Automotive lithium-ion battery costs aren’t going to fall “dramatically”

by SR on July 12, 2012

Japan lithium-ion battery production expressed in total battery capacity (kAh)

  • Conclusion: new McKinsey report on automotive batteries just implies “more of the same”. The media has picked up on a recent McKinsey report forecasting a “dramatic” decline in costs for lithium-ion batteries over the next 8 to 13 years (from 2012 to 2020~2025). By using real data for Japanese lithium-ion battery output over the past 15 years, we show that the declines forecast by McKinsey are in line with the historical norms and do not imply major technological breakthroughs. I expect no significant acceleration in cost reductions for automotive lithium-ion batteries and accordingly I do not expect the price of hybrid vehicles to drop sharply as a result of cheaper batteries. This view is congruent with testimony from Toyota to the National Academy of Science in 2009, when EVP Masatami Takimoto remarked that “significant reductions in cost will require major technological breakthroughs”. My opinion is that the hybrid market will continue to expand but that we’re unlikely to see a sudden rise in growth rates over the next 5 years. The McKinsey report is a talking point but not something on which investors need to spend time.

McKinsey’s lithium-ion battery cost forecast

The core claim made in the McKinsey article seems to be that cost of a lithium-ion battery pack will fall from “$500 to $600” per kilowatt-hour (kWh) in 2012 to $200 in 2020 and $160 by 2025. Let’s call the starting point $550, the mid-point of the range they give for 2012. A reduction in costs per kWh of $350 over the 8 years from 2013 to 2020 is equivalent to a compound annual growth rate (CAGR) of -11.9% in costs.

A cut in costs of 10-15% per year in the technology industry is, frankly, pretty poor. And that is why lithium-ion has been in some ways a disappointing technology. Compared to what we have observed in the the hard disk drive (HDD) and semiconductor memory industry, the lithium-ion battery supply chain just hasn’t made much headway in terms of reducing costs.

In 1995 a hard disk drive with a capacity of 250 megabytes would have cost you around $250. In other words, to get a hard disk with a gigabyte of capacity you would have needed to pay $1,000 or actually a little less since larger disks were cheaper in terms of dollars per megabyte. Today you can buy a 750 gigabyte drive for about $90 (a Western Digital Caviar Blue, for example). The cost per gigabyte of HDD storage now is roughly $0.12 (yes, that’s twelve cents) so over that 16-year period from 1996 to 2011 the CAGR was -43%. The following crude chart makes the point.

Price per gigabyte of hard disk drive storage 1995-2011

Note that the CAGR calculation is sensitive to start and end points, so you may get slightly different numbers depending where you cut the cake. I have seen larger figures (50% is a common rule-of-thumb) and if you take a slightly later starting point, such as 1999, I think you’d see even higher figures, closer to 100% than 50%. That’s because giant magneto-resistive (GMR) hard disk drive heads only started to ship in volume at the end of the 1990s and it was the introduction of GMR that allowed a surge in areal density. In essence the manufacturing process for GMR heads is very similar to that of a semiconductor production process so a semiconductor cost reduction curve is pretty much what we got.

If you can find a good source of DRAM prices you should find something similar, but of course you should take the price per bit rather than the price per memory module. I’ll leave that as an exercise for the interested reader, but the rule of thumb in the past has been a cost-per-bit CAGR of -35% to -50% per year.

Real historical data for lithium-ion batteries

Above we established that the cost declines implicit in the McKinsey predictions are not very impressive compared to those achieved in certain other consumer technology industries. How does this forecast of a CAGR of -11.9% from 2012 to 2020 compare to the past?

The first point to make is that the use of lithium-ion batteries in automotive applications is so new that there is no long-term data available. On the other hand, if you know where to look, there is data for consumer electronics applications. Specifically, METI has been quietly collating monthly data on lithium-ion battery production in Japan for more than 15 years. This reflects the fact that Japanese lithium-ion suppliers have always been serious players in this industry, although the Koreans have now taken the lead globally. So, off we trot to take a look at the METI data.

Japan lithium-ion battery production

Amongst other items, we have is lithium-ion battery production in millions of yen, thousands of units and in kiloamp-hours, which is the usual measure of storage capacity for a battery. From this data we can calculate an average price per unit and an average price per kAh. Unfortunately we don’t have any further detail on the composition of production, for example how many batteries of which sizes, so these figures are very much a “gross” estimate.

I’m going to assume that prices have fallen in line with costs i.e. the gains made in cost reductions are nearly all passed on to customers. I’m pretty confident that is the case, because it does not appear that lithium-ion battery manufacturers in Japan have made comfortable margins over the years. Prices expressed in terms of capacity fell from 1,178,440 yen per kAh in 1995 to 118,571 yen in 2011 – just one-tenth of the price. On the other hand, the compound annual growth rate in price per kAh was -13.4% from 1995 to 2011. Compared to the declines we mention above for the price of hard disk storage it’s not very impressive. If we take just the past 8 years, from 2003, the price has fallen from 303,223 yen to 118,571, a CAGR of -11%. Here’s the chart of the price per kAh.

Japan lithium-ion battery cost per kAh

So the decline in battery pricing that McKinsey is forecasting (-11.9%), expressed as a compound annual growth rate from 2012 to 2020, is actually very similar to what we have seen in Japan over the past 8 years. Far from being a dramatic change in the landscape, it is more of the same. The first Toyota Prius was launched in Japan in 1997. It was not until 2011 that a lithium-ion battery version went on sale, 15 years later. One reason was the natural caution of auto companies, which put the priority on safety and reliability rather than on rapid adoption of “untried” new technologies, especially volatile ones like lithium-ion. The other issue, according to auto parts suppliers, was the higher cost. Although the price of lithium-ion batteries fell, it didn’t fall enough to make lithium-ion price-competitive with competing solutions based on nickel metal-hydride (NiMH) chemistries.

Brainpower has been thrown at lithium-ion for 20 years

A lot of clever people have been working very hard on lithium-ion battery technology for a couple of decades now. The results have been disappointing. Sure, there have been regular incremental improvements and new form factors and chemistries adopted (lithium polymer, lithium-iron phosphate) and I expect there to be more in future. The fact remains that we simply have not seen a step-change in costs per unit of storage. When hard disk drive heads moved from thin-film to MR to GMR (with corresponding improvements in platter and suspension technology) the areal density curve steepened dramatically and the cost of storage plunged. No such revolution has taken place in lithium-ion batteries.

I am not saying that the technological revolution in lithium-ion can’t and won’t happen, but I’m cautious. I had lunch a while back with a very smart young guy working at the cutting edge of academic lithium-ion battery research. His team are working on getting a better understanding of certain properties of electrolytes and separators that could lead to a ‘great leap forward’ – this is genuinely exciting stuff. On the other hand I was surprised that after twenty years of commercial product development, we still know so little about such basic issues, a fact that my academic friend acknowledged. The key to lithium-ion performance seems to be a bunch of very hard, interlinked problems that have so far defied the kind of analysis that leads to order-of-magnitude improvements.

Raw material concerns

Another concern I have is raw material prices for lithium-ion batteries. Process costs should go down as volumes rise, but if supply of any of the main materials and components is constrained, costs might not fall as they have done in the past. Note that automotive batteries are huge compared to consumer electronic cells. As of January 2012, METI has started to collect and report automotive lithium-ion battery statistics. In April 2012, production of lithium-ion batteries for automotive use came to 3.4 million units, compared to 72.6 million for other lithium-ion batteries (about 5%), but the capacity in kAh was 58,227 kAh compared to 130,780 kAh for other lithium-ion batteries – about 45%. There’s little doubt in my mind that the rapid rise in demand for automotive batteries is going to put pressure on the lithium-ion supply chain, probably lead to higher raw material costs (putting pressure on margins) and hampering efforts by battery companies to reduce prices.

Stocks in Japan

The perennial favourite automotive lithium-ion play in Japan is GS Yuasa. This is a favourite of individual investors and can be very volatile due to crowds of, ah, not necessarily well-informed people surging in and out of the stock. Of course, being able to make lithium-ion batteries is not enough: you need to be able to sell them to somebody. A couple of brokers have recently cut their ratings on the stock to reflect slow sales of electric vehicles at Mitsubishi Motors, a major customer.

{ 1 comment }

Sterling Brenner November 21, 2013 at 11:22

Interesting read. In regards to the price of raw materials for li-on batteries, I encourage you to check out this article, which details the possible use of Rice (yes, Rice!) for longer lasting batteries:

http://firstlook.pnas.org/battery-materials/

{ 1 trackback }

Previous post:

Next post: