Bloomberg runs a model that tracks Model 3 production, using a combination of registered VINs and VINs of delivered vehicles. According to the model, Tesla (TSLA) is currently producing 2,395 Model 3s per week. What’s more, the trend data indicates a surge to well above 3,000 per week before the month is out.
For now, Bloomberg’s model provides evidence that the Model 3 ramp at the end of Q1 is sustainable, not a temporary burst as some critics had speculated. That forecasted 3,000+ per week rate would be a sudden increase from the Q1 exit rate of 2,020 Model 3s per week. If Tesla can raise the weekly production rate by 1,000 per month in April, then in May, then again in June, that gets the company to its goal of 5,000 Model 3s per week by the end of Q2.
Six months of delays left some fans disheartened and some critics emboldened. What I continue to think is that Tesla’s ambitious future is not cancelled, simply delayed. And now it looks like the delay might soon be over.
Two Model 3s charging at the University of Dayton in Ohio. Photo by DatTr0waway.
While some critics argue that a ramp to 5,000 Model 3s per week is not realistic, others are agnostic on that point. Their contention is that the Model 3 cannot be profitably sold at its advertised price points. Some critics are even eager for Tesla to ramp production, so it can start losing money faster. Well, we should find out in about ten months if that’s going to happen.
If the ramp proceeds as described above, I think the decisive moment could come when Q4 2018 earnings are reported. (If Q4 2017 is a guide, that should be in February 2019.) My reasoning: if the 5,000/week rate is achieved by the end of Q2, it will take another quarter for production to be optimized. So, until the end of Q3. Only in Q4 will we finally get the first full quarter of representative data. Then we can test the claim that the Model 3 can’t be sold profitably.
What might surprise many people is how critical software is to the Model 3’s gross margin. Tesla’s work on Enhanced Autopilot, led by Andrej Karpathy, has outsized importance here. So does work on features beyond the Enhanced Autopilot package, which will be bundled into the Full Self-Driving Capability package. To avoid confusion in the interim before actual full self-driving, that package might have to be renamed something like… Ultra-Enhanced Autopilot.
Software margins are deliciously thick because the cost of copying and transmitting computer code is typically on the order of pennies per unit sold. Other costs include tech support, which is usually also small on a per unit basis. For this reason, 95% is a reasonable estimate for the gross margin on a software product like Enhanced Autopilot. Enhanced Autopilot costs $5,000, so that’s $4,750 in gross profit per purchase.
If 67% of Model 3 owners buy Enhanced Autopilot and the average selling price of the car is $42,000, then the car’s hardware could be sold at break-even and the gross margin would be 7.6%. Again, that’s assuming the hardware is sold at cost — an incredibly bearish assumption.
The uptake rate depends a lot on how much Enhanced Autopilot improves. The 67% figure comes from a survey conducted long before the 2018.10.4 update that added remarkably improved lane keeping and lane changing abilities to Enhanced Autopilot. Here’s an example of its new capabilities. With the update, Enhanced Autopilot no longer needs painted lane lines in order to detect where a lane exists. This allows it to navigate through improvised lanes like this one created with pylons in a construction zone:
If enough future improvement bumped the uptake rate up to 80%, gross margin (with hardware sold at break-even) would be 9%. At a 90% uptake rate, gross margin would be 10.2%.
Then there’s Full Self-Driving Capability (which may later be rebranded as something like Ultra-Enhanced Autopilot). Currently, it’s sold for $3,000. At an eventual 50% uptake rate, that would take gross margin up to 13.6%. What’s more, CEO Elon Musk has indicated the price of this option will increase once some actual functionality is offered.
Under these optimistic assumptions for the Model 3’s software, the gross margin on the Model 3 hardware would only need to be 13.7% for Tesla to hit its 25% gross margin target. Is it realistic to expect a 13.7% gross margin on the hardware?
The base price of the Model 3 is $35,000. The average price for a new car in the U.S. is $35,285. The average gross margin for the U.S. auto industry is 12.2%. At a 12.2% gross margin on the hardware, the Model 3’s overall gross margin would be 23.7%. Would we call that close enough?
It’s interesting to note that the average gross margin of the U.S. auto parts industry is 17.6%, higher than for companies that do final assembly. Tesla is so vertically integrated that it has been described as a hybrid between an auto manufacturer and an auto parts supplier. If I take the average of the gross margins for the two industries — 14.9% — and apply it to the hardware of the averagely priced base Model 3, the result is an overall gross margin of 26%.
In other words, Tesla only has to hit a U.S. industry average gross margin on the Model 3’s hardware to achieve its 25% gross margin target. To make up the rest of the distance, it needs to perform exceptionally well in selling high-priced software options — something that Tesla, unique among automakers, excels at. Critics sometimes scoff when I say Tesla is a software company. Well, it sells software. That software accounts for a significant share of its gross margin. What else do you want?
I’ll switch gears now from gross margin to EBIT margin. This offers me another angle to discuss Model 3 profitability.
UBS has a rather bearish or conservative model of the costs that go into making and selling a Model 3. I think this is a good starting point to talk about the EBIT margin of the Model 3. UBS (UBS) estimates that Tesla will have a 2% EBIT margin at a selling price of $42,000 (see page 45). Since Tesla previously guided as the Model 3’s likely average selling price, I’ll use UBS’s $42,000 version of the car as a surrogate for the average of all versions of the car sold. So, even on UBS’s conservative assumptions, we still get a positive EBIT margin. Let me explain what those assumptions are.
The first conservative assumption is that Tesla’s battery packs cost $165 per kWh. Tesla has indicated that its battery packs cost no more than $133 per kWh. The base Model 3 sports a battery pack “just over 50 kWh.” Let’s call it 52.5 kWh. At $133 per kWh, that comes to $6,983, which is $2,092 less than the $9,075 that UBS estimated. If that $2,092 were to drop straight into the EBIT for the $42,000 version, the EBIT margin would jump to 6.6%.
The Model 3’s optional long range battery pack is 75 kWh and costs $9,000 extra. Let’s speculate with survey data as a guide and say that half of customers will choose each battery pack. That puts the average battery pack size at 63.75 kWh and average cost at $8,479. However, since the cost of the extra 22.5 kWh for the long range battery pack should only be $2,993, the contribution of this option should be $6,007. Since we’re speculating the take rate is 50%, it should contribute $3,003 on average.
So, by subtracting $596 from the average battery pack cost and adding $3,003 to the average contribution from options, the contribution margin should increase by $3,599. Again, if that drops into EBIT for the $42,000 version, the EBIT margin jumps to 10.2%. That’s above UBS’s EBIT margin estimate of 9% for the base version of the BMW (OTCPK:BMWYY) 330i.
The second conservative assumption is that at a $42,000 selling price, options will only add $3,500 to the contribution margin. As discussed above, Enhanced Autopilot should contribute at least $3,182 on average (at a 95% gross margin, 67% uptake rate, and $5,000 price). If we go with UBS’s estimate of a 50% contribution margin on hardware options, then the average contribution from options should be $4,182. That’s $3,182 from software options and $1,000 from hardware options.
Dropping this additional $682 into EBIT brings EBIT margin up to 11.8%. This is not far off from the 14% EBIT margin that UBS estimates for the $45,000 version of the BMW 330i. So, after revising these two overly conservative assumptions, UBS’s model has the Model 3 making a roughly BMW-level EBIT margin.
Model 3. Photo by Leo Nguyen.
Sustaining the growth engine
My idiosyncratic investment thesis is that the auto sales business will all but disappear within about 10 years of the widespread commercialization of fully self-driving cars. Auto sales will be replaced with autonomous ride-hailing. In my view, Tesla is uniquely well-positioned to grab a share of that new market. That’s the primary reason I’m invested in Tesla.
For this reason, I feel pretty laid back about the auto sales business from a valuation perspective. The auto sales business is primarily a means to an end. It’s necessary to sustain the company long enough to get to autonomous ride-hailing. In the big picture, that’s what matters most. It’s not that I don’t think Tesla would be a good investment based on long-term growth of the auto business alone. It’s just not where I think the future is likely to go, or where the greatest opportunity for Tesla lies.
This perspective will no doubt sound strange to some people. However, it’s how I really think about investing in this company, and for what it’s worth, I think it’s rational. The empirical test of this investment thesis will probably take at least two years to come. Eventually, we’ll see how it shakes out.
In ten months, Tesla has the chance to prove that it can produce the Model 3 at scale and sell it profitably. If Tesla can indeed prove this, the long-raging debate on Tesla will have to be refocused on other points of contention. Tesla will probably remain controversial among investors for years to come. However, after Q4 2018’s earnings, this particular chapter of the debate may finally be closed.
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.