I'm reminded a little of how the Blackberry board supposedly watched the iPhone announcement and insisted it was fake, as there was "no way" someone could build and profitably sell a phone with that feature-set.

Tesla might not do what they're promising. Hell, they probably won't do what they're promising. But pronouncing something impossible on the basis of industry experience can be a dangerous thing. Some things are considered unthinkable and impossible only until someone finds a way to do them, after which they are retroactively declared obvious and inevitable.

Certainly, Tesla seem to have the right ideas: Electric cars are inherently simpler, mechanically speaking, than gasoline cars, and they're leveraging this further by reducing the options and variability on the Model 3 to the absolute minimum. They also hired Peter Hochholdinger from Audi as VP of production, and he is pursuing a pretty extreme form of Industry 4.0[1] — with everything in their production process being instrumented, automated and designed for easy replacement.[2]

[1] https://en.wikipedia.org/wiki/Industry_4.0 [2] http://www.mljournal-digital.com/meleadershipjournal/october...

The body on an electric car is basically no different to an ICE car. The floor is flatter and thicker to accommodate the battery pack and no need for a tunnel to accommodate the drive train, but other than that they're about the same.

Tesla will almost certainly be using a combination of materials on the model 3; mild steels (cheap and cheerful, easy to press), aluminium (lightweight) and boron steels (strong). To join these materials you need a variety of techniques including resistance spot welding, aluminium RSW (altogether more difficult than steel because of the low melting point and high conductivity, and the fact that weld spatter sticks to the tip of the gun), riveting (for steel-Al joints), structural adhesives, flow drill screws, laser welds etc. All of these processes take serious time to calibrate and get right, and some are not all that robust (by high volume standards).

There are many quality loops with a vehicle launch. The tooling needs to be proven out. Tools get shimmed relentlessly to make up for panel variations - panel quality can vary from batch to batch within a press run. This is especially true for mixed material cars.

Believe me, these are not easy problems to solve and nothing I've seen and heard out of Tesla makes me think they've addressed any of these issues.

Is it possible that Musk has solved the problem, much like he or his employees have solved countless other tech problems, but Musk also knows what makes a sexy headline. Spot welding is hard to sell as flashy and new. If you can't do it on mars, it might not make good headlines.

If Musk solved the problem, they'd be producing the Model S faster already. ;)

Spot welding isn't hard, and wasn't claimed to be hard. Mixed Materials welding and joining is Hard (steel + aluminium/mild steel + high strength steel/ etc).

It isn't even that it's hard. I mean, it's been done before, many times. But there are processes here with a stack of key control parameters that can and will drift.

Panels will not be consistent - they never are (and you'd be surprised at the variation). Rivets will tumble in feed tubes or get trapped in escapements. Vision systems will get confused by ambient light or reflections at certain times of day. Weld spatter will accumulate on tips, and frequent tip dressing will change weld parameters (timing, pressure, current etc). Adhesives will be effected by temperature, humidity. Batches of sealer might not be consistent. Weld timers will fail. Power supplies will fail. Flow drill screws will deform, go off centre and be a bitch to get right. Laser welder lens will accumulate weld crud. People will mis-load parts. PIA bits will come in below the required quality. Tools will get shimmed wrong. Robots will be badly programmed.

All of this stuff will take time to get right. It isn't impossible, but it is very difficult to get a line performing reliably. Tesla will be integrating a lot of new technologies, and at volumes that they are not used to, in an area with no serious history of car building, in a high wage state. I doubt they can get the people to manage the line and maintain this equipment.

You make it sound like we already know about most of the problems and there is a ton of room for improvement in the current systems.

Isn't this exactly what Musk and his employees are good at fixing? I have seen no reason to bet against Musk, Your complaint is the same as the others saying he cannot do something. You seem to have expertise in how a thing is done now, not in how well it could be done.

He might be late, but he won't fail. As it is this is the scheduled shortened by 6 months.

If electric cars are so simple, why is the Tesla Model S so unreliable?

https://www.truedelta.com/Tesla-Model-S/reliability-1095

Conversely, the Toyota Prius is one of the most reliable cars. Yet it is mechanically complex with both an internal combustion engine and an electric motor, plus a CVT. So clearly engineering skill and manufacturing quality are bigger factors than the drive train.

The Model S drivetrain is reasonably reliable. Most of the reliability problems (and I believe they have an average score now, not great but not terrible) are other parts of the car. Tesla likes their unnecessary "cool" tech, which also tends to break. The Model S has the extending door handles, and the X has ridiculous doors. Not that it's limited to these aspects of the vehicle, of course.

Note that the Prius doesn't have a CVT the way you'd normally think of one. It has a fixed planetary gear set which is able to somewhat emulate a CVT by using two electric motor-generators in conjunction with the engine. I mention this only because "CVT" usually makes people think of painful friction-based transmissions, whereas the Prius's transmission is much simpler than even a normal transmission. Your point that engineering skill and manufacturing quality are the main factors is spot-on.

How unreliable are ICEs these days - maybe its just me, but I've never had an engine problem in nearly 20 years of owning cars. Electrical problems, suspension problems, drainage problems - had all of them but never had a problem with the actual engine.

Car guy here. Internal Combustion Engines have some pretty complex systems that pure electric systems don't need. Water cooling and engine oil spring to mind. Timing belts need to be changed, transition systems have problems with some regularity. (Granted, most performance electric cars have a transmission too, but you can almost get away without one, so I'm going to assume more simplicity of design)

Not arguing one way or another just providing data. My only electric car is an old converted Porsche. (Transmission is still there but you can basically slap it in third and leave it there) I'll never have to change the oil on that one again, and the electric motor is simple as can be. Battery management is kind of a pain, but that's because I don't have the kind of advanced battery management a mass market car would have.

I don't have enough data yet, but my experience suggests that electric cars will have a much lower maintenance load simply because of the reduced complexity. Of course all the automation and bells and whistles will still break, and the computer control systems will get wonky.

The last internal engine problem I had was with a 200cid (3.2 liter) inline six from Ford - the pistons had to be de-carboned after a cross-country drive. This was in 1985, so I blame the 55 mph national speed limit for not letting the engine run harder. Ever since - I just do 5000 mile oil changes and 3 year coolant changes, and all is well. They're really quite reliable these days, even as the complexity has gone up with turbos, direct injection, variable valve timing, and so on.

Maybe it depends on the age of your car and how you drive it? I've had 4 Honda Accords from an 88 up to an 07 that all started exhibiting the same engine problems around the 200k mile mark. Blown head gaskets, burning oil, timing belt issues, etc.

I assume you were replacing the timining belts on schedule. Given the near legendary reliability of the Honda Accord it makes me wonder if there's something about the way you drive.

Even so, 200,000 miles is not too shabby. I don't think we have useful data on electric vehicles with that kind of mileage.

I have a honda with 226,000 miles. Starts and runs like new.

I have replaced timing belt and water pump and motor mounts in addition to regular fluid changes. The timing belt and water pump work was proactive maintenance (the timing belt drives the water pump so they suggest replacing both at once since it's not really any extra work).

The motor mounts simply wore out. Not expensive to replace. Rubber mounts will wear out on a Tesla too, eventually.

What is a drainage problem?

I had a Passat a few years back and it had a design flaw where the tray that holds the battery has drain holes that can get blocked and if they do it overflows into the interior of the car.

I used to park my car under trees that dropped leaves which blocked the battery compartment drainage holes and as I am in Scotland it can get quite wet. Ended up with car filled with 1" of water.... Interior of car had to be stripped, dried out and put back in by the dealer.

I'm always tempted to test drive stylish German cars but I swear 90% of the maintenance horror stories I've heard in the last 15 years have involved a German car.

VW in particular has a pretty poor reputation for long term reliability (also see Consumer Reports et al). The stories I hear about American cars, by contrast, tend to be poor reliability soon after purchase (followed by solid long term reliability)

Not sure I would describe a Passat as "stylish" :-)

It's when you realize your OEM sunroof leaks just like an '70s aftermarket sunroof but there is a drain that keeps you from noticing. Or there was a drain, now it's full of dirt and debris and doesn't drain very well.

Window and trunk leaks

Turbo-diesel fuel injection pumps are ... not the most reliable thing to have in a car.

Because (personal opinion) the Tesla Model S isn't a particularly great car. Their success is largely in marketing, network, and visual appeal, not in the product itself.

If you pick a better (albeit less appealing) electric car, the reliability rate skyrockets. For instance, using your own source -- https://www.truedelta.com/Nissan-LEAF-vs-Tesla-Model-S-relia...

^ This. The Model S has a lot more in common with a mid-high range Mercedes or a low end Ferrari. Anybody who owns an exotic car like that will tell you, you do not buy those for the reliability.

It's also worth noting the Model S is, and is marketed to the audience for, a performance car. When you put down that kind of money, you're going to drive it like one. If you drove a Honda like a Ferrari you'd likely break a whole lot of it's "reliable" parts too.

A Mercedes limousine is hardly exotic, and neither is it unreliable.

My own observations of Honda vehicles suggest differently, but I have very few observations of Tesla vehicles, so I don't really have enough of a point of comparison to really disagree with that assertion.

I'm not drinking anybody's kool-aid. If you think a 691 bhp Tesla is going to wear it's parts the same way as a 278 bhp Accord, I have no words for you. That's not even taking into account the much higher acceleration forces, G-forces, etc. that the parts in a Tesla are going to be dealing with, plus the fact that it has a completely different center of gravity and that affects how every component of the suspension wears.

I'm not saying Tesla build quality is higher or lower, I'm saying it's comparing apples and oranges. Once Tesla has a production everyman's car available, then you can actually say something. I know five people who own Model S's and they do not drive them like Accords, and if their stories are anything to go by, most people who own them drive them like performance cars. That means hard corners, rapid acceleration and deceleration, and high speeds (and given the crap quality of our roads around here, a lot of turbulence too).

It's also worth noting that Honda's have decades upon decades of previous models to build off of. Tesla's been around since what, 2003? And they didn't have a car to show until 2008.

> If you think a 691 bhp Tesla is going to wear it's parts the same way as a 278 bhp Accord...

Literally no one said that.

And that wasn't the criticism of your comment. The criticism was that you asserted that Hondas would be as reliable as Ferraris if they were driven hard. Lots of people do drive their Hondas hard. They replace the exhaust and lower the car and maybe throw on a wing and pretend they're on a track all the time. They drive these things hard as hell and they hold up very well.

> I'm not saying Tesla build quality is higher or lower, I'm saying it's comparing apples and oranges.

It's not apples and oranges. It's cars and cars. Yes, people drive their Teslas harder than their Nissan Leafs. What about M5s? How do Teslas compare? They should be far more reliable due to the vastly simpler drivetrain. Are they?

> It's also worth noting that Honda's have decades upon decades of previous models to build off of. Tesla's been around since what, 2003? And they didn't have a car to show until 2008.

So your argument about Tesla reliability is essentially that they are unreliable, because they're new?

What I found the most interesting in that comparison is that the Tesla, although having a lot more problems in other areas, had 0% issues in brakes and suspension.

Are they using some different much more advanced technology for those parts, or it's just some problem with the repair data?

I can only speak for the brakes. Teslas use regenerative braking on the motor before falling back to traditional frictional brake pads as I understand it. Essentially the motor is put in reverse and that drains off speed. That means less wear on the brakes and so they last much, much longer than an ICE car.

My ICE car has regenerative braking. So this is not something exclusive to the EVs.

Regenerative braking only works if there is spare battery capacity to absorb the energy. In general that requires a large battery to be useful. A normal car battery is nowhere near large enough and is generally topped off by the alternator.

My ICE car has regenerative braking as well, but I certainly didn't buy it for that, nor have I been able to discern the impact of it. I have driven EVs before and the regenerative braking is unmistakable.

Really? Don't you need a big electric motor for that... and some substantial lithium ion batteries? Is your ICE car a hybrid?

No, it is not a hybrid. But it uses brake energy regeneration to charge car's battery when you are not accelerating. The effect is not as pronounced as on EV cars, but enough to not use brakes often in normal traffic. In the context of the GP statement, it does reduce some brake pads wear.

ICE hybrids use regenerative breaking.

Actually the Prius is less mechanically complex than a normal ICE-Car. The planetary gear is really simple and you don't need parts which are prone to wear like a high torque clutch or a complex gearbox.

It is very reliable not despite but because of it's unique powertrain concept.

My friend put 100,000 miles on his Tesla in a year and was never stranded. He drove 30,000 miles on a known faulty motor.

It's not more unreliable than an ICE vehicle. It's significantly more simple in design and for repair. These claims are FUD.

This seems unbelievable. That is 500 charge discharge cycles at 200 miles apiece. Since there are only 365 days in a year This means that your friend was driving more than 270 miles a day meaning he was at 55 mph that is 5 hours a day. This is on the order of a professional trucker once you count weekends.

This also discounts the difficultly of finding well space out chargers. It also discounts the schedule of your friend. This also makes me wonder what time you have left to socialize with your 'friend'.

If you are going to make bold claims some explanation of the detail would go a long way towards making them seem plausible.

Well when someone gets a million dollar settlement from their employer and gets to keep their Tesla paid for outright and has nothing better to do for a year they just drive around the country. So that's what he did.

> insisted it was fake, as there was "no way" someone could build and profitably sell a phone with that feature-set

A little context here might be good. The supposed disbelief of Blackberry was their carrier deal, as they themselves were not _allowed_ to sell phones with that feature set.

It wasn't unthinkable to build it engineering-wise. (It was expensive yet with underwhelming features, a 2G phone with no apps at a time when even feature phones had 3G and apps for almost five years.) It was unthinkable business-wise. Those rules are easier to play hard and fast with.

That there is unfounded criticism doesn't mean all criticisms is unfounded.

The carrier deal was also a surprise to them, but if you believe this anonymous former RIM engineer they also didn't think it was possible to do what the iPhone did and still have acceptable battery life:

"RIM was even in denial the day after the iPhone was announced with all hands meets claiming all manner of weird things about iPhone: it couldn't do what they were demonstrating without an insanely power hungry processor, it must have terrible battery life, etc. Imagine their surprise when they disassembled an iPhone for the first time and found that the phone was battery with a tiny logic board strapped to it. It was ridiculous, it was brilliant."

https://news.ycombinator.com/item?id=2044389

The iPhone is now a 10 year old device still selling like crazy. Thanks for the link.

I'm not sure how realities of production get beaten by "ideas". Yes, Tesla has excellent design and approach. That won't help them magically build in volume. Or quality - even very expensive Model S cars aren't really a paragon of build quality in their class.

Steve Jobs didn't promise production of 300mil iPhones in a year when they announced it.

I don't know if you've heard of this little thing called The Law of Accelerating Returns.

It's the law.

Ok, I get it, do not mock The Law of Accelerating Returns.

Please stop posting unsubstantively like this on HN.

Even just using Tesla's numbers, it's unlikely they can ramp up production.

They currently produce ~25K Model X and S per quarter, and they're having supply issues with battery packs. (see the other, recent thread on them)

How in the world are they going to be producing 20K Model 3s by December, without cannibalizing their more profitable existing sales, and judging by their own previous growth rate and demand problems?

> they're having supply issues with battery packs.

They are building a huge factory for making battery packs in Nevada. That's already operational and will be more so by December. I imagine that's part of the solution.

> I suspect the first cars that they sell will roll off the VP tools and not off the main lines anyway.

There a story a couple of months ago about the fact that they've saved a lot of time by going straight to the main production lines rather than building a testing line first?

Is this a good plan? Time will tell but I'm sure glad I'm not involved. :S

Sources:

http://www.newsweek.com/tesla-skips-traditional-manufacturin...

http://autoweek.com/article/green-cars/tesla-wont-test-model...

Edit: From the second link:

> Let's recap: Tesla's flawed Model X launch was partly due to not taking the time to adjust for problems identified in the soft-tooling phase of pre-production. The lesson Tesla learned? Don't bother with soft tooling for the Model 3.

Wow. Is it just me or does that sound like "The unit tests kept failing and made us miss a deadline. No more unit tests!"

No, this is not a good plan at all, it's incredibly risky.

To get the car all road legal it needs to be crash tested - that is usually what the pre-volume builds are used for. If they fail (and they sometimes do, because you can't do a finite element analysis on an assembly as complex as a car in the variety of scenarios that you would crash test it) then you need to reengineer the car. This usually means panel gauge increases, sometimes it means additional reinforcement parts (boron steel inserts and brackets a lot of the time). If that happens the production could be delayed by months as they'd need to find a way to feed those parts to the line and modify any tooling that might be effected.

So these all sound like very true and good things. No reason to doubt there is a reason current big auto does it the way they do.

On the other hand, Musk hasn't been one to just say stuff and totally fail. By and large he is getting what he says he is getting, and on a close to reasonable time frame. If we stuck to what big auto said, there would be no autopilot, and EVs would still be a toy with a tiny range.

I have at this point little reason to doubt that they will pull this off.

Clearly making cars is a solved problem, millions are made every year. It SEEMS like Musk hires people that know their stuff.

So I am going long on Tesla, and short on existing big auto. Time will tell what happens ;)

> and on a close to reasonable time frame

woah, let's not get ahead of ourselves here. when elon says years, it's martian years as the meme goes, and not without reason. he's getting a better at estimation but is still an extreme optimist.

But Musk's numbers are much lower than everyone else's and he usually some percentage late, Not decades. Sometimes he is even on time.

He does things that many claim are literally impossible, who cares if they come 10% later and that is 3 months. If Telsa hits the production numbers by June they are still doing something that GM and Ford once said was decades in the future.

This is where I am too, I could see him aiming for 500k cars in 2018 and only shipping 425k due to some shortage. I could not see it taking until 2022 for him to get to the 500k mark.

Elon tweeted a couple of days ago that Model 3 had all needed regulatory approvals, so I guess it passed the crash test. I can believe that you know a lot about building cars, but you don't seem to have any actual information about what's going on inside Tesla's Fremont factory.

Aren't we already past that point, though? This seems a bit like worrying that the horses might bolt today because the barn door was left open yesterday.

Yes, and that's the scary thing.

People said the exact same thing when Musk announced SpaceX. Then they said no big deal after the Falcon 1 worked, and laughed when Musk said he was going to land the Falcon 9 on a barge in the ocean, then he did. Next they said there was simply no way SpaceX would be able to refly a used first stage, which they've now successfully done twice.

History of his accomplishments show us "industry knowledge" doesn't really matter as he's trying to do things explicitly different than the rest of the industry.

Musk seems to set high goals with the full intention of meeting them, but shrugging it off if he's wrong in the end so long as progress was made. In the end the goal doesn't matter, just the progress, which was certainly affected positively by the goal.

Very well said.

Some goals are more strategic than realistic, and I think anyone can learn what strategic goals look like from Musks' ventures.

In my view strategic goals don't have the same form as SMART goals:

Traditional Smart - Specific, Measurable, Attainable, Relevant, Time-bound

Strategic Smart - Specific, Measurable, Advancing, Relevant, Terminable

The two differences, Advancing and Terminable, reflect the idea that strategic goals are geared towards innovation, not iteration. Advancing means that you don't actually know whether it is Attainable. Terminable means that the goal may be cancelled without major disruption to the business.

Entirely agreed. He loves to set lofty goals, and he does not always achieve them, but sometimes he does! Either way, he's unafraid to challenge virtually any status quo or existing understanding of things. This is why industry experience while relevant, doesn't always matter. Also, he moves his top talent between his companies at will, giving him an edge over other companies.

I've got a friend who has worked for Aerospace Corporation for close to 11 years as a structural engineer. He people when they have to stop removing things to prevent rockets from exploding (aka a rocket scientist). He told me ~5 years ago there was literally no way SpaceX could land orbital class rockets. Yeah, he doesn't say that anymore.

What's more, a lot of the materials science and metalworking "lessons learned" (particularly when it comes to working with aluminum) at SpaceX are available to Tesla.

Really not the case. Space X have built, um, not many rockets. Tesla are planning to churn out hundreds of thousands of cars.

The key isn't being able to do it (as someone else said joining metals is a solved problem), it's being able to do it with high quality and in a reliable way. Do you think Space X joining techniques are optimised for making 100k joints in a factory environment?

You are entirely incorrect. SpaceX perfected friction stir welding for large sheets of aluminum for their fuel tanks. That tech was transferred that technology entirely to Tesla, who happen to make cars from large pieces of aluminum (Model X and S).

https://electrek.co/2015/05/24/spacex-transferred-novel-weld...

There is no way that they are using friction stir welding in a high volume line. It's a very slow process, visually it's horrible (big scar along the joint) and, though I've no experience with it I would have thought that pushing a spinning piece of metal at high temperatures through two thin abutting pressings would be highly unreliable.

"though I've no experience with it" - aka you're not qualified to answer whatsoever. At least you're honest in your trolling :)

SpaceX did not invent friction stir welding, but they did develop a mastery of the technique. Luckily, unlike you, they hire people who have experience in it. The strength of friction stir welds is one of the reasons they use it in the Model X and Model S vehicles, both which blew it out of the park regarding NTHSA safety standards.

Musk himself describes it here (for SpaceX):

https://www.wired.com/2012/10/ff-elon-musk-qa/

They even mention a few of the aerospace bits they have in the Model S:

https://www.tesla.com/blog/tesla-model-s-achieves-best-safet...

Disclosure: I've never worked in or around the automotive industry.

The amount of focus and effort Tesla has put toward the tech in the Model 3 pales in comparison to the amount they have put toward factory automation. Musk has said as much in multiple interviews. He's talked about how it's 10x harder to 'build the machines that build the machines'.

See also their acquisition of Grohmann Engineering.

Having said all that, we will have to wait and see. One thing I do know is that Musk's ventures have a long history of surprising the status quo.

There is no way anyone could ever profitably produce a rapidly reusable rocket. No one can make an online bank. No one can do solar profitably. No one can replace the yellow pages (Zip2 got passed on when an investor threw the yellow pages on the table). No one can make an electric vehicle that doesn't suck...

You might not be wrong, but the shorts get squeezed Everytime someone says this. I wouldn't bet against Tesla and the market is more than tolerant of these delays

I would have more confidence in this announcement if it was not made a day before Tesla announced that they are experiencing falling demand for their existing line and barely hit the bottom of their production forecast.

Their initial "hand over" is to employees, not real customers, so it is really just internal testing and does not indicate anything about when actual deliveries will begin.

Hitting delivery deadlines is actually worse than missing them if you are hitting them by pushing an unfinished product which is losing tons of money per unit due to inefficient production processes and creating huge future liabilities for warranty repairs, recalls, and negligence lawsuits if someone gets injured by a defect.

There is a reason why Goldman is revising their price target down on this news and their target is half the current stock price.

Falling demand, really?

"After delivering 25,051 vehicles during the first quarter, it pushes deliveries to “approximately 47,100” according to the company. [For the year.] It represents a 53% increase over deliveries during the same period last year and both Model S and Model X deliveries were higher than during Q2 2016"

Maybe a plateau could be argued looking at the graph here.

https://electrek.co/2017/07/03/tesla-delivers-vehicles/

"After delivering 25,051 vehicles during the first quarter" ... they delivered 22,000 vehicles in the second quarter.

That sounds like a 12% decline to me. It sounds like a 12% decline to Goldman Sachs. I am not sure what you call it.

Car sales can be seasonal, or like Tesla said in their press release, they where supply constrained by their new 100 kWh battery. Just comparing numbers from the last two quarters is not really very informative. The previous quarter was an increase, the one before that a decrease, the one before that a substantial increase. More important is a trend. And I see a plateau when looking at the mast four quarters.

Those employees are still paying their own personal money for the car. They're as much "real customers" as you or me.

I suspect they're not paying what customers will pay and they're likely to be leases since they're effectively pre-production models and you'd have to be a big knucklehead to bet on the resale value of a pre-production Model 3.

Tesla offers zero employee discounts on their cars. Even Musk pays full price, on principle.

https://www.quora.com/Do-Tesla-employees-get-a-discount-on-T...

I imagine the value of a pre-production Model 3 will be quite good in ten years. Look at what the iPhone (original release) is worth. Or perhaps a Ford Model T?

This is not gods earth. I would be interested in how they solved these challenges but you come off being someone who just points out problems which is not as helpful as someone who thinks of solutions, also not as abrasive either.

> There is no way on god's earth that Tesla can ramp up production like that

I'm not clear what like that you are referring to, or what would be a more reasonable plan.

What do you think is a realistic goal for its production (numbers wise) of the Model 3 by the end of the year?

Commenting to remind myself to check back later and see if you were right!

By clicking on thr time string in the post, you will get to a page which will let you favorite the actual comment, without your own comment.

There was an article recently where some higher ups talked about the engineering. They described focusing work on the factory itself instead of the product with the idea being that every time you improve or fix the factory, you gain many upsides in the final output.

Can't find the article but I was an interesting read

(Warning: biased source)

Electrek has an article quoting it. Elon was describing the idea of designing the "Machine that builds the Machines"[0].

“We realized that the true problem, the true difficulty, and where the greatest potential is – is building the machine that makes the machine. In other words, it’s building the factory. I’m really thinking of the factory like a product.”

[0]:https://electrek.co/2016/06/01/elon-musk-machines-making-mac...