The 5 Lakes bots
Controversy has haunted Tesla at every step of its extraordinary journey, making it hands-down the most argued-over automaker in the world. These controversies have touched on nearly every aspect of the California-based car company’s business, but they also share one core commonality: at their core, every Tesla controversy has been about disclosure.
Countless issues that might have been easily forgivable if disclosed early and fully have escalated into major controversies because of how the company communicates (or doesn’t) about the challenges they face. Never has this been more true than it is in the case of the most recent controversy about Model 3 production, which was kicked off last week by a Daily Kanban report.
Every serious auto analyst has been saying for years that ramping up to mass-scale production of the Model 3 will be the biggest challenge Tesla has ever faced, so the fact that the company missed its internal Model 3 production goal for Q3 by some 85% isn’t intrinsically controversial. But Tesla’s statement accompanying this massive miss stated that “there are no fundamental issues with the Model 3 production or supply chain,” even after confirming its missed goal at the start of production. After Daily Kanban‘s report, Tesla PR reps “repeatedly” and “vehemently” denied that Model 3 parts were being made by hand according to Jalopnik’s Ryan Felton, and then completely changed their tune after the Wall Street Journal’s Tim Higgins reported that aspects of Model 3 manufacturing were indeed taking place by hand. Had Tesla simply disclosed what was causing such a massive delay to the Model 3 production ramp instead of misleading, obfuscating and attacking reporters the controversy would doubtless have died down… at least until the next production number disclosure. The mismatch between Tesla’s ambitious guidance and underwhelming performance, as well as its aggressive attempts to discredit any reporter attempting to add to the public’s understanding of the situation, turned a wholly unsurprising production “hiccup” into a full-blown controversy.
Though this approach fits Tesla’s established pattern of PR practice and though it plays well with the firm’s fanatical “cult” followers, it was hardly the only logical approach to questions raised by its Q3 Model 3 production miss. After all, a look back through CEO Elon Musk’s statements about the Model 3 production ramp provides some evidence to suggest Tesla already disclosed –in its uniquely ambiguous manner– its current predicament well in advance.
Musk first started discussing Model 3 production in-depth during Tesla’s Q1 2016 earnings call, when he set a July 1 2017 start of production deadline but warned that “volume production” would take place “a number of months later.”
“Now, will we actually be able to achieve volume production on July 1 next year? Of course, not. The reason is that even if 99% of the internally produced items and supplier items are available on July 1, we still cannot produce the car because you cannot produce a car that is missing 1% of its component. Nonetheless, we need to both internally and with suppliers take that date seriously, and there needs to be some penalties for anyone internally or externally who does not meet that timeframe. This has to be the case, because there’s just no way that you have several thousand components, all of whom make it on a particular date.
So the reality is that the volume production will then be some number of months later as we solve the supply chain and internal production issues.”
Musk tempered this warning by reminding analysts that the Model 3 had been “designed for production” and would thus be unlikely to follow the ramp patters established by Model S and X. As a result, Musk said his “expectation” was that Tesla would produce 100,000-200,000 Model 3s in the second half of 2017.
In the subsequent Tesla earnings call, Musk revealed that the Model 3 production ramp would not in fact be a single ramp at all, but rather multiple ramps using multiple production systems.
“The Model 3 – the internal name for designing the machine makes the machine is the – we call it the alien dreadnought. At the point at which the factory looks like an alien dreadnought, then you know you’ve won. It’s like, what the hell is that? So we’ve got alien dreadnought version 0.5 will be Model 3. It will take us another year get to version 1 and probably a major version every two years thereafter. By version 3, it won’t look like anything else. It might look like a giant chip pick-and-place machine or a super high-speed bottling or canning plant, and you really can’t have people in the production line itself. Otherwise you’ll automatically drop to people speed. There’s still a lot of people at the factory, but what they’re doing is maintaining the machines, upgrading them, dealing with anomalies. But in the production process itself there essentially would be no people.”
As usual, the main reaction to Musk’s vision focused on the promised end state rather than the process by which he planned to reach it. But as exciting and provocative as Musk’s resurrection of the fabled “lights out” factory concept was, the apparent plan to repeatedly upgrade the Model 3 production system after the start of production was arguably even more provocative. If, as Musk had repeatedly claimed, the Model 3 was “designed for production” the plan to dramatically change its production system over the course of a product cycle was an odd one. Yet nobody asked how Tesla could have designed the Model 3 for a production system that was several iterations from its ultimate state, or what kind of impact each new iteration would have on production rates. Not only does each plant retooling typically disrupt production for several months, but each major change to production systems also require new processes, worker retraining and a whole new ramp. Having struggled with “production hell” in each of its previous production ramps, Tesla’s apparent decision to endure several of such disruptive transitions over the course of the Model 3 product cycle was a curious one.
A quarter later Musk gave more color on the Model 3 production system, saying
“I mentioned this before, but as we go to high volumes, what really matters is the factory, the machine that designs the machine – the machine that creates the machine is – becomes actually of greater significance, much greater significance than the machine itself. That’s where we have most of our engineering team working on. So sort of an internal codename for the factory machine that builds machine is the alien dreadnought so a point in which our factory looks like an alien dreadnought and we know it’s probably right. So we think with Model 3, it will be alien dreadnought version 0.5 approximately, and then it will take us about another year or so, I don’t know, summer 2018 to actually get to alien dreadnought version 1.”
The last sentence of this quote was widely overlooked, but turns out to have been extremely important. Not only was Musk reiterating his statement that the Model 3’s initial production system would be “version .5”, he had given a timeline for version 1 of the production system. Already it was possible to read through the lines: Model 3 would be produced using an incomplete production system (version .5) for roughly the first year of Model 3 production, and that the first complete version of the system would come online a year later in Summer 2018. In the world of software, particularly open source software, versions below 1.0 typically denote an incomplete version. By repeatedly using the version .5 designation, Musk was effectively disclosing that the Model 3 wouldn’t launch with a “real” production system. Moreover, by disclosing that the Model 3 production system would be upgraded to version 1.0 about a year after the start of production, Musk was signalling that the first year of production would be akin to a pre-production validation run rather than a mid-cycle update to a fully-developed production system.
The Q1 2017 call saw Musk evolving this vision:
Yeah, so with Model 3, I think we’ll be roughly comparable with the best high-volume vehicle production lines in the world. Better in some respects, a little worse in others. But roughly comparable, and then with some further iteration, I think it will probably be a little bit better than the next-best automotive production line. Then where things will really be a step change, I think, beyond any other auto manufacturer, will be the Model Y factory. And this is all a function of designing the product to be easy to manufacture and easy to automate, as well as designing the factory itself. So Model-wise, I think, we’re really the common step change, but Model 3 is going to be at or probably slightly better than I think the next best automotive production in the world. I just think that’s pretty good outcome. And then Model Y will be – there will be nothing close to it, I think.
Here Musk refers to the Model 3 line as being comparable to existing high-volume production lines, suggesting that he is talking about the 1.0 production system referenced earlier. Further versions of the production system would be reserved for the subsequent Model Y crossover, which at that point was planned for an entirely different platform than Model 3. Musk backtracked on this in the subsequent quarterly call, saying his executive team had “reeled him back from the cliffs of insanity” and convinced him to share the 3’s platform for the Y. But in an exchange later in that call, he also made what seems to have been one of the most significant statements about Model 3 yet:
Antonio M. Sacconaghi – Sanford C. Bernstein & Co. LLC
Yes. Thank you. I have one question and one follow-up as well, please. In terms of the Model 3, at the delivery event, 20 of the units were for engineering validation. And the first several thousand it appears are going to be going to employees prior to going to the general public. So I guess the question is, what are you hoping to learn or what might you learn from these engineering validation units that have come out from your employees? And then, realistically, what’s
Elon Reeve Musk – Tesla, Inc.
They’re not engineering validation. They’re fully-certified, fully-DOT-approved, EPA-approved production cars. These are not prototypes in any way. They’re not validation anything. They’re full-production cars. The reason they are initially going to employees, and some cases, investors, or anyone has been a long-time investor is that for the first several thousand vehicles, there are problems that crop up that are rare.
On a percentage basis, they might be like 0.1% likely to occur. But then there are a whole bunch of these things that only show up 1 in 1,000 cases. And it’s good to iron out these things with a than to with customers. It also reward for those who work on the underlying development and creation of the vehicle. Yes. Yes.
Musk’s statement that the initial run of Model 3s is not “validation anything” and that “they’re full-production cars” is an incredibly definitive statement. If his previous statements about the Model 3 launching with a “version .5” production system implied that the first year of production was a pre-production run, he seemed to be refuting that analysis. But as the exchange continued it appeared as if he was playing with the commonly-understood definitions of the word “pre-production.”
Jonathan McNeill – Tesla, Inc.
I mean, it’s important to note too that all those people paid full price for their car.
Elon Reeve Musk – Tesla, Inc.
Yes. Full price. There was no discount internally at all.
Jonathan McNeill – Tesla, Inc.
Exactly.
Antonio M. Sacconaghi – Sanford C. Bernstein & Co. LLC
Right. No. I mean, the root of the question is if you realistically uncover something and even if it’s 1 in 1,000, what flexibility do you think you have to actually be able to rectify that concern in a way that won’t impact your ramp? And if this is being done arguably later in the process than a traditional OEM, that’s not trying to ramp necessarily as aggressively as you need to, I’m, again, just trying to understand realistically what can be done and what kinds of things, like perhaps you can give an example of what you might uncover and what the rectification might be?
Elon Reeve Musk – Tesla, Inc.
Yes. These are not (61:13), obviously. These are – what we’re talking about here are inconsistencies in the production process or in the quality control from a supplier. So they’re relatively easy to correct. They tend to be quite a large number of them. But, again, only rarely occurring. Usually, it involves like a tolerance stack up, or some combination of factors that we didn’t anticipate, but they’re almost always very easy to correct, but there’s just a bunch of them. It’s a lot of work.
Musk argued that the first year of Model 3 production was not “pre-production” or “validation” for several reasons: the cars were DOT certified for public sale, and employees taking delivery were paying full price for them. These characteristics do distinguish this initial run from most pre-production vehicles, but in general the purpose they seem to serve appears to be precisely the same as any other pre-production or validation run. Releasing a limited run of vehicles to employees of the company so that they can find quality issues to be fixed prior to public sales is precisely the role pre-production or validation cars serve. The more accurate way to phrase the initial run of Model 3 production is pre-production validation that happens to have been monetized through employee purchases.
Outside of Tesla itself, nobody knows exactly what the state of Model 3 manufacturing is. Between the WSJ’s report that initial production was done by hand and the subsequent Wards Auto report that quoted Tesla sources as saying a second body shop would be required to hit the 5,000 units per week rate Musk has forecast for the end of 2017, it seems clear that there are indeed fairly “fundamental” issues with Model 3 manufacturing. Further quotes from the Wards story suggest that Tesla skipped a number of traditional validation techniques, further suggesting that the initial rollout to employees is indeed a validation test period by another name (and with employee-derived cashflow). With version 1.0 of the Model 3 production system still about a year away, it seems likely that this is the pilot system that Daily Kanban reported is in development at Thai Summit and Five Lakes Automation.
