Generic selectors
Exact matches only
Search in title
Search in content
Search in posts
Search in pages
Filter by Categories
Asset Allocation
Data update
Development News
Equity markets
Exploration News
News & Views
The Drill Bit
The Future
Trade & Demand


The dangers of understating the magnitude of the battery material supply/demand imbalance

  By: Matt Fernley

Posted on - 20 Aug 2021

I wanted to talk about the Nature article on battery raw materials that’s been doing the rounds this week. The article, Electric Cars: The Battery Challenge (Nature, 19 August 2021), is an otherwise excellent discussion of a lot of the issues with sourcing materials for electric cars. Unfortunately, there’s a big “but”. And the “but” is in its treatment of primary battery raw materials.

While the author, Davide Castelvecchi, has clearly spoken to a lot of experts on batteries, recycling and other elements of the supply chain, maybe he hasn’t known exactly which questions to ask, because we get a discussion almost entirely on ternary batteries with little to no mention of LFPs (and their ability to lower demand for Nickel, Cobalt and Manganese) and we also get only three paragraphs on the impact of extractive industries on the battery industry. All the “analysis” on raw materials is effectively based on BNEF’s Long-Term Electric Vehicle Outlook for 2021 and the general conclusion, based on a quote from the BNEF analyst, is that “temporary shortages [of battery raw materials] and dramatic price swings… [will] work themselves out”.

I very much disagree with this view and that’s what I want to talk about in this article. BNEF is not the only organisation touting this view that battery raw material supplies are not at critical levels. A number of global investment banks are also in on this act. Ordinarily this wouldn’t be a problem, but the issue is that these groups are unfortunately being listened to by automakers which are therefore not making capital available for new supply of materials. This, as much as anything else in my view, is endangering the EV event and therefore the whole Energy Transition.

Most of those organisations that believe that battery raw materials are likely to stay in oversupply are basing their conclusions on faulty analysis, in my view. I’ve noted in two recent blogs why I think that analysts are understating demand and overstating battery raw material supply forecasts. I’m now going to turn my attention to BNEF’s analysis on battery raw materials supply/demand balances included in its Long-Term Electric Vehicle Outlook.

The two key elements of BNEF’s analysis that I disagree with the most are:

  • As I’ve stated in my Blog a number of times, I believe that near-term forecasts for EV sales are understated and believe that BNEF’s are as well; and
  • BNEF has used USGS reserve data to illustrate how much supply is left, but it used it incorrectly and not corrected for metal recovery from the production process. Effectively, BNEF is comparing apples with oranges in comparing material in its raw state in the ground with material demand in its processed state. There are lots of iterations between raw and processed materials.

BNEF (and consensus) EV sales forecasts understated in near-term

So, lets talk about EV sales forecasts first of all. I’m not going to labour the point here, but I wrote a detailed Blog on 24 November 2020 highlighting why I felt that consensus was too low on EV sales forecasts, and I still believe consensus is too low.

To summarise, my view was that:

  • Consumer adoption rates have gotten steeper and steeper over time;
  • The effect of government support and demographic drivers would make EV demand growth faster than in previous consumer adoption cycles;
  • In a normal take-off event, growth rates are higher at the beginning, slowing as the event matures, while most sell-side analysts have growth rates slower at the beginning.

Given the continuing rapid growth in EV sales since November 2020, a number of analysts (including both myself and BNEF) have raised their EV sales forecasts, but I’m still some way above BNEF (and consensus) on my forecasts in the medium term.

Now, one million vehicles doesn’t look like a big difference, but just to illustrate that for a likely average battery size of 60kWh (by that time), that translates into c.8Kt of lithium metal or c.40Kt of lithium carbonate equivalent (LCE), equivalent to c.10% of current lithium production. So in 2030, likely with larger battery sizes, the variance could be c.30-40% of current lithium production and similar magnitudes in other materials.

Use of reserves data

In the BNEF analysis, the team makes the argument that there’s no scarcity of battery raw materials and that, compared to current known reserves, forecast demand is practically irrelevant. In the Nature article the author makes the quite reasonable assertion that as battery raw material prices rise, more resources will become economically extractable and hence reserves will increase.

Unfortunately, the BNEF analysis is flawed and way too simplistic. BNEF has taken a common approach across different battery raw materials, but that in itself is too simplistic because it doesn’t appreciate the specific intricacies of each material in a battery industry context. To examine this, I’m going to consider the arguments on a material by material basis.


Even though it’s not sourced in the BNEF charts, it seems likely that BNEF based their reserves estimates on USGS data. In its 2021 Lithium Mineral Commodity Summary, USGS reports 21Mt of global lithium reserves (in the form of lithium metal), of which 9.2Mt or 44% are from Chile. I contacted USGS to understand how they collected their reserves data for lithium because this seemed a bit high to me. I understand that the Atacama is the world’s premier basin for lithium brine but given the focus on ESG, groundwater draws and the likely impact of higher Chilean royalty rates I wondered whether all this material was genuinely reserves, or actually resources.

USGS noted that a number of brine producers don’t disclose reserves which I already knew, although it continues to floor me. How it’s possible for the SEC (and ASX, by the way) to allow listed companies that rely on a natural resources not to publish a regular update of their reserves data is beyond me. Surely that’s the first thing that an investor needs to know? How long a company’s source of material will last…

However USGS pointed me in the direction of SQM’s annual report from which it had sourced the data for Chilean reserves (and noted that since they couldn’t access Albemarle’s data, that in fact the Atacama reserves were probably understated).

While the reserve data was there as advertised, I was very interested in the notes to that table. To quote, SQM said “metric tonnes of…lithium…considered in the…reserves are shown before losses from evaporation processes and metallurgical treatment. The recoveries of each ion depend on both brine composition and the process applied to produce the desired commercial products” and “Recoveries for lithium vary from 34% to 60%”. [SQM Annual Report, 2020, p48].

So SQM is saying that only 34-60% of its stated reserves are recoverable using the processing methods that it applies. I would assume that that is the case for other brine resources in the world, so, straight away, one has to correct brine reserve tonnages down by c.50%… In addition to that you’ve got the processing recovery from lithium chloride to lithium carbonate and, as I’ve noted in previous articles, a large proportion of lithium production from brine is not battery grade (it can range between 10-40%) but let’s call those corrections together 20%. So, effectively, for every one tonne of brine reserves, 0.4 tonnes is usable as battery grade lithium. So that correction needs to be used for most of the reserves from Chile, China and Argentina.

In the hard rock sector we see a similar situation. Lithium recoveries when hard rock deposits are mined and concentrated probably average out at c.70%. Conversion into lithium hydroxide and lithium carbonate then probably has an 85-90% recovery. Taking the two together, recovery to battery grade lithium chemicals from hard rock sources is probably of the order of 60%. We apply that correction factor to the reserves for the rest of the world.

So, straight away the reserves data used is overstating the actual realistic supply of lithium by over 50%.

If we then add in BMR’s forecasts for lithium demand based on our EV forecasts, one can see a very different situation for the reserves/demand balance compared to BNEF’s analysis.

Within our analysis, current lithium reserves (corrected for recovery) would be exhausted by 2040. If EV take up is faster than forecasted, then reserves could potentially be used up much faster.

I should also note that within BMR’s corrected reserves we included an estimate of lithium metal reserves for Albemarle’s Atacama asset to ensure we are being conservative.

I will make one final point on the lithium analysis. The article refers to “temporary” supply shortages and dramatic price swings. In a recent issue of Battery Materials Review I compared current supply forecasts for lithium with the material that would need to be utilised to meet our current EV sales forecasts. The analysis was chastening because it suggested that EV sales forecasts need to be coming down, not up. Because by the second half of this decade there won’t be enough lithium supply to hit our EV sales forecasts. I don’t call nine years a “temporary” supply/demand gap. I call it an emergency…


As with many studies on high purity manganese, the BNEF “analysis” is, in my view, a complete disaster. The reason: The BNEF analysts have not separated the core material of high purity manganese out from the manganese in use in the steel industry. I’ve written on high purity manganese previously in my LinkedIn blog, which you can find here and discussed the situation in more detail in my webinars here and here, so I won’t regurgitate all that in this piece.

But in summary, not all manganese deposits are created equal, and (in my estimation) probably less than 1% of currently known manganese reserves globally are economically, environmentally and chemically suitable to be upgraded to the kind of high purity manganese utilised in high performance EV batteries. Current resources (not reserves) of manganese for high purity battery grade production are just 3% of current global manganese resources…

On top of that there’s the processing recovery, which is likely to be of the order of 60-70%. When we correct for all of those factors we get a corrected reserve number of 8.5Mt of manganese metal, vs the BNEF estimate of 1300Mt of manganese metal.

So for BNEF to just whack in USGS’s reserve estimate for steel grade manganese is just plain wrong.

As with lithium, in manganese we see a very different situation to BNEF; in our analysis current reserves for high purity manganese only have an eleven-year life. That’s close to critical levels…


Now cobalt is an industry that I’m less excited about than many battery materials industries and I don’t have too much to say about BNEF’s analysis. I share the Nature article’s view that cobalt is being thrifted out of most EV batteries (particularly with the greater usage of LFP batteries which is not discussed in detail in the article).

But I will note on the supply side that cobalt is a by-product in most of the mines in which it is produced and, as a result, its recovery is much lower than for primary products. Cobalt recovery in sulphide orebodies can be as little as 65% and then further processing can take total cobalt recovery to c.60%. In laterite orebodies, total cobalt recoveries are often higher at c.85% (including downstream processing), but those recovery issues have also not been factored into BNEF’s analysis.

Similarly to in the other materials, when recoveries are factored into the analysis, the BNEF exercise comes up short and our data suggests that there are c. 14 years of remaining cobalt reserves extant, substantially less than implied by BNEF’s analysis.


In my view, the analysis for nickel is also extremely flawed. The charts shown in the report only seem to consider nickel usage for EVs, while we know that there is quite a substantial additional market for nickel out there. It’s called stainless steel. So showing a chart with Class 1 nickel reserves and then only considering EV demand for nickel (which is currently only a small percentage of demand for class 1 nickel) is extremely misleading, in my view. The “other” applications of nickel, a number of which also require class 1 nickel, currently contribute more nickel demand than the EV sector.

This, on its own, makes the analysis lopsided, and when one adds in the impact of metallurgical recoveries, it again moves the goal posts.

In my charts, I’m including a correction for class 1 nickel demand for non-EV uses. I’m not sure of the provenance of the BNEF reserve number. USGS quotes 94Mt of total global nickel reserves, while BNEF utilises 73.8Mt for Class 1 nickel reserves. BMR suggests that 50/50 is a reasonable level for Class 1 vs Class 2 nickel, which implies Class 1 nickel reserves of c.45Mt of contained nickel.

If we then correct for recoveries, we are left with a much lower number.


BNEF doesn’t include graphite in its analysis at all, which is surprising considering that graphite is a core component of lithium-ion batteries. All lithium-ion batteries of whichever cathode chemistry include graphite anodes, whether manufactured from naturally-derived or synthetic graphite or a mix of the two.

We’ve decided to recreate the BNEF analysis for graphite so that readers can truly compare all of the key lithium-ion battery inputs.

There are a number of caveats in our analysis of graphite demand vs graphite supply:

  • USGS records 320Mt of natural graphite reserves globally. c.31% of that material is amorphous graphite which is not suitable for battery anode manufacture.
  • Flake graphite is the primary natural graphite product used for battery anode manufacture, but it comes in six different size fractions, with super fines material and super jumbo size material generally less-preferred for anode manufacture (fines because it is often not crystalline and super jumbo has non-anode applications which make it too highly-priced for use in anode manufacture). Our coverage of graphite projects suggests that on average 27% of flake material is fines and 4% is Super-Jumbo.
  • That means that 31% + 21% = 52% of global natural graphite resources are likely not suitable for battery anode production.
  • Graphite recoveries from concentration of mined material are of the order of 88% and processing recoveries (into spherical graphite) are of the order of 50%.
  • It also behoves us to make a correction for the number of occurrences where the chemical or physical characteristics of the graphite make it unsuitable for use in battery manufacture. We use a correcting factor of 10%.
  • That means that BMR’s estimate of global graphite reserves corrected for recovery is of the order of 64Mt.
  • On the demand side, synthetic graphite is used extensively in battery anodes. In the recent past it has been preferred as a feedstock, but its high economic and environmental cost of production (it is power-intensive), plus a diminishing resource of feedstock (most feedstock is derived from blast furnace steelmaking or petroleum refining) mean that we see synthetic graphite output growth being substantially less than is possible from natural graphite in coming years. BMR forecasts that the percentage of synthetic graphite used in battery anode production will decline from c.55% now to c.35% by 2040.
  • On the demand side, there are a number of other high growth segments that use natural graphite, most particularly within the expandable graphite segment.
  • Historically, graphite has not been recycled too much from battery anodes, although there are a number of technologies under development that would enable this.

The takeaway then is that there’s c.17 years of natural graphite reserves in the world. While this is better than for other battery raw materials, it’s still not a great outcome.

Flawed approach

In my view, while interesting to illustrate how much material is available, the reserve life analysis is somewhat flawed. The major issue as I see it is the near-term differential between demand and supply which, as I noted for lithium and could note for most of the battery raw materials, is going to be significant.

BMR is forecasting major supply/demand imbalances in all battery materials in the second half of this decade. Even in the first half, our EV forecasts can’t go any higher because of a lack of material availability. The scarcity factor means that prices will go high and be volatile. That’s not a good result for industry participants.

In contrast to what the Nature article suggests, this is not a short-term phenomenon. In our view it could be 7-9 years before supply truly catches up. The question is, can EV producers continue to absorb what I believe will be higher battery costs, while continuing to lower the prices of EVs? Absent that, it’s hard to see EVs hitting the mass market. But if it does happen, EV makers are likely be losing money for an extended period…

We need to get building and exploring

If we’re going to develop an entire Energy Transition, which could last of the order of 40-50 years, off the use of these materials then we really need to get cracking in finding some more! Recycling will be important eventually, but over the next 20-30 years we have to put enough primary material into the industry to have sufficient amounts for recycling.

Less than 20 years of extractable reserves is not something that you can hang an entire Energy Transition on.

Given that Auto companies have allocated over US$100bn to electric vehicles over the past three years, it seems amazing that nobody has thought to allocate enough capital to the upstream end of the business to make sure that there will be enough material for the conversion to Electric Vehicles.

Capital spending in battery raw materials has increased markedly over the past nine months, but is still in no way enough to fend off a raw material shortage which could last for an extended period. It’s not just construction-ready projects we need to finance; it’s also the next stage of projects. We’re just not raising money fast enough.

And that is extremely concerning.

Matt Fernley is Editor of Battery Materials Review and Head of Research for Westbeck Capital’s Volta Energy Transition Fund.


Anode Batteries BNEF Cathode Cobalt EVs Graphite Lithium Manganese Nickel Reserves