In recent discussions, blogs and issues of Battery Materials Review (such as here, here and here) I’ve highlighted my view that battery materials in general and lithium in particular are going into a supercycle. I’ve stated that because my forecasts show a substantial supply/demand gap emerging.
While a growing number of analysts are positive on the outlook for battery materials, many (particularly those at bulge bracket investment banks) are more conservative. I suggest that it’s a lack of understanding of what is a small and increasingly specialised sector that is causing this issue.
In a recent Blog post (Why most analysts are understating battery raw materials demand, 2 June 2021) I discussed why I believe that many analysts are understating the demand side of the supply/demand equation.
Now I’m going to focus on the supply side and discuss why I believe that many analysts are getting it wrong here as well.
One investment bank (Morgan Stanley) has been more outspoken than most on its views on the lithium market over the past few years. While MS was right to be cautious in 2018, I do believe that they’ve got the situation very wrong in 2020-21 and it’s because, in my view, MS is still treating the lithium market as a commodity market when, in fact, lithium carbonate and lithium hydroxide are very much specialty products.
I thought it would be worthwhile recapping on the main reasons why I believe that many investment banks are wrong on lithium. They are:
- Understating EV forecasts
- Understating battery yield losses
- Lack of understanding regarding fungibility of lithium chemicals for cathode makers
- Overstating amount of material produced which is battery grade
I’ve already discussed the first two points in my previous articles and now I want to go on and discuss the supply-side reasons in this article.
Understanding why lithium projects are not fungible
So let’s start with the fungibility of products. This is the essence of why lithium chemicals are specialty, not commodity, products, in my view.
The key issue here is the purity of products required to make batteries. Most industry specialists will tell you that it’s actually not about purity per se, it’s about impurities. Adjacent is a comparison of specifications for battery grade lithium carbonate products.
You can see from the data that there are different purity requirements for the lithium carbonate, between 99.5% and 99.8% and then, within that, a huge level of spread for the impurities, for instance between 150-650ppm in sodium (Na), 30-300ppm in chloride (Cl) and 150-500ppm in sulphate (SO4).
But for some elements, there isn’t much variation. For instance, iron (Fe) is extremely deleterious in battery manufacture, only between 2-10ppm are allowable in each of the four specs. Similarly the metals, such as aluminium (Al), lead (Pb) and zinc (Zn) have very tightly-permitted concentrations. [There’s a reason for this; if the metal content is too high they can form metallic dendrites during charging which will puncture the separator, creating shorts].
The specification of the lithium output will be agreed between the cathode maker and the lithium producer. The producer must go through an extended period of qualification pre-production to ensure that they hit the specified levels of purity and impurities. This can take between 12-18 months, before a cathode maker or battery producer (or their client) will sign off on an offtake agreement.
In the past, metallurgical testing for a commodity project would likely take place on only a few kilograms or hundred kilograms of material. In battery qualification testing, up to six tonnes of material may be needed for testing all in all. Tests will take place on a lab-scale, a pilot-scale and eventually on a demonstration-scale basis to ensure that the developer can produce to the required specification on a consistent basis.
When it comes to cathode manufacture, producers will often trade the specs of one material off with another. For instance, if the nickel they’re using has a high iron level they may require the lithium, cobalt and manganese to have a lower iron level. If the lithium has a high calcium level then they may require a slower calcium level in the other cathode materials. The materials, and the impurities in the materials, then fit in like a jigsaw puzzle; they are genuinely complementary.
As a result of this though, material from one lithium project is in no way directly fungible with material from another. Most cathode producers will use lithium from two to three suppliers to ensure that there are no supply issues if one of those suppliers goes down, but it won’t be easy for them to bring a totally new supplier into the process if they have a different product specification (it could take months or years).
This is something that analysts looking into the sector often get wrong. They mistake lithium and other battery materials for interchangeable commodities. They are most definitely not. They are entirely specialty products.
Overstating the amount of battery grade material produced
The above consideration leads almost directly into this point. Namely the amount of current production and planned output that can actually be classed as battery grade. Not many people outside the industry know this, but a fairly significant proportion of the lithium carbonate currently produced by Latin American brine producers is not suitable for use in batteries.
It varies by producer and by asset, but industry sources say that it could be up to 40% for some companies and assets. SQM is a notable major brine producer that produces a lot of material that is (currently) not battery grade. Orocobre is another. You can tell that those producers are making a fair proportion of sub-battery grade material because their calculated or reported prices are materially below realised prices for Japanese and Korean imports (which are mostly battery-grade tonnage).
Such material that is not battery grade at source either needs to be upgraded, with a commensurate drop in lithium recovery, or be set aside for other uses.
It's not just producers that have this issue either. Development projects from Latin America are also likely to have an issue with non-BG production. In its 25 March 2021 release Galaxy Resources noted that it had achieved “battery grade” purities on pilot runs from its Sal de Vida project in Argentina. It showed that “apart from Ca and Mg, all specifications for battery grade were met.” The work was on a continuous pilot which produced 100kg of material. You can see from the specifications above that that’s not a lot of material for qualification and the company goes on to note that it has done lab-scale tests to lower Ca and Mg contents. Realistically it will need to do larger scale tests to be certain of producing battery grade material.
Neo Lithium is developing the 3Q project in Argentina. In its preliminary feasibility work it suggested that early production from the plant would not be battery grade and it could take years, rather than months, to reach 100% output of battery grade material.
And not producing at battery grade is not just an issue for lithium carbonate producers. In its recent study Piedmont Lithium initially suggested that it could take two years (later reducing that estimate to one year) to optimise its hydroxide plant to produce 100% battery grade material. And indeed, in China, there is an existing problem with optimisation of hydroxide plants; some take at least two years to hit spec.
I don't include these mentions to have a dig at the quality of products being produced or being planned to be produced by these companies, but to highlight the difficulty of getting up to battery grade spec and staying there.
And given my experience in 20 years as an equity analyst that one of the only constants in the universe (beside death and taxes!) is that development management teams will always overpromise and underdeliver, that suggests to me that at least 10% of lithium carbonate from Latin America will not be battery grade at steady state output and probably as much as 50% of production from the first two years of production at Greenfield project developments will not be battery grade.
And that’s just the situation for brine and hard-rock based projects. Who knows how complex it will be to optimise projects based on new technology? The sedimentary-based projects in North America, the geothermal projects in Europe and North America are a key component of some analysts’ supply forecasts. Rio Tinto’s new Jadar project is based on processing a newly-discovered compound. We have no idea what the percentage of battery grade product will be in these new projects and how long it will take to reach planned levels. But, whether these projects succeed or fail, based on recent experience in the industry it’s likely that it will take much longer than we expect for them to hit their production targets and product specifications…
Supply overstated, demand understated
That means that many analysts are overstating their supply forecasts by a minimum of 10% and a maximum of 20%. And that can make a big difference.
It can make an even-bigger difference when we add in that a large amount of analysts are understating their demand forecasts because either their EV sales forecasts are too low or they are not factoring in battery yield effects (as I discussed in more detail in my recent blog).
In a recent research note (What’s On My Mind? Our EV Fcsts May Be Too Low; Elon Farewell to Analyst Calls is a Good Thing, 27 July 2021, Morgan Stanley) Adam Jonas, Morgan Stanley’s US autos lead analyst, noted that his team's EV sales forecast may be too low. He team was (and still is) forecasting 6.1% global EV penetration on average in 2022, yet global EV sales penetration hit 6.5% in June 2021. We’ve written before about how we believe that the shape of the EV penetration curve will be steeper at the beginning than most autos analysts are using, and how this can have substantial impacts on battery materials demand forecasts…
Morgan Stanley wouldn’t be the first to raise it’s EV sales forecasts.
In fact, check out our chart to see the rise of EV sales forecasts for 2025E and 2030E over the past few years.
UBS (with its March 2021 update) is the only bank or organisation that’s above us in its forecasts for EV sales in the medium term. All of the others have been slowly playing catch up and bumping up their forecasts by a few million each time they publish…
Only a matter of time
In my view it will only be a matter of time until consensus forecasts show a supply/demand deficit for lithium and other battery materials in the near term. The problem is that that will likely be too late for the auto makers. Autos is a conservative business and most auto makers are presumably utilising industry analysis done by major investment banks on which to base their capital allocations.
However, as I noted above, UBS is the only major investment bank which has what I would call a viable EV sales forecast. Many other major investment banks don’t really cover battery materials because it’s such a tiny part of the market. So whether their autos or raw materials supply/demand models are a fraction out doesn’t really matter to them (if they even have raw materials supply/demand models).
But it’s going to matter to the autos industry. As I’ve stated in previous blogs (going back over two years now), the autos producers are in a brilliant position to make capital available for the mining industry by nature of the fact that they have the ability to access capital so easily (which mining juniors don’t). But they haven’t.
And, as a result, a substantial supply/demand imbalance has developed. And that’s going to lead to higher raw materials prices. Which is not great for the autos industry and will not be great going forward. We need to see capital flowing into mining now if we’re to head off a long-term supply/demand imbalance. And we’re not yet seeing it. And bulge bracket analysts not understanding this industry and having poor assumptions in their models is contributing to the problem.
Matt Fernley is Head of Westbeck Capital's Volta Energy Transition Fund and Editor of Battery Materials Review.
