Tesla Insider Speaks Out On Problems Of Soft Tooling


“Soft tooling” complicated the launch of the Tesla Model X, so the electric automaker has foregone that step with the Model 3.

According to a Tesla inside source who wishes to remain anonymous, “soft tooling” resulted in the debut of an initially problem-plagued Tesla Model X. To avoid a similar issue with the launch of the Model 3, Tesla is skipping the “soft tool” step.


Inside Tesla’s Fremont factory

What’s “soft tooling” you ask? As Reuters explains:

“Typically, automakers test their design with limited production using lower grade equipment that can be modified slightly to address problems. When most of the kinks are worked out, they order the final equipment.”

So, tools, stamps, etc. that are sort of meant to be disposable and/or redesigned before arriving at “hard” production tooling.

Those “soft tools” are linked to several Model X problems, most notably the alignment of the Falcon Wing doors. In a rush to get the X out, initial production vehicles seemingly rolled off the line with some parts made on these “soft tools,” according to the source. This led to various problems. Quoting Reuters:

“Working on a tight deadline, Tesla had no time to incorporate lessons learned from soft tooling before having to order the permanent production tooling, making the former’s value negligible, the source said.”

“Soft tooling did very little for the program and arguably hurt things,” said the person.”

As such, Tesla is foregoing this step with the Model 3 and instead is using “hard tooling” right now on the various release candidates we see from time to time on public roads.

We’ll find out soon enough if this decision pays off for Tesla.

Source: Reuters

Category: Tesla

45 responses to "Tesla Insider Speaks Out On Problems Of Soft Tooling"
  1. R.S says:

    “Working on a tight deadline, Tesla had no time to incorporate lessons learned from soft tooling before having to order the permanent production tooling, making the former’s value negligible, the source said.”

    How about taking a bit more time to learn something from soft tooling? I agree that you don’t need it, if you won’t use it, but why not take a bit more time to get it 100% right? Scrapping hard tooling will be a lot more costly.

    1. Pushmi-Pullyu says:

      “…why not take a bit more time to get it 100% right? Scrapping hard tooling will be a lot more costly.”

      Yup. This is basically a case of “We’re gonna cross our fingers and hope for the best.” Definitely a risky bet. That may pay off… but considering the thousands of parts made for the Model 3, it’s mathematically almost certain that at least a few are going to have to be redesigned or replaced with something entirely different after production starts.

      I understand Tesla’s desire to get the M3 into production as soon as possible. Billions of dollars have been and/or will be spent on tooling up for production and on building the Gigafactory. Having all that equipment sit around unused, but having to pay monthly fees on utilities and servicing Tesla’s debt, is an expense Tesla really, really needs to avoid. But if the Model 3 gets saddled with the albatrosses of a reputation for unreliability and problems which are not fixed in a timely manner, as the Model X has been, then it will be a disaster for Tesla, and even worse for the company’s future than delaying production by a few months.

      If there are serious problems with early production Model 3’s, then Tesla really needs to bite the bullet and wait until those problems are fixed before it sells even one car to anyone who isn’t a Tesla employee.

      But those caveats don’t stop me from saying: Go Tesla!

      1. cab says:

        I agree that this is a very risky move for Tesla and quality issues foisted on a legion of buyers used to decent quality today won’t be well received…at all.

        Of course “risky move for Tesla” basically described everything they’ve ever done! Let’s just hope the 4th step after their “3 step plan” isn’t “fix critical mistakes made in step 3”.

    2. (⌐■_■) Trollnonymous says:

      “why not take a bit more time to get it 100% right? ”

      There’s no such thing as 100%.
      No such thing as 95% either.
      If they waited to attain those, the products would never ship.

      1. Nix says:


      2. ModernMarvelFan says:

        That is what SW engineers say all the time.

        *sigh*. I guess Tesla is really a SW company which is a good thing because that is where the profit and growth is. But then again, that type of attitude will result in issues with hardware…

        The initial quality of both Model S and Model X are good proof of that.

        The fact that Tesla is using early “pilot” material (from soft tooling) in the customer sales unit is a practice that would have been frowned upon by most engineering companies out there.

        I hope Tesla it get right the first time.

        1. Kieran Mullen says:

          Hmm when a hard drive says MTBF that’s really software eh? When a network has 6 nine’s of availability, that’s just hardware ? https://en.m.wikipedia.org/wiki/High_availability

    3. JIMJFOX says:

      Are ‘clays’ still used for final body/panel shaping?
      Can ‘soft tooling’ be considered a parallel case?

      1. JIMJFOX says:

        Oops! Failed to read the comments further down…
        I really can’t see that a clay will give you more visual information than a properly-rendered 3D model viewed on a VERY large screen [4K / 75″]


  2. mx says:

    With current CAD/CAM systems, sheesh, the systems are at least 20 years old, there should be no need for soft tooling.

    1. Pushmi-Pullyu says:

      CAD systems are a very useful tool. But just like all tools, it can be used well or used poorly.

      Look what happened to Boeing with its attempt to go directly from CAD design to finished parts for the Dreamliner, without producing prototype parts to make sure everything fit right before proceeding to actual production and assembly.

      It was a disaster. Despite everything fitting perfectly in the computer, in real life many things didn’t fit so well. Boeing made the mistake of letting each vendor develop parts on their own, without close supervision by Boeing to make sure every vendor was following the exact same procedures.

      Reportedly, Tesla is working very closely with vendors, in many cases directly overseeing production in the vendor’s factories. So hopefully the problems there will be far fewer. But given the very large number of parts in the Model 3, it probably isn’t realistic to believe that there will be no problems at all.

      1. philip d says:

        Your point is valid but a passenger airliner is an exponentially more difficult machine to build this way than a car.

        While cad can design the components down to a small fraction of an inch, parts of differing sizes and materials will expand and contract differently.

        In a passenger airliner there is a much more dynamic environment with which these parts have to work together. From ground level to 30,000 feet temperatures swing drastically in a short amount of time.

        Also these components undergo a large range of movements and stress across the airframe from small vibrating ocillations to larger scale bending and torquing.

        Cars do experience varying vibrations and forces from varying road surfaces and speed but it is a lot more predictable. At least it ihopefully is for Tesla’s sake.

        1. zzzzzzzzzz says:

          Airliner is actually easier by some aspects. It is slow production, few per month, you can go back and fix what was wrong.
          Try going back and manually fixing 500,000 mass production cars on road – a company without reserves may go under.

      2. bro1999 says:

        Same deal with clay modeling. With all the computer modeling programs available, you’d think clay modeling would be extinct like the dinosaurs. But even Tesla still uses clay models!

        Some stuff you just can’t quite replace with computer simulations.

        1. Pushmi-Pullyu says:

          I agree, clay modeling seems positively medieval.

          Why don’t they use large-scale 3D printers to print out sections and glue them together for a mockup? Seems like a much better way to control the exact dimensions, and they can easily make another if they don’t like the first attempt.

          But come to think of it, the clay can be reworked much faster than waiting on the 3D printer to make 2nd copies of everything. So maybe not such a good idea.

          1. wavelet says:

            I read something about this a while back.
            AFAIR, clay modeling isn’t used for engineering purposes, but to try out the overall look of a car visually; as we all know, aesthetics are a vital factor in car sales.

            An onscreen simulation isn’t the same thing as a physical model, and 3D printing, besides being quite slow, requires modeling the parts accurately in CAD first — the accuracy isn’t needed to test out various looks for the car.

      3. JIMJFOX says:

        Misfits are NOT the fault of the CAD program; anything can be built in 3D cyberspace to 0.00001 mm BUT the problem arises in the manufacturing processes.
        But if you use monkeys on your CAD system you will indeed get peanuts.
        [UK nuclear sub CAD models were flawed because of inadequate skills & training in CAD]

    2. sven ¯\_(ツ)_/¯ says:

      Apparently, the Reuters article is nonsense and completely wrong. A Forbes article claims other automakers don’t currently use soft tooling and haven’t used soft tooling so for decades. Can anyone confirm this?


  3. Pushmi-Pullyu says:

    While the problems related in the article may have contributed to Tesla’s problems with the MX’s falcon-wing doors, it seems to me this is an attempt to shift blame. Jay Cole said that Tesla didn’t get around to even soliciting submissions for FW door actuator designs from vendors until only a few months before MX production was supposed to start. This was far too short a time for proper testing, spotting problems, diagnosing them, producing a corrected design, and re-testing. To some extent at least, Tesla’s problems with those doors were an “unforced error” which they created for themselves.

  4. Bacardi says:

    The Irony is, because the M3 has the “it factor” you’ll still look cool broken down on the side of the road…see a Ferrari broken down on the side of the road? Far more people would pull over and offer help compared to say a Prius…

    1. Taser54 says:

      People are more likely to help the Prius, IMHO. The Ferrari owner can sweat it out on the side of the road, that is unless the driver is Christie Brinkley.

      1. Bacardi says:

        Corrected your statement for you:

        “People are more likely to help the Ferrari. The Prius owner can sweat it out on the side of the road, that is unless the driver is Christie Brinkley.”

  5. John in AA says:

    After Elon’s adolescent humor at the D reveal, it’s almost irresistible to riff on “soft tool”.

    1. georgeS says:

      It’s the ambien and the wine:)

  6. Taser54 says:

    Looks like this PR release is not well received, and why should it be? It’s revisionist history. Even diehard Tesla fans know what happened with the Model X delays.

    Now we have Tesla going straight to production tooling without completing testing of the Model 3. As others have written here, that is a big gamble.

    I’m really thinking that the employees’ cars are essentially part of a captured beta fleet. That is, Tesla will buy them back. That way Tesla can claim that production has started, yet prevent these beta cars from reaching the public.

    Tesla will be able to incorporate lessons learned and the public will get a better car.

    It might get real interesting if an employee’s Model 3 purchase agreement leaks.

    I wish Tesla success, but I still don’t believe this release.

    1. georgeS says:

      “Now we have Tesla going straight to production tooling without completing testing of the Model 3. As others have written here, that is a big gamble. ”

      Check out svens post. Apparantly shorting to hard tools isn’t something new.

      ….but getting your employees to buy the car and test it for you certainly is.

      Pretty ingenious IMO. When Tesla ships it’s first “production car” to an employee the press will all go wild that he made his bogey.

      1. zzzzzzzzzz says:

        Automaker employees testing cars isn’t new either really. They do it all the time. Except that they get paid for it, not vice versa 😉

  7. georgeS says:

    “Yup. This is basically a case of “We’re gonna cross our fingers and hope for the best.” Definitely a risky bet.”


    Meh. The risk is mitigated because No falcon wing doors and a steel body. Besides sven already posted an article that said this practice is already being followed.

    1. Pushmi-Pullyu says:

      I’m quite surprised any regular reader here would take Sven’s word for it. When is the last time he posted anything about Tesla or its cars which was actually true? It’s not impossible I suppose, but I’d put the odds of truthfulness there as extremely low.

      1. sven ¯\_(ツ)_/¯ says:

        Bite me, troll.

  8. georgeS says:

    “Look what happened to Boeing with its attempt to go directly from CAD design to finished parts for the Dreamliner, without producing prototype parts”


    Boeing started that practice way before the 787. They started it on the 777–That was 25 years ago.

    1. Pushmi-Pullyu says:

      *** @Staff: I’ve had this and similar attempts at a post disappear three times. Perhaps the ‘bot doesn’t like my links, so I’m replacing them with indirect goo.gl links. *** — PMPU

      Okay, I used the wrong term. I should have said “mockup” rather than “prototype”.

      I can’t find the “deep dive” article I once read about Dreamliner problems, but one of the points made in that was that Boeing used a CAD “virtual model” to ensure that all parts from diverse suppliers would fit together, and that none would try to intrude into the same space. But when actual assembly was first tried, Boeing found that indeed there were parts that needed to fit into the same spaces. As I said, at least according to what I remember from an article I read some time ago, the different vendors did not use the virtual design in a consistent fashion, and Boeing failed to oversee the suppliers’ design processes adequately.

      The article linked below isn’t exactly the most objective or neutral one I’ve ever read, and thus is perhaps a questionable source, but it does claim that multiple mockups were used during the design of the Boeing 777:


      Those interested in reading details of how Boeing’s reliance on outsourcing parts and assemblies for the Dreamliner, and its lack of oversight, contributed greatly to a three-year (!) delay in getting the Boeing 787 into production, can find details here:


    2. JIMJFOX says:

      Perhaps 787 problems were more to do with extensive use of composites than the CAD?

  9. Ron M says:

    I’m surprised that some people think that the Model 3 is going to have all kinds of problems getting production off to a fast start. If Musk admits that he made some mistakes with the roll out of the Model X then I’m sure he learns from his mistakes and doesn’t repeat them. Model 3, Solar Roofs, Autonomous Driving and Energy Storage are gonna propel Tesla forward. Musk’s vision and it’s work force that’s as committed to his vision as he is will make Tesla a force to be reckoned with like Amazon has.

  10. Nix says:


    It seems to be working….

    1. georgeS says:


      Have you been partying with Elon? You better lay off that ambien:)

      1. Nix says:

        *laugh* It started with just a regular caps-lock error, and I decided to just go with it…

    2. georgeS says:

      PS Nix,
      I have to hand it to you. Seems like you were one of the first to propose that Tesla went right to final production line tooling to shorten up the development process.

      1. Nix says:

        Yup, Tesla actually started building out the Model 3 production assembly line last summer, and began unit testing individual sections of their production assembly line last October. They ordered enough parts from vendors to build 300 cars late last spring and have been using those parts for testing assembly. I swear I can type that in my sleep, since I’ve been posting it since last Nov.

        So while people are fretting endlessly like Tesla is somehow just now trying to fit all the parts together a month away from production, nothing could be further from the truth. They’ve already got 8 months worth of work under their belt of testing how their final production assembly line will fit parts together. (Oct-May)

        And while there certainly are parts that Tesla is reworking with suppliers, not all changes in parts trigger a change in the size or shape of the part. Often parts can be changed and still use the exact same mounting and fit in the exact same space and get installed by the exact same robot with no additional programming.

        Tesla just isn’t where many people think they are at in getting ready to build production cars off their production assembly lines.

        1. JIMJFOX says:

          Can ANYBODY explain to me the differences between “exact same”,,, “exact”,,, & “same”???

          Is it like ‘half dead’ or ‘partly pregnant’?

  11. Spider-Dan says:

    So Tesla was forced to use soft-tooled parts because of “tight deadlines”… on a vehicle that arrived two years behind schedule?

    This is not a very convincing excuse.

    1. Ron M says:

      I think you mean two years ahead of schedule. Right

      1. bro1999 says:

        Model X was 2 years ahead of schedule? In what universe?

    2. Josh Bryant says:

      Have you ever worked on a project that is 2 years behind schedule? It sucks.

      Every deadline is yesterday. Every PM is asking what can be cut out. It is a recipe for poor engineering quality.

      Model 3 is in much better shape than X, but I am sure it will still have some hiccups. The massive fleet of employee owners should help to curb how much of those issues make it out into the volume production though.

      It is a brilliant move if the employees buy into it.

  12. Ron M says:

    Sorry I was talking about the Model 3