Battery Pack Costs Plus Overlooked or Underappreciated Items from Q4 Call

[MitchJi]

I apologize for any mistakees, typo's etc. If I had known how long this was going to take me to do write this...

Battery Pack and Gigafactory
Gigafactory - Production Cars and Storage Growth

Given the dependency on the Model 3 profitability on the Gigafactory, can you talk about timing for a full ramp on both anode and cathode assembly?
<snip>
This is a modular facility, right? So you've got machines up and running, so just the first tranche of equipment that you've installed. When is that going to be up and running at full capacity?

Yes, they're pretty large increments. But those will be up and running at the full capacity for the first instances within just a few months. We're already in the stages of – we'll be doing the second instance of anode and cathode electrode assembly. So, these are needed for the ramp of Model 3. So you can get a sense of the timing here.
<snip>
First instance of the electrode manufacturing line.
<snip>
Yes, and that first instance would be achieving full volume in Q3.

Would you need to build another Gigafactory prior to hitting 1 million units per year in 2020, or do you think that the current Gigafactory will have enough capacity?

...but if you say it's somewhere around the 60-kilowatt-hour to 70-kilowatt-hour level, then you need 70 gigawatt-hours to get to 1 million units. And we think that's the cell level. And then we think the current Gigafactory should actually be able to do in excess of 100 gigawatt-hours. So, that leaves – probably a big – Gigafactory 1 can manage – can support – it can support probably 1 million vehicles a year, plus maybe something like 30 gigawatt-hours or so of storage, depending upon how fast the storage market grows. But really, I think the storage market's probably going to grow maybe twice the rate of the automotive business. Something like that.
<snip>
Yes, I currently think that we should build 500k vehicles next year and 1 million vehicles by 2020. That's 500k vehicles in total, Model S, Model 3, and Model X...

Gigafactory - Car Pack Costs

Tesla believes the Gigafactory will drive down the cost of its battery packs 30% by at least 2017 and 50% by 2020.

In your question you had [something that] should be corrected, like the -- so the 30% savings is not just due to logistics. Logistics is a big factor. We are ---

It's not even the biggest though.

Logistics [indiscernible] the fact that it's just go to one station to the next instead of going from multiple entities to multiple entities. But really when you get to the kinds of scale that we're talking about, you really get to design custom equipment that's much better at processing each step. And you really get to design the machine that makes the machine, not just do so with off-the-shelf equipment. So it took -- everything about it is going to get a whole lot better. That's why we think the 30% number when the Giga Factory is at full production is a conservative number.

The following is clearly less important than using custom equipment and vertical integration for achieving their 30%-50% cost reduction estimates:

Further price reductions are achieved by manufacturing cells that have been optimized for electric vehicle design, both in size and function.

Summarizing they said that the use of custom large scale manufacturing equipment and secondarily colocation (aka vertical integration) are the main reasons for the GF cost reductions. This scale of this equipment is so large, that each instance of anode and cathode electrode assembly is big enough to produce a significant percentage of the worlds current battery production. Designing and building this custom large scale equipment only makes sense because of the massive scale of the GF, and also partially because they need to build more than one of these production lines. So they will setup some kind of production system to reduce the cost and the time required for of producing the succeeding cell production lines.

In other words the major product is not GF1, it’s GF Production Line 1 (GFPL1). This has huge implications for the flexibility of setting up smaller factories, or setting up factories in phases to obtain most of the cost benefits of a larger factory.

That means, now that it's completed and working satisfactorally, they will be able to produce succeeding production lines at a substantially lower cost and more quickly than they produced the first one.

These lines can be installed anywhere that Tesla needs a multiple of one or more of these lines. I believe that it will be possible for Panasonic to produce 2170 cells in Japan, at a similar cost to the cells produced at the GF by doing two things:
1. Using one or more units of the large scale manufacturing equipment they developed for use at the GF.
2. Vertically integrating that factory. Obviously converting their existing 18650 line to produce 2170's is relatively trivial. But getting the price down requires much more than that. It probably makes more sense to shift MS-MS 2170 cell production to the NV GF (to take advantage of colocation), and setup and use more efficient production lines in Asia for Asian TM and Asian TE.

I think that it’s pretty clear the idea to use customized equipment and to use vertical integration were Elon-Tesla’s ideas, and that Panasonic’s hesitancy to invest ended when they were convinced that a 30% cost reduction was a conservative number. So it's more complicated than just scaling production. *Elon-Tesla figured this out using first principles. If it seems obvious why didn't Panasonic figure this out without having to be convinced by Tesla?

Tesla is now claiming 35% battery cost reduction at ‘Gigafactory 1’ – hinting at breakthrough cost below $125/kWh

Tesla is now claiming 35% battery cost reduction at ‘Gigafactory 1’ – hinting at breakthrough cost below $125/kWh.

The company was aiming for at least a 30% reduction from its battery cost before the Gigafactory and it now claims a “35% cost reduction” in a new promotional video.

Tesla has never officially disclosed its battery cost beyond claiming to be “below $190/kWh” in early 2016. That was before any battery cell manufacturing was taking place at the Gigafactory.
<snip>
The “holy grail” of battery cost, meaning when most battery-powered vehicles will be cost competitive with gas-powered ones even before accounting for gas saving, is believed to be $100/kWh. In the past, Musk said that he would be “disappointed” if Tesla doesn’t hit the milestone before 2020, but that was before they accelerated the Gigafactory production plan by two years in order to meet the new Model 3 production plan.

The tripling of capacity (aka alien dreadnaught production), which were also due to **Elon-Tesla innovations will also have an important impact on cell and pack costs. Because they will get 3x the production for the same fixed costs. Tesla originally said that they conservatively expected cost reductions of 30% by 2017 (beating their projections on price), and 50% by 2020. I believe that the additional 20% by 2020 projection came largely from the fact that they were planning on the cost reductions from producing multiple units of the large scale cell manufacturing equipment. If that's correct (what else could it possibly be?) their costs will be even less when they hit 35 GWh of cell production in the current GF, because when they hit 35 GWh production they will have achieved all of the cost reductions originally planned for the 35 GWh goal, but have only incurred about a third of the building costs.

So I believe that when the GF production hits 35 GWh (2018 or 2019) Tesla's car pack costs will be less than half of $190 per kWh (~50%-55%) or under $95 per kWh. Probably under $85 per kWh. I'm sure some of you are thinking that 50% reduction in costs is crazy, but the second generation TE products have about a 50% reduction in price and they include the invertors!

I've already pointed out two things which are major advantages for Tesla which were due to Elon Musk. I don't understand the criticism he receives on this forum. IMO he should be regarded in a similar fashion as Lebron James is regarded by the Cleveland fans. IMO he is clearly Tesla's main moat.

You talked about Gigafactories, 3, 4 and possibly 5. It sounds like you're pretty covered with Gigafactory 1 in terms of the 1 million vehicles, but I wonder if you could just speak to the strategy and thinking and timing there. And then also Panasonic would be your partner on those as well. Thanks

I think we'll reserve some dry powder for announcements later this year. This is surely more than enough news for today. But I think those announcements will be really quite exciting later this year.

What I believe would be quite exciting is not a cap raise that involves either selling stock or a loan tied to a stock hedge. What's exciting about those options, other than the prospect of a SCTY type dip (two or three more GF's before the first one is complete). I think that they will do something like one of the suggestions here:
Suggestion for Tesla to accelerate their trajectory for reducing carbon emissions
Model 3 Production

Tesla First Quarter 2016 Update
• Volume Model 3 production and deliveries to start in late 2017

The date, I'm sure this will leak, it's hard to keep a secret really, the date we are setting with suppliers to get to volume production capability with the Model 3 is July 1 next year.

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 components.

Tesla Fourth Quarter & Full Year 2016 Update
• Model 3 on track for initial production in July, Volume production by September

Well, I feel pretty confident that we should get there by the end of this year, to 5,000 a week.

So, when we place parts orders with our suppliers, we've told them 1,000 a week in July, 2,000 a week in August, and 4,000 a week in September.
<snip>
And no matter what date we set the exam paper for, when the term paper was due, there's always like some number of people that are late. It's just the way it goes. People sometimes – well, and I'm guilty of this too. Like too optimistic about the timing or they get unlucky or something like that.

So, we have to set these really strict dates, then some number of people are late, but it only has to be 1%...

So, I think it's going to be a very compelling car, but it's just a simpler design and we also understand manufacturing a lot better than we did in the past, and we're also able to get usually the A team at the A supplier for Model 3. It's rare that we're not able to get that, whereas, particularly for Model S and to some degree for Model X, when we were trying to get suppliers for Model S, a lot of the top tier suppliers wouldn't even work with us. They thought we'd go bankrupt.
<snip>
Now, in fact, we are building something on the order of 50,000 Model S's per year. And so, having shown the results of the Model S and the Model X, the interest from suppliers went from basically getting like the worst team on second tier suppliers to getting the best team on first tier suppliers. Really a big difference.

The "A team at the A supplier" means that Tesla's business is important to the suppliers. I believe this is a major reason that both the Letters, and Elon sound more confident about starting production in July.
Model 3 Production Line and CapEx

Yes, I mean, there's obviously going to be a fair bit of incremental investment to go from 5,000 cars a week to 10,000 cars a week, but it's going to be a lot less than getting to 5,000 cars a week in the first place. We don't know exactly what that's going to be except I'm confident it'll be less. Because the first thing we'll try to increase output is going back to rocket equation is to increase exit velocity of the line. And we don't know exactly where the trouble points are going to be. We tried to model it out as carefully as possible, but there'll be things that aren't captured in the model.

But I think in a lot of cases, we'll simply be able to run the lines faster as opposed to duplicate the line. That's by far the best CapEx maneuver is just to make it go faster. But I would say it's going from 5,000 to 10,000 is probably – this is a total wild-ass guess, so right way to think about, but it's like somewhere between 50% to 70% of the cost of the 5,000 line. Something like that.

If you're lucky and smart, 50% is only half the game, which obviously is pretty awesome from a CapEx standpoint. I can't imagine it being more than about 70% as much as there is. So, JB, what do you think?

Yes, I think that's right. And it's maybe helpful to realize, but a lot of the infrastructure investments to get all the way to 10,000 are already completed, Gigafactory in particular.
<snip>
...great efficiencies in the way to layout of facility, for instance, as there were at Gigafactory. So, we don't anticipate needing to build much new square footage, for instance, to go all the way to 10,000, even though we would be expanding the internal production lines while we speed them up and add new instances of production, but the Tesla CapEx would not be a one-to-one scaling, not even close.

1. Tesla is modeling the line to get to 10k per month, but they are saying that the chances of being able actually dial up to that speed are initially are low. It will be really interesting to see the rate they get from doing that.
2. The concerns that Elon was crazy to suggest the possibility of 100k to 200k by the end of 2017 are exaggerated.
3. They are not building one line and then replacing the entire line.

Model 3 Cash Flow and CapEx

- Deutsche Bank Securities, Inc.
I guess, just still trying to calibrate to this cash flow and cash needs, maybe a different way to ask this, is it reasonable to expect that you would hit for the Motors company free cash flow breakeven at the 250,000-unit a year level for Model 3, assuming what we know today?

It depends on how quickly we want to ramp production to go from 5,000 a week to 10,000 a week for Model 3. There could be an argument that you don't want to go to cash flow breakeven or positive because you're losing a lot of sales and when you calculate the present value, huge cash flows, then it's like actually not smart to be in that case free cash flow positive. Or maybe you want to be a little negative at least and not give up a huge number of sales, because we would be talking – the numbers get just so crazy, it's 0.25 million units a year, roughly $1 billion a month revenue.

And so then you double that and it's $2 billion a month. So, maybe spending an incremental $0.5 billion on CapEx would be pretty smart move if it advances things by two months or three months.

If the capacity of the production system is X, until you are at least like half X, your gross margin is going to be weak, and it's going to be terrible when you're like an order of magnitude below, or if you're 10% of X, or less. It's going to be terrible. But then it'll get really good as you start to approach 100% capacity. Like, then it gets great. And then, as we get to the initial phase of capacity of 5,000 a week, I would expect to see gross margins comparable to that of the Model S and Model X.

Yes, I'm not sure if it's going to make sense for us to show the final version before start of production or after.

So, I think in terms of showing the final version, it's probably at least a few months away, maybe as far as July itself.

Final version. I think that has nothing to do with the part 3 (probably HUD or something else that is also on the MS-MX) reveal. They are obviously trying to do everything that they can to not cannibalize MS-MX sales. "Model 3 won't have self presenting door handles" (who cares!).

Elon Musk Down-Sells Model 3 On Twitter In Favor Of Model S

Starting at $35,000 before the tax credit, customers were eager to get into a Tesla for half the price of a Model S.

But with production scheduled to begin in just a few months, Musk took to Twitter on Friday and seemed to be encouraging customers to opt-in for the more expensive Model S — or lower their expectations.

But that's a little hard to do since Elon is committed to making compelling cars. I believe that the reasons that they are going to delay the full reveal and the opening of the Design Studio as long as they can because the Model 3 will be compelling, especially the prices. For example Elon tweeted that pack upgrades will cost less on the M3. I also believe that they are planning to make some upgrades to the MS-MX, and reduce the prices of some of the options at the same time. They are in an awkward position until they start making profits on the M3.

No Planned Obsolescence - Ludicrous Mode is a SW upgrade

...like we increased the design lifetime of the powertrain from roughly 0.25 million miles to aspirationally 1 million miles. So, that should really help with service.

Three Dog Day

Ludicrous Mode
While working on our goal of making the power train last a million miles, we came up with the idea for an advanced smart fuse for the battery. Instead of a standard fuse that just melts past a certain amperage, requiring a big gap between the normal operating current and max current, we developed a fuse with its own electronics and a tiny lithium-ion battery.

That was combined with upgrading the main pack contactor to use inconel (a high temperature space-grade superalloy) instead of steel, so that it remains springy under the heat of heavy current. The net result is that we can safely increase the max pack output from 1300 to 1500 Amps.

These is no planned obsolescence designed into the driveline, unless you consider a million mile design life planned obsolescence .

The million mile driveline, built into every Tesla, is the hardware required for Ludicrous Mode. In other words Elon just confirmed my belief that Ludicrous Mode is a software upgrade.

 

[dennis]

OMG @MitchJi my head hurts- but in a good way. I need a piece of chocolate.

 

[neroden]

FWIW I was thinking planned obsolesence in more pedestrian items like the touchscreen (will it really last more than 10 years?).

 

[EinSV]

@MitchJi -- great post!

An important assumption in the quote below seems to be that the additional cost reductions by 2020 that Tesla expected (reducing pack costs from <$125/kWh to <$95/kWh) are due to scaling of GF production units. I wonder whether there could be other factors that might contribute to Tesla's expected cost reductions, such as improvements in the technology/process for building GF production units (better/faster robots), improved pack design (both efficiency and manufacturability) as well as improved cell efficiency over time. It seems to me that these factors could contribute to lower pack production costs by 2020, and might be factored into Tesla's predicted >50% cost reduction estimate. Would be interested in your thoughts about that.

I believe that the additional 20% by 2020 projection came largely from the fact that they were planning on the cost reductions from producing multiple units of the large scale cell manufacturing equipment. If that's correct (what else could it possibly be?) their costs will be even less when they hit 35 GWh of cell production in the current GF, because when they hit 35 GWh production they will have achieved all of the cost reductions originally planned for the 35 GWh goal, but have only incurred about a third of the building costs.

 

[Snapdragon III]

I apologize for any mistakees, typo's etc. If I had known how long this was going to take me to do write this...

Battery Pack and Gigafactory
Gigafactory - Production Cars and Storage Growth




Gigafactory - Car Pack Costs




The following is clearly less important than using custom equipment and vertical integration for achieving their 30%-50% cost reduction estimates:

Summarizing they said that the use of custom large scale manufacturing equipment and secondarily colocation (aka vertical integration) are the main reasons for the GF cost reductions. This scale of this equipment is so large, that each instance of anode and cathode electrode assembly is big enough to produce a significant percentage of the worlds current battery production. Designing and building this custom large scale equipment only makes sense because of the massive scale of the GF, and also partially because they need to build more than one of these production lines. So they will setup some kind of production system to reduce the cost and the time required for of producing the succeeding cell production lines.

In other words the major product is not GF1, it’s GF Production Line 1 (GFPL1). This has huge implications for the flexibility of setting up smaller factories, or setting up factories in phases to obtain most of the cost benefits of a larger factory.

That means, now that it's completed and working satisfactorally, they will be able to produce succeeding production lines at a substantially lower cost and more quickly than they produced the first one.

These lines can be installed anywhere that Tesla needs a multiple of one or more of these lines. I believe that it will be possible for Panasonic to produce 2170 cells in Japan, at a similar cost to the cells produced at the GF by doing two things:
1. Using one or more units of the large scale manufacturing equipment they developed for use at the GF.
2. Vertically integrating that factory. Obviously converting their existing 18650 line to produce 2170's is relatively trivial. But getting the price down requires much more than that. It probably makes more sense to shift MS-MS 2170 cell production to the NV GF (to take advantage of colocation), and setup and use more efficient production lines in Asia for Asian TM and Asian TE.

I think that it’s pretty clear the idea to use customized equipment and to use vertical integration were Elon-Tesla’s ideas, and that Panasonic’s hesitancy to invest ended when they were convinced that a 30% cost reduction was a conservative number. So it's more complicated than just scaling production. *Elon-Tesla figured this out using first principles. If it seems obvious why didn't Panasonic figure this out without having to be convinced by Tesla?

Tesla is now claiming 35% battery cost reduction at ‘Gigafactory 1’ – hinting at breakthrough cost below $125/kWh

The tripling of capacity (aka alien dreadnaught production), which were also due to **Elon-Tesla innovations will also have an important impact on cell and pack costs. Because they will get 3x the production for the same fixed costs. Tesla originally said that they conservatively expected cost reductions of 30% by 2017 (beating their projections on price), and 50% by 2020. I believe that the additional 20% by 2020 projection came largely from the fact that they were planning on the cost reductions from producing multiple units of the large scale cell manufacturing equipment. If that's correct (what else could it possibly be?) their costs will be even less when they hit 35 GWh of cell production in the current GF, because when they hit 35 GWh production they will have achieved all of the cost reductions originally planned for the 35 GWh goal, but have only incurred about a third of the building costs.

So I believe that when the GF production hits 35 GWh (2018 or 2019) Tesla's car pack costs will be less than half of $190 per kWh (~50%-55%) or under $95 per kWh. Probably under $85 per kWh. I'm sure some of you are thinking that 50% reduction in costs is crazy, but the second generation TE products have about a 50% reduction in price and they include the invertors!

I've already pointed out two things which are major advantages for Tesla which were due to Elon Musk. I don't understand the criticism he receives on this forum. IMO he should be regarded in a similar fashion as Lebron James is regarded by the Cleveland fans. IMO he is clearly Tesla's main moat.




What I believe would be quite exciting is not a cap raise that involves either selling stock or a loan tied to a stock hedge. What's exciting about those options, other than the prospect of a SCTY type dip (two or three more GF's before the first one is complete). I think that they will do something like one of the suggestions here:
Suggestion for Tesla to accelerate their trajectory for reducing carbon emissions
Model 3 Production






The "A team at the A supplier" means that Tesla's business is important to the suppliers. I believe this is a major reason that both the Letters, and Elon sound more confident about starting production in July.
Model 3 Production Line and CapEx


1. Tesla is modeling the line to get to 10k per month, but they are saying that the chances of being able actually dial up to that speed are initially are low. It will be really interesting to see the rate they get from doing that.
2. The concerns that Elon was crazy to suggest the possibility of 100k to 200k by the end of 2017 are exaggerated.
3. They are not building one line and then replacing the entire line.

Model 3 Cash Flow and CapEx






Final version. I think that has nothing to do with the part 3 (probably HUD or something else that is also on the MS-MX) reveal. They are obviously trying to do everything that they can to not cannibalize MS-MX sales. "Model 3 won't have self presenting door handles" (who cares!).

Elon Musk Down-Sells Model 3 On Twitter In Favor Of Model S


But that's a little hard to do since Elon is committed to making compelling cars. I believe that the reasons that they are going to delay the full reveal and the opening of the Design Studio as long as they can because the Model 3 will be compelling, especially the prices. For example Elon tweeted that pack upgrades will cost less on the M3. I also believe that they are planning to make some upgrades to the MS-MX, and reduce the prices of some of the options at the same time. They are in an awkward position until they start making profits on the M3.

No Planned Obsolescence - Ludicrous Mode is a SW upgrade

Three Dog Day

These is no planned obsolescence designed into the driveline, unless you consider a million mile design life planned obsolescence .

The million mile driveline, built into every Tesla, is the hardware required for Ludicrous Mode. In other words Elon just confirmed my belief that Ludicrous Mode is a software upgrade.

This was great. Thanks for taking the time

 

[Fallenone]

Very nice post, thanks Mitch!

 

[schonelucht]

It's really exciting to see battery pack costs dropping. It really won't take long before the $100/kWh is the norm instead of the exception.

 

[MitchJi]

• @MitchJi -- great post!
  
  An important assumption in the quote below seems to be that the additional cost reductions by 2020 that Tesla expected (reducing pack costs from <$125/kWh to <$95/kWh) are due to scaling of GF production units. I wonder whether there could be other factors that might contribute to Tesla's expected cost reductions,
  1. such as improvements in the technology/process for building GF production units (better/faster robots
  2. improved pack design (both efficiency and manufacturability)
  3. as well as improved cell efficiency over time.
  
  It seems to me that these factors could contribute to lower pack production costs by 2020, and might be factored into Tesla's predicted >50% cost reduction estimate. Would be interested in your thoughts about that.

  1. I think that this is the same thing as the bold text.
  2. Between 2018-2020? I think that it's pretty mature technology.
• 3. You are correct. I missed that! When describing the 30-50% reductions Tesla specified that they didn't include any improvements due to cell chemistry, which means that we should deduct another 5-10% from their pack costs.
  
  I didn't mean to claim that the only factor is the ramping of production of the cell production equipment, but I believe that accounts for the bulk of the cost reductions.
  
  I once built some wood containers for holding five gallon glass bottles for hauling spring water in our pickup truck. The two main purposes were to proctect the bottles from breakage and to make it easier to move the bottles. The first one I built took me a lot longer. When it was completed I measured all of the individual pieces and set stops on a chop saw. It took me less time to build the next seven or eight, than it took to design and build the first one.

We know that the cell manufacturing equipment accounts for a huge percentage of the total cost of Gigafactory. We also know that for the first two sets of equipmentt that they build they not only need to design the equipment, but they also need to design or purchase and set up the equipment (programming robots for example).

They will still no doubt make incremental improvements in both the cell manufacturing equipment and the equipment that produces that equipment but I believe it's safe to assume that when they have completed the first two iterations, from that point on they will be able to produce the equipment much faster and cheaper than the first one or two sets. At that point they are ready to start producing the equipment for much less, but the unit costs will include the added costs of the first two sets. So building the capacity to produce ~100 Gwh will cost less per unit than building the the capacity to produce 35 gwh, not including the cost savings of the building and land per gwh, which I didn't include in my pack cost estimate.

Two key takes aways are that they will be able to ramp the increased capacity of the Gigafactory much more quickly and for less money than the the first two sets of units. The main thing to keep in mind (when buying options ) is that the Gigafactory production will increase faster than most observers believe possible.

Another key takeaway is that we don't need to obsess over Tesla's pack costs. I believe that my pack cost estimates, based on $190 -50% ($95) are extremely conservative. When I did my first pack cost estimates I believed that their maximum cost was $170, but there was no point in saying that because almost nobody would believe it. So I provided two figures $190 and $170. When Tesla announced that their costs were less than $190, that proved that at least I was correct with my figure of under $190. How much under $190 do you want to use for a starting figure? I think that they are concealing their costs. So concealing $10-20 makes more sense than concealing $5, why bother?

Both Tesla and Panasonic said that their 30-50% reduction estimates were conservative. And my estimates didn't include the reductions due to cell chemistry improvements. So you could conservatively start at $185 and deduct another 5% for cell chemistry improvements, plus you could assume 55% based on their recent statements that they were at 35% instead of 30%. The result is $185 x .4 = $74, and that's still a conservative figure.

So when Elon said that nobody is even close to their pack costs it's not because he's delusional. The reason that I believe that it's a waste of time to obsess about the exact costs because whatever it is, it's way less than their competition. So far ahead that the GM employee in charge of the Volt clearly didn't believe that $190 was even possible.

 

[EinSV]

Both Tesla and Panasonic said that their 30-50% reduction estimates were conservative. And my estimates didn't include the reductions due to cell chemistry improvements. So you could conservatively start at $185 and deduct another 5% for cell chemistry improvements, plus you could assume 55% based on their recent statements that they were at 35% instead of 30%. The result is $185 x .4 = $74, and that's still a conservative figure.

I'm not sure there's much left to say after that.

FWIW, I don't think "scaling of production units" is the same thing as improved production technology. It might be a distinction without a difference with GF1 since massive improvements in production technology and methods were likely necessary to triple the production capacity (maybe that is what you are saying?). Either way, I expect Tesla's continued emphasis on manufacturing technology to continue to bring costs down in later phases of GF1 and future GFs over and above the effects of scaling alone.

 

[MitchJi]

These lines can be installed anywhere that Tesla needs a multiple of one or more of these lines. I believe that it will be possible for Panasonic to produce 2170 cells in Japan, at a similar cost to the cells produced at the GF by doing two things:
1. Using one or more units of the large scale manufacturing equipment they developed for use at the GF.
2. Vertically integrating that factory. Obviously converting their existing 18650 line to produce 2170's is relatively trivial. But getting the price down requires much more than that. It probably makes more sense to shift MS-MS 2170 cell production to the NV GF (to take advantage of colocation), and setup and use more efficient production lines in Asia for Asian TM and Asian TE.

Clarification:
Actually there are three major cost advantages of the GF, the third is the alien dreadnaught technology. Elon said that (paraphrasing) if your competitors require five factories to produce what you can produce with one that they aren't really competition. Can competitors compete on price if they need three GF's to produce what Tesla can produce using one GF?
Requires correction?:

These lines can be installed anywhere that Tesla needs a multiple of one three or more of these lines. I believe that it will be possible for Panasonic to produce 2170 cells in Japan, at a similar cost to the cells produced at the GF by doing two things:

 

[MitchJi]

"$80M per GWh of machinery" is an extremely high estimate.

Have you updated your estimate since January?

@MitchJi - I would also be interested in your thoughts on this.

Given Alien Dreadnought plans (exit velocity from 0.2 m/s to 3+ m/s and volumetric efficiency from 5% to 50%), it may be less than $10M per GWh.

Most of my thoughts are in this thread.

Responding to bolded part of the sentence:
The cells are already being produced at a speed that's faster than the rate of fire of a machine gun. Why do you believe that your statement about exit velocity is relevant?

Now that they've Now that they've designed and built the custom cell manufacturing equipment for the first time, for the first phase, additional phases, and factories, will be cheaper and faster.

Indeed. Learning curve may well apply to specialized equipment and overall process design. This may be a 15% reduction in cost per unit of productive capacity every time cumulative capacity doubles. So if the roll out pace grows by 50% per year, which it should to keep pace with growth in EVS, then that would lead to an annual cost reduction of about 9%.

I don't think that it works like that. I think that most of the learning curve is complete, and if you read my posts upthread you will see why I believe it accounts for the 20% (part of the 50% GF1 cost reductions) previously mentioned by Tesla.

Specially, making the equipment more compact so that 3 times the capacity can fit into the same size building is enormous. First the cost of the building and land was about 20% of total capex. Cutting that to a third is a 13% cost reduction on the cost per capacity. Densification may also reduce the length of cables, pipes, and pathways. So the capex per capacity reduction may be greater than 13%. Moreover, I suspect that this densification of the plant could potential reduce some operating costs by reducing travel distances.

I'm not sure that works, but if it's correct it implies that Tesla's pack cost is "less than $195" -63% ='s under $73 per kWh!

 

[SageBrush]

I find Tesla cost cutting pronouncements related to kwh cost in a pack to be an exercise in guesses because for the most part we are not sure if cell or pack is being talked about; and secondly because Tesla usually talks about fractional reduction in cost without specifying the baseline.

For now I'm presuming baseline is $190 per pack kWh -- but that surely is a moving target.

So ....
A "35% reduction" implies a pack kWh cost of $123
A "50% reduction" implies a pack kWh cost of $95

The latter in the context of a (maybe ?) 55 kWh Model 3 then means $5200 cost for the pack. Now if only I knew the Tesla motor cost and what a transmission+ICE for a somewhat comparable conventional car costs I would have a frame of reference to say how close Tesla is to killing the ICE.

 

[Cattledog]

I find Tesla cost cutting pronouncements related to kwh cost in a pack to be an exercise in guesses because for the most part we are not sure if cell or pack is being talked about; and secondly because Tesla usually talks about fractional reduction in cost without specifying the baseline.

For now I'm presuming baseline is $190 per pack kWh -- but that surely is a moving target.

So ....
A "35% reduction" implies a pack kWh cost of $123
A "50% reduction" implies a pack kWh cost of $95

The latter in the context of a (maybe ?) 55 kWh Model 3 then means $5200 cost for the pack. Now if only I knew the Tesla motor cost and what a transmission+ICE for a somewhat comparable conventional car costs I would have a frame of reference to say how close Tesla is to killing the ICE.

Under $100/kWh and Tesla will be killing ICEs.

 

[SageBrush]

Under $100/kWh and Tesla will be killing ICEs.

Is there something other than blind faith in your statement ?

 

[MitchJi]

I find Tesla cost cutting pronouncements related to kwh cost in a pack to be an exercise in guesses because for the most part we are not sure if cell or pack is being talked about; and secondly because Tesla usually talks about fractional reduction in cost without specifying the baseline.

Incorrect. Tesla has always specified which they are using, which has either been mostly or always pack prices.

Under $100/kWh and Tesla will be killing ICEs.

Tesla will be killing ICEs (financially) by the end of 2017 (probably) or 2018 (definitely).

 

[Cattledog]

Is there something other than blind faith in your statement ?

Costs to then produce their cars power plant and drive train would be lower than the equivalents to ICEs, so Tesla would no longer have to rely on lower total cost to own, superior drive experience and environmental friendliness as their main selling points (which are more important to some, but initial cost is still an issue for many).

 

[MitchJi]

Both Tesla and Panasonic said that their 30-50% reduction estimates were conservative. And my estimates didn't include the reductions due to cell chemistry improvements. So you could conservatively start at $185 and deduct another 5% for cell chemistry improvements, plus you could assume 55% based on their recent statements that they were at 35% instead of 30%. The result is $185 x .4 = $74, and that's still a conservative figure.

So when Elon said that nobody is even close to their pack costs it's not because he's delusional. The reason that I believe that it's a waste of time to obsess about the exact costs because whatever it is, it's way less than their competition. So far ahead that the GM employee in charge of the Volt clearly didn't believe that $190 was even possible.

And so far ahead of Mercedes that they believe that the semi violates the laws of physics (you couldn’t make those things up ). Competition?! What competition?

There initial statements were very clear the 30-50 percent reduction estimates were due to the Gigafactory, not cell improvements. So their recent disclosures about cobalt reductions and energy density improvements of ~20% (semi range increase from five hundred miles to about six hundred miles are additional cost reductions. So it’s time to update my conservative estimate to under $70 per kWh by 2019. The 20% energy density increase should reduce their pack costs proportionally and the cobalt cost reductions are an additional reduction. Those cell improvements are very conservative because I didn’t fully account for any improvements between then and before the semi.

 

[jhm]

Thanks, this thread is good to update.

If Tesla is really poised for $70/kWh by end of the year, this can go a long ways toward improving cash flow. Once they've got volume for Model 3, what is prevent them from using new battery in Model S/X. Imagine the 100kWh pack cost dropping from say $14k to $7k. On say 100k/y Model S/X, this is a savings up to $700M that goes directly to bottom line.

Also in the shareholder letter, "We had our highest ever Q1 for [Model S and X] orders. With demand exceeding supply, we are making considerable progress with margin improvements." So could these margin improvement include battery production costs?

With so much attention right now on Model 3, Tesla could be making some really strong gains with Model S/X that analysts are not paying attention to.

Flufferbot is not now needed because the fluff has be designed out of the Model 3 pack. Might this learning and redisign apply to other packs that used fluff?

 

[TheTalkingMule]

Just like you take Elon's timelines with a grain of salt and add six months, I was also cutting in half the Tesla "5 year lead" that most there espouse. Nope, it's pretty much a legit 5 year lead.

China will be able to ramp in Giga-style and meet their own needs, but Tesla is going to have the high-end market relatively to itself for a while. You can't snap your fingers and replicate this effort or cost advantage.

 

[jhm]

ramp in Giga-style

Sounds like a K-Pop song.