As much as I appreciate
@luvb2b's work - and use it as one of multiple baselines when making my own estimates (tweaking various parameters, so I don't have to model every little detail on my own) - it's important to point out that for Q3 luvb2b predicted:
Auto revenue: $5,2B (actual: $6,0B)
GAAP auto margins: 19,9% (actual: 25,8%)
GAAP net income: -$238M (actual: $261M)
EPS: -$1,33 (actual: $1,91)
I remember calling that forecast (and others in that thread) way too pessimistic in several regards, and was considered to be unreasonably bullish at that time.
Lots of people around here were deleveraging based on that forecast, and partly missed out on a big jump. Of course, the actual results were more bullish than I was!
That said: a 2019 GAAP profit is just
not going to happen. Or anything close to it.
Firstly, a correction, I misremembered the GAAP profit
@luvb2b is estimating for Q4: it's +$295m.
Secondly, Q3 was arguably an outlier, it was profitable
despite a 10% increase in production and an only 1.8% increase in deliveries and price reductions, because:
- a significant improvement in efficiencies: my guess is that the primary driving factor was that Tesla re-negotiated the battery purchase contract and its pricing formula in particular with Panasonic, to better tie cell prices to the (dropping) world market prices of key raw materials,
- this can be suspected from the fact how 'storage' margins improved from near zero to 20% in a single quarter,
- excellent opex control - historical weak area of Tesla under Deepak as a CFO - Zachary seems much better at this,
- there was a ~$100m GAAP income from foreign exchange shifts - it's real income but almost impossible to model due to how it shifts randomly between quarters.
None of which could be reliably modeled. If just two of those factors were included (storage margin improvement and FX effects), his model would have shown a small profit for Q3.
More importantly, these improvements from Q3 are now included in the Q4 model.
It also makes sense to take a broader look at the GAAP income track record of luvb2b's model:
As you can see there's no inherent bullish or bearish bias in the model, it's been missing in both directions. Also note that all 6 predictions of him got the direction of the quarter-to-quarter shift in GAAP income right, and only a single quarter (Q3'2019) got the sign wrong - and this was over an extremely challenging 1.5 years of Tesla's history, with a lot of strategic and operational shifts going on within the company.
More importantly, in quarters where opex and cost structure was mostly unchanged, luvb2b's model had very good accuracy. He was pretty much the only modeler who expected Q3'2018 to be profitable, and got within 16% of the final profit, which was amazing.
I expect Q4 to be a similar quarter to Q3 in terms of cost structure and opex control: Model 3 production was up 10% from Q3, total deliveries were up 15%. In Q4'2018 Model 3 production was up 15%, deliveries were up 9%.
In Q4'2018
@luvb2b's model overestimated GAAP profits by $83m, so I don't think there's much basis to suggest that his $295m estimate for Q4'2019 is overly bearish. If you click on the link above luvb2b also lists a few wildcards that could improve Q4 results, but again those are hard to model.
In any case I'd warn against too bullish expectations for Q4, in addition to the potential wildcards, there's various potential headwinds as well:
- the FX gods could move against Tesla this time around,
- Tesla could have more opex from China and from the accelerated Supercharger expansion,
- they could also be recognizing some costs now instead of Q1 to help Q1's results,
- or they could decide to not sell any ZEV credits or recognize FSD revenue in Q4 but save them all for Q1, etc.
- they might write down any remaining 75D inventory with steep discounts.
The threshold number is $266m Q4 profits IMO, because that would make even marginal $1m profits in Q1 trigger S&P 500 inclusion eligibility.
Anyway, not advice.