch major components, suppliers or sensors and other units. The more you change, the greater the chances you’re going to have some problems.” [1]
[1] https://www.cnbc.com/2021/11/18/consumer-reports-2021-auto-r...
"5. Improve constantly and forever the system of production and service, to improve quality and productivity, and thus constantly decrease costs." - Edwards Deming
https://en.wikipedia.org/wiki/W._Edwards_Deming
This does make version and revision management and service harder, but that you can manage with software and its worth the cost.
However, what you say is partially true, its not all positive. If you endlessly make the same car with very little change you can also hit very high quality and reliability. That is why sometimes cars like Jeep look good in quality rankings.
Overall however, I think there is good reason for iteratively interdicting changes.
What you're saying is true, with one important distinction we may disagree on:
>"but that you can manage with software"
This is a take that seems accurate on it's surface but belies a lot of additional problems. Leaning on software for improving reliability can be a dangerous path. It's tempting because it's cheap and easy to iterate on. Software doesn't wear out like mechanical components. But software makes reliability pretty tough when you constantly iterate because it tends to fail at interfaces. Not only do the number of these interfaces tend to grow exponentially, they are notoriously difficult to test fully. Nancy Leveson has some good articles that demonstrate this distinction.
Since this thread is about aerospace, you can see where software failure caused catastrophic failures by lulling engineers into complacency about how it can enhance reliability. Ariane V, to Falcon, to CST-100 all have examples of how software can not fulfill its promises to increase reliability and often it does the opposite. Don't confuse the ability to move a design change faster with an inherent ability to make things more reliable.
>"but that you can manage with software"
This is a take that seems accurate on it's surface but belies a lot of additional problems. Leaning on software for improving reliability can be a dangerous path. It's tempting because it's cheap and easy to iterate on. Software doesn't wear out like mechanical components. But software makes reliability pretty tough when you constantly iterate because it tends to fail at interfaces. Not only do the number of these interfaces tend to grow exponentially, they are notoriously difficult to test fully. Nancy Leveson has some good articles that demonstrate this distinction.
A automotive example from my own experience is a comparison with mechanical vs. electrical robots used in the stamping process of body panels. Plant managers like electrical retrofits because they held the promise of being much cheaper to build and modify. But they needed near-constant tweaking of the control system. The number of interfaces causes a lot more failure modes and impacts the reliability. By contrast, a mechanical system is, by it's nature, constrained by mechanical linkages. Once the mechanical system is tuned and working, it works for a long time. Gradually the parts wear, but they tend to be much easier to predict and troubleshoot.
Since this thread is about aerospace, you can see where software failure caused catastrophic failures by lulling engineers into complacency about how software can enhance reliability. From Ariane V, to Falcon, to CST-100 all have examples of how software did not fulfill its promises to increase reliability and often it does the opposite. Don't confuse the ability to move a design change faster with an inherent ability to make things more reliable.
What I mean by 'manage with software' is that each car has an exact description of what hardware is in it. Meaning what revision of every part. Tesla has some advantage because they manage the process end to end, knowing exactly what vehicle was sold to what costumer and the control the service center and write the diagnostic software so the service center should know exactly what car they are working on and what potential problems each revision of the software has.
In the 70s when Deming advocated for this such extensive version tracking would likely be much harder.
The software management issue I was talking about was a different animal. Take the Ariane V; the idea was the could iterate faster if they leveraged software reuse. Since it was previously verified software, they didn't think it needed much testing. But the new rocket design exceeded the assumptions in the previous iteration and the vehicle was lost. There's a lot of instances where software allows you to iterate faster, but you have to very careful about the baked in assumptions.
But that brings ups up a specific example of where Tesla has struggled. They used to keep HUGE backstocks of excess inventory (I believe they have since tamed this) which is a classic manufacturing mistake. So if you have any production mistakes or improvements in the part, you have to scrap or just cycle through the entire inventory until the improvements show up in the car.
This is why Just-In-Time is such an important concept in manufacturing. You can iterate constantly. Meanwhile, Tesla would crank out huge batches of identical fenders and doors and mirrors to maximise efficiency (economies of scale) but then struggle with part fits and panel gaps.
Tesla has since dealt with a lot of this when they went through their production hell period. But a lot of what they had to do was re-learn things like Lean Manufacturing and JIT.
Yes Tesla had to re-learn a lot about manufacturing. They certainty didn't start out being good. But by now they are doing pretty well.
Just 5 years ago, they had 10k cars per quarter.
You need to look at rate of change of improvement. Or, we can check back on Tesla 40 years from now.
Are you comparing the rate of change of improvement of Tesla to Toyota during their first 5 years?
For comparison, if Toyota was as efficient as Tesla they would've posted 2022 Q3 profits of ~$24 billion, not $3.1B. Tesla achieved $3.9B from only 344k cars. Operational efficiency is their leverage.
Cars sold (2022 Q3)
Toyota: 2.6M (+5% y/y) Tesla: 344K (+42% y/y)
Operating profit excl. FX impact Toyota: $3.1B (-38% y/y) Tesla: $3.9B (+97% y/y)
They have a different business model than OEMs, this is not apples-to-apples. OEMs act as wholesalers.
Besides the fact that you're talking about the efficiency of a company that literally produces 10x the number of cars. Let's see what happens to Tesla if they ever reach that scale.
>Toyota: 2.6M (+5% y/y) Tesla: 344K (+42% y/y)
Base rate fallacy. Why not use absolute numbers of cars, rather than percentage increase? It would tell a different story, wouldn't it?
10x higher production should in itself bring a lot in terms of economies of scale. Generally, the more you make of something, the cheaper one unit is to make.
So Tesla's position is certainly more impressive in this comparison.
I'm not even saying they don't have great manufacturing practices objectively - if you compare them to a lot of other industries they absolutely do. But auto manufacturing has always been the cutting edge of manufacturing tech and they simply are not at the top of that highly competitive field yet.
I'm curious, because sometimes exceptional supplier expectations come from a customer not really knowing what they need, so they ask for the world.
In doing something as simple as making their own seats, they were able to also improve the safety. They've put sensors directly in the seats to reprogram airbag deploy sequences, in software, on the fly. They know if there is a child sitting on the very edge of the seat vs an adult laying down across the seat and have the airbags fire differently based on this knowledge. Recently[2], they stepped this up a notch and are in the process of using millimeter wave radars to confirm this. They're industry leading in this regard and previously just used pressure sensors plus the cabin camera above the rearview mirror.
In short, Tesla is not perfect, but they are trying to rethink the problems from first principals. In doing this, they are coming up with very novel approaches that are actually big improvements over the entire industry. In other areas (fit and finish) they're towards the middle of the pack when compared against peer auto manufacturers.
[1] https://www.teslarati.com/tesla-patent-refrigerant-seat-temp...
[2] https://www.teslarati.com/tesla-driver-monitoring-system-fcc...
One of the ultimate "business metric" for auto manufacturers is the gross margin aka "how much profit does the company make as a percentage of the cost of the vehicle"
Before Tesla, Toyota[1] generally held the crown as the highest with just under 18% over the past five years. Ford[2] averages around 14%. Tesla[3], is industry leading with an average of 20.61%. GM[4] has always sort of been a laggard with the average at 11.26%. The German manufacturers, known for their efficiency, also do very well with VW[5] at 18.75% (this includes Porsche, which significantly tips the scales upwards for Volkswagen) and Mercedes[6] at 19.66%.
So Tesla makes approximately 6% more profit per vehicle sold at the low end (GM) and almost exactly 1% more than Mercedes on the higher end. Mercedes has ~173,000 employees and (per your own comment) Tesla has ~110,000 employees. So doing "less with more" is a pretty nice way to say they're more efficient, is it not?
[1] https://ycharts.com/companies/TM/gross_profit_margin
[2] https://ycharts.com/companies/F/gross_profit_margin
[3] https://ycharts.com/companies/TSLA/gross_profit_margin
[4] https://ycharts.com/companies/GM/gross_profit_margin
[5] https://ycharts.com/companies/VWAPY/gross_profit_margin
[6] https://www.cnbc.com/2022/10/02/tesla-tsla-q3-2022-vehicle-d...