Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

The broader implications of Semi's ~$70/kWh battery pricing

This site may earn commission on affiliate links.

KarenRei

ᴉǝɹuǝɹɐʞ
Jul 18, 2017
9,618
104,600
Iceland
As can be seen from the $30k difference in price between the 500 and 300 mile range semis, the cost for Tesla to produce the battery packs for Semi appear to have plunged.

Assume: 900kWh pack = 540kWh for 300mi = 360kWh for $30k = $83/kWh; assuming the usual 25% margin = $67/kWh. This is actually getting close to - but still plausibly above - raw materials costs for the packs, which should be somewhere around $50/kWh. Approaching raw materials prices has, after all, always been a goal of gigafactory.

But this isn't the point of this discussion; the point is something else entirely. Namely: Tesla's promise of $0,07/kWh renewable electricity for the trucks.

I think some of us may have blown this off as read-the-fine-print hype. Along the lines of "they'll just put up solar awnings which contribute an insignificant fraction of the total, and call that solar energy". But even if we were to assume that, how do we account for the $0,07/kWh power costs? That's cheaper than commercial rates, ignoring the expensive demand charges, and Tesla has a lot of capital costs to amortize (chargers, powerpacks, etc). For example, the recent project to deploy 100MW/129MWh in Australia cost $50M. A typical 8-stall supercharger station without a battery buffer is said to cost $250k, and that's 1/12th the power, so that's meaningful, but the real issue is those powerpacks - there's serious capital costs here!

Industrial power rates are usually in the $0,06-0,07/kWh range. But with industrial power, you pay for all of the connecting infrastructure and a lot of additional overhead. And again, what about the megacharger capital costs?

Some analysts have suggested that the Megacharger stations are a temporary loss leader to sell trucks - that by selling power for cheaper than they can buy it, they're just deferring costs to the future, to a time where they have more revenue and better economics, while in the meantime jumpstarting a switch to electric semis which won't revert. But that would be a hugely expensive loss leader - and selling T-shirts in a truck stop certainly won't pay for it.

I think there's something much bigger that we're missing here, and it comes down simply to: look at the costs on those battery packs. Tesla has stated that Semi is using the same battery chemistry as their grid storage products (aka, NMC). Meaning that these figures should also be directly applicable to future powerpacks. So: 8 stall megacharger storing 12 hours of power with a 30% usage rate and 1MWh allocated per vehicle charge? Only $4m (for 58MWh). This says nothing about Tesla's costs in transforming power, only that in terms of storing it, their powerpack costs should be going way, way down.

But here's where things get interesting. Let's double our battery size to 24h / $8m. Add in a (proportionally) small grid link as backup (or a peaking plant - NG or biomass). Now build your own solar farm. Not mere solar awnings, but a whole solar farm somewhere near the megacharger. Musk said the megacharger stations will be solar powered - let's try taking him at his word. You have a battery buffer and a grid backup; it's effectively baseload. Solar is getting almost absurdly cheap, with average plants coming in at $1,5/W installed (and falling). Let's go with a 25% capacity factor and $1,2/W. Our 8-stall megacharger is averaging ~5MW (counting losses). That's a $24M plant. Add in various overhead (such as transmission to the megacharger) and say $28M. Let's amortize (3% interest) the battery pack over 15Y ($55k/mo) and the solar plant over 30Y ($118k/mo). So $173k/mo (omitting any backup / grid costs). The megacharger takes in $245k per month in charging fees - +$72k/mo in the black. So long as everything else (backup, chargers, facilities, etc) is less than $15M amortized over 25 years, the station as a whole is in the black - before accounting for "truck stop sales" (which are where most gas stations / truck stops today make most of their money).

And solar prices keep on falling. In India some plants are coming in nearly down to $0,60/W. US prices may well average under $1/W in a few years - and expect Tesla to try to stay on the forefront of that curve. Solar capacity factors will of course vary depending on location (wind is also an option), but then again, the above 25% is lower than the current US average. Now, maybe Tesla will want longer than a 24h buffer - maybe 48, even 72h, to reduce the required backup grid connection / peaking needs - but even that shouldn't ruin the equation. 15Y is the current expected lifespan of powerpacks, but it may improve, and Tesla gets to recycle them at the end for those raw materials (which are, as noted, increasingly a large chunk of the battery costs). The solar plant should last longer than 30y (although assuming limited lifespans on the powerpacks and solar farm helps account for maintenance costs, which aren't listed above). Seasonal fluctuations change solar power output, but then again, Tesla can sell excess solar power in high times just as readily as they can buy grid power in low times; solar tracks demand well. In short: I think they can actually do this. Tough? Yeah. And they'll need to prove their economic case to get those kind of interest rates. But I think they can do it.

What happened here to make it possible? Two things.
  • Solar is cheap... with a catch: The low cost of solar power comes with peaking requirements; it needs to be buffered for the night and days with little sun. But the megachargers need their own battery buffer either way. Tesla gets a two-for-one here.
  • Tesla appears to have changed the game on grid storage prices. Seriously, if they can manufacture storage for under $70/kWh and sell it for $85/kWh or so with a 25 margin, solar and wind have just become a lot cheaper and easy to deploy for everyone.
In short, this what I think is going on with those pricing figures. And if there's any word that describes the implications, it's this: profound.
 
Last edited:
Interesting thoughts, thanks @KarenRei

I stumbled over a Seeking Alpha article that was pondering that if Tesla sells 1,000 kWh battery packs in Semis it is foregoing 10x 100 kWh car sales, and they would be more profitable, and thus the Semi is doomed ...

... the article bases this on a Cobalt Cliff which will limit (overall) battery production.
 
"Cobalt cliff" and the talk of "exhausted planetary reserves" is nonsense (just another person who doesn't understand what is meant when someone talks about "reserves" figures in general). Cobalt can be found anywhere nickel, copper, silver, and a number of other associated minerals are, and is more common than lead, 2/5ths as common as copper, and 1/3rd as common as nickel (all of which are produced in *vastly* larger quantities). Primary cobalt is mainly is produced in Congo today (with most mines of associated minerals not recovering cobalt) simply because they have the highest cobalt ratios in their ores. But at current cobalt prices you can support getting it from almost anywhere.

Reserves figures are all based on explored assets, with the recoverable percent based on a given market price. But exploration dwindles once reserves figures reach a point where producers think they have "sufficient" reserves, while a doubling in market price of a resource generally leads to an orders-of-magnitude increase in the recoverable amount. Sudden increases in demand always lead to price spikes (as we've seen), but those price spikes almost invariably collapse once some combination of new exploration and production re-adjusts to the new demand curve. In the case of cobalt, it should if anything be quicker to catch up because recovery can just be added to existing tailing streams (although these sort of quick fixes aren't as cheap as the longer-term cost reduction you get from better exploration and processes)

Or to put it another way: see the Simon-Erhlich wager.

And remember: Tesla's prices today are with today's raw material prices at spike levels. Not what they'll be like in the long run when the production side catches up.
 
Last edited:
Love the OP, yet I'd be careful concluding that Tesla has already attained these cheaper battery costs. If I'm Elon, I take a look at the declining cost trend on batteries, look out 2-3 years when Semi will hit the market, and price the Semi today using future battery cost expectations. I can't imagine he'd price Semi any other way, frankly. So it doesn't necessarily portend anything near-term about reduced cost/improved margin/reduced pricing for S/X/3 (not that you implied that it did, in your OP, it's just where my brain headed).
 
Rather than a grid tie for their solar farm, they could have their own natural gas electrical generation using fuel cells. That would avoid paying any tiered or peak energy prices. And be more reliable.

This assumes that natural gas delivery is more reliable than the grid. It also ignores the high cost and short lifetime of fuel cells. Fuel cells are too expensive and short-lived to be economically feasible. And finally it ignores the fact that the grid is becoming more and more efficient as more and more utilities install sustainable power generation.
 
You don't put a high capital cost generator (aka, fuel cells) as peaking. Peaking plants are used infrequently. They can be inefficient; the primary constraint on them is simply "lots of power for little money when fired up". You're ideally only using them for a couple percent of any given year.

A grid interconnect (giving access to a diverse array of generation options over a broad region) should be cheaper backup than having your own peaker (which costs something like $1/W, not counting fuel). It also would let them sell power when it's in excess.
 
As can be seen from the $30k difference in price between the 500 and 300 mile range semis, the cost for Tesla to produce the battery packs for Semi appear to have plunged.

Assume: 900kWh pack = 540kWh for 300mi = 360kWh for $30k = $83/kWh; assuming the usual 25% margin = $67/kWh. This is actually getting close to - but still plausibly above - raw materials costs for the packs, which should be somewhere around $50/kWh. Approaching raw materials prices has, after all, always been a goal of gigafactory.

But this isn't the point of this discussion; the point is something else entirely. Namely: Tesla's promise of $0,07/kWh renewable electricity for the trucks.

I think some of us may have blown this off as read-the-fine-print hype. Along the lines of "they'll just put up solar awnings which contribute an insignificant fraction of the total, and call that solar energy". But even if we were to assume that, how do we account for the $0,07/kWh power costs? That's cheaper than commercial rates, ignoring the expensive demand charges, and Tesla has a lot of capital costs to amortize (chargers, powerpacks, etc). For example, the recent project to deploy 100MW/129MWh in Australia cost $50M. A typical 8-stall supercharger station without a battery buffer is said to cost $250k, and that's 1/12th the power, so that's meaningful, but the real issue is those powerpacks - there's serious capital costs here!

Industrial power rates are usually in the $0,06-0,07/kWh range. But with industrial power, you pay for all of the connecting infrastructure and a lot of additional overhead. And again, what about the megacharger capital costs?

Some analysts have suggested that the Megacharger stations are a temporary loss leader to sell trucks - that by selling power for cheaper than they can buy it, they're just deferring costs to the future, to a time where they have more revenue and better economics, while in the meantime jumpstarting a switch to electric semis which won't revert. But that would be a hugely expensive loss leader - and selling T-shirts in a truck stop certainly won't pay for it.

I think there's something much bigger that we're missing here, and it comes down simply to: look at the costs on those battery packs. Tesla has stated that Semi is using the same battery chemistry as their grid storage products (aka, NMC). Meaning that these figures should also be directly applicable to future powerpacks. So: 8 stall megacharger storing 12 hours of power with a 30% usage rate and 1MWh allocated per vehicle charge? Only $4m (for 58MWh). This says nothing about Tesla's costs in transforming power, only that in terms of storing it, their powerpack costs should be going way, way down.

But here's where things get interesting. Let's double our battery size to 24h / $8m. Add in a (proportionally) small grid link as backup (or a peaking plant - NG or biomass). Now build your own solar farm. Not mere solar awnings, but a whole solar farm somewhere near the megacharger. Musk said the megacharger stations will be solar powered - let's try taking him at his word. You have a battery buffer and a grid backup; it's effectively baseload. Solar is getting almost absurdly cheap, with average plants coming in at $1,5/W installed (and falling). Let's go with a 25% capacity factor and $1,2/W. Our 8-stall megacharger is averaging ~5MW (counting losses). That's a $24M plant. Add in various overhead (such as transmission to the megacharger) and say $28M. Let's amortize (3% interest) the battery pack over 15Y ($55k/mo) and the solar plant over 30Y ($118k/mo). So $173k/mo (omitting any backup / grid costs). The megacharger takes in $245k per month in charging fees - +$72k/mo in the black. So long as everything else (backup, chargers, facilities, etc) is less than $15M amortized over 25 years, the station as a whole is in the black - before accounting for "truck stop sales" (which are where most gas stations / truck stops today make most of their money).

And solar prices keep on falling. In India some plants are coming in nearly down to $0,60/W. US prices may well average under $1/W in a few years - and expect Tesla to try to stay on the forefront of that curve. Solar capacity factors will of course vary depending on location (wind is also an option), but then again, the above 25% is lower than the current US average. Now, maybe Tesla will want longer than a 24h buffer - maybe 48, even 72h, to reduce the required backup grid connection / peaking needs - but even that shouldn't ruin the equation. 15Y is the current expected lifespan of powerpacks, but it may improve, and Tesla gets to recycle them at the end for those raw materials (which are, as noted, increasingly a large chunk of the battery costs). The solar plant should last longer than 30y (although assuming limited lifespans on the powerpacks and solar farm helps account for maintenance costs, which aren't listed above). Seasonal fluctuations change solar power output, but then again, Tesla can sell excess solar power in high times just as readily as they can buy grid power in low times; solar tracks demand well. In short: I think they can actually do this. Tough? Yeah. And they'll need to prove their economic case to get those kind of interest rates. But I think they can do it.

What happened here to make it possible? Two things.
  • Solar is cheap... with a catch: The low cost of solar power comes with peaking requirements; it needs to be buffered for the night and days with little sun. But the megachargers need their own battery buffer either way. Tesla gets a two-for-one here.
  • Tesla appears to have changed the game on grid storage prices. Seriously, if they can manufacture storage for under $70/kWh and sell it for $85/kWh or so with a 25 margin, solar and wind have just become a lot cheaper and easy to deploy for everyone.
In short, this what I think is going on with those pricing figures. And if there's any word that describes the implications, it's this: profound.
I have been posting this graphic for the last year. The projection is bases on the vast volumes that the Gigafactory will be pumping out. EVs are happening....at a pace that is breathtaking.
2170 Cost Prediction.jpg
 
  • Love
Reactions: 1 person
$55's a bit much. The raw materials alone cost basically that much***, and there's going to be *some* costs to turn them into batteries. But I don't think they're going to be that much over it. Even if one wants to pretend that the difference between 300mi and 500mi Semi costs isn't linearly correlated with battery costs, the batteries have to be cheap regardless to justify the Semi prices, period (everyone underestimates how much motors, inverters, chargers, etc add to EV costs).

The whole point of Gigafactory was to make battery costs get closer to raw material costs. And by all signs, Tesla has either done so, or thinks they will be there by the time that Semi launches. Which has amazing implications for the future :)

*** While the raw materials appear to be around $50/kWh today, once the raw materials production curve matches back up to the new demand curve - which it will - raw materials should to back down to around $35/kWh or so. And that will be a lovely thing :)
 
  • Like
Reactions: David99
$55's a bit much. The raw materials alone cost basically that much***, and there's going to be *some* costs to turn them into batteries. But I don't think they're going to be that much over it. Even if one wants to pretend that the difference between 300mi and 500mi Semi costs isn't linearly correlated with battery costs, the batteries have to be cheap regardless to justify the Semi prices, period (everyone underestimates how much motors, inverters, chargers, etc add to EV costs).

The whole point of Gigafactory was to make battery costs get closer to raw material costs. And by all signs, Tesla has either done so, or thinks they will be there by the time that Semi launches. Which has amazing implications for the future :)

*** While the raw materials appear to be around $50/kWh today, once the raw materials production curve matches back up to the new demand curve - which it will - raw materials should to back down to around $35/kWh or so. And that will be a lovely thing :)
Well said, but my prediction is based on the fact that competition always races back through the supply chain, all the way back to the mines....reducing costs. There are so many companies working to "catch the wave" in LiIon batteries that I believe it will happen very quickly. A good example is solar panels, where prices have fallen like a stone in just a few years.
Price_history_of_silicon_PV_cells_since_1977.svg.png
Whoosh!
 
  • Like
Reactions: KarenRei
I do think it's a likely expectation that - with higher production rates - some of the lower-volume EV raw materials (cobalt, lithium) may well end up lower than their historic prices once EV demand pushes up the scale in which they're mined and causes a new wave of exploration and extraction technology. There's a lot of historical analogues to point to in this regard. I only differ in that I don't see that happening in the short term. Somewhere in the 2,5 to 6 year timeframe :)

EVs shouldn't meaningfully impact copper or alumium prices, hydrocarbon prices (even fluorinated ones), or manganese. Depending on what form of carbon is used on the anodes, if it's a broader market than just EV anodes, they may impact that market. It'll be a while before EVs significantly impact the nickel market since it's so large; 2025 or so it should be pretty meaningful, although not huge. But with cobalt and lithium, the expansion of the market has the potential (and I stress potential) to lead to discovering new deposits and production methods that drive prices even lower than their pre-spike levels.
 
  • Like
Reactions: David99
I installed PV at home that amortizes out to 2.5 cents a kWh. Admittedly my labor, but first install and lots of errors that cost me money. I know that I could build it for well under 2 cents a kWh now. For another data point of just how cheap PV can be, international PPA contracts are well under 2 cents a kWh today. I think the cheapest thus far is 1.6 cents a kWh. Note that these kWh prices include a profit for the developer. Also note that Tesla will presumably be using its own PV factory in Buffalo in production. No Trump BS to worry about. Financing at the cost of Tesla's ability to raise cash or their opportunity cost.

Regarding pack price:
If a mile is 2 kWh as Tesla has reported,
then a 200 mile range is 400 kWh
$30,000/400 kWh = $75 a kWh retail

I agree with OP's conclusions. There is profound change here, and clean energy/clean transport is going to kill diesel on money considerations alone. The only real unanswered question here is when to short diesel truck manufacturers.

 
Last edited:
Even if one wants to pretend that the difference between 300mi and 500mi Semi costs isn't linearly correlated with battery costs,
What are you trying to say here?
Any two (different) points form a line, thus a linear relationship. The price difference could be 1x battery cost, it could be 2x, it could be 0.5 x. We don't know which of the versions of semi Tesla makes more profit on, so to directly infer that the price difference exactly matches the battery price delta (with or without fixed margin on top) has nothing to supporting. Same as we can't say that <2kWh means 2kWh ==> 1 MWh pack.
At 1.6kWh/mile (physics based assumptions):
800 kWh pack * $125/ kWh = $100k leaving $80k for chassis/ motors and overhead. 300 mile version pack could be 480 kWh or $60k, giving Tesla an extra $10k profit on it.

the batteries have to be cheap regardless to justify the Semi prices, period
Depending on the power supplying agreement, Tesla could be using the razor strategy. 1 million miles @ 1.6 kWh/ mile @ 2 cents profit per kWh = $32,000, or an additional 18% margin on the 180k tractor sales price (a million miles is many years, with improvement in solar production costs, profit could be higher) .

Analogy warning: does the sales price difference between 16GB/32GB/64GB IPhones have anything to do with flash memory costs?