5 ms·

Do you think batteries are getting more expensive over time?

XorNot 2y ago
I think the assumption they would get cheaper given a massive change in their adoption patterns is unfounded. We know that lithium prices have gone up from the rise in EVs, and this has in part driven Tesla to switch to LiFePO4's rather then Li-Ion packs in some of their cars. Same effect with coltan. (although lithium is declining currently because EV sales in China have slowed). What would be the effect of the demand for terawatt-hour levels of batteries? Even if lithium is replaced with sodium in the application, the manufacturing capacity isn't there and won't be for quite some time - and the higher-value applications will get first take (i.e. EVs). Even if the lithium demand was saturated battery wise, at the steady state point the cost can then simply stabilize around the refurbisment/recycling cost of lithium/sodium/whatever from batteries. Sigmoidal adoption curves look exponential at the start, linear in the middle,and then become asymptotically flat.
- kragen 2y ago
  > although lithium is declining currently because EV sales in China have slowed have they? 'one in three' of all new vehicle registrations in china were electric last year https://www.iea.org/reports/global-ev-outlook-2024/trends-in-electric-cars https://www.iea.org/reports/global-ev-outlook-2024/trends-in... or 37% https://autovista24.autovistagroup.com/news/ev-registrations-in-china-nearly-double-from-2023/ https://autovista24.autovistagroup.com/news/ev-registrations... and the absolute number seems to be doubling this year, though they only expect a 45% market share these figures do include both hybrids and pure battery models, though in general the 'experience curve' says that things get exponentially cheaper as you produce exponentially more of them
  - XorNot 2y ago
    The experience curve is asymptotic[1]. What do we get if we look at say, FLOPS per US $ over time? This[2]. There is no reason to think batteries are any different, the only margin you have is to argue over which part of the curve you think we're in. A sigmoid curve[3]. [1] https://www.extension.iastate.edu/agdm/wholefarm/html/c5-208.html https://www.extension.iastate.edu/agdm/wholefarm/html/c5-208... [2] https://www.researchgate.net/figure/Number-of-flops-per-United-States-dollar-by-year-Almost-mirroring-the-change-in-computer_fig3_367400092 https://www.researchgate.net/figure/Number-of-flops-per-Unit... [3] https://en.wikipedia.org/wiki/Sigmoid_function https://en.wikipedia.org/wiki/Sigmoid_function
    - jaggederest 2y ago
      That FLOPS graph is an exponential scale... so that straight line "after the sigmoid" is an exponential growth curve. I'm not sure it makes the point you're trying to make.

I think the assumption they would get cheaper given a massive change in their adoption patterns is unfounded. We know that lithium prices have gone up from the rise in EVs, and this has in part driven Tesla to switch to LiFePO4's rather then Li-Ion packs in some of their cars. Same effect with coltan. (although lithium is declining currently because EV sales in China have slowed). What would be the effect of the demand for terawatt-hour levels of batteries? Even if lithium is replaced with sodium in the application, the manufacturing capacity isn't there and won't be for quite some time - and the higher-value applications will get first take (i.e. EVs). Even if the lithium demand was saturated battery wise, at the steady state point the cost can then simply stabilize around the refurbisment/recycling cost of lithium/sodium/whatever from batteries. Sigmoidal adoption curves look exponential at the start, linear in the middle,and then become asymptotically flat.

kragen 2y ago
> although lithium is declining currently because EV sales in China have slowed have they? 'one in three' of all new vehicle registrations in china were electric last year https://www.iea.org/reports/global-ev-outlook-2024/trends-in-electric-cars https://www.iea.org/reports/global-ev-outlook-2024/trends-in... or 37% https://autovista24.autovistagroup.com/news/ev-registrations-in-china-nearly-double-from-2023/ https://autovista24.autovistagroup.com/news/ev-registrations... and the absolute number seems to be doubling this year, though they only expect a 45% market share these figures do include both hybrids and pure battery models, though in general the 'experience curve' says that things get exponentially cheaper as you produce exponentially more of them
- XorNot 2y ago
  The experience curve is asymptotic[1]. What do we get if we look at say, FLOPS per US $ over time? This[2]. There is no reason to think batteries are any different, the only margin you have is to argue over which part of the curve you think we're in. A sigmoid curve[3]. [1] https://www.extension.iastate.edu/agdm/wholefarm/html/c5-208.html https://www.extension.iastate.edu/agdm/wholefarm/html/c5-208... [2] https://www.researchgate.net/figure/Number-of-flops-per-United-States-dollar-by-year-Almost-mirroring-the-change-in-computer_fig3_367400092 https://www.researchgate.net/figure/Number-of-flops-per-Unit... [3] https://en.wikipedia.org/wiki/Sigmoid_function https://en.wikipedia.org/wiki/Sigmoid_function
  - jaggederest 2y ago
    That FLOPS graph is an exponential scale... so that straight line "after the sigmoid" is an exponential growth curve. I'm not sure it makes the point you're trying to make.