|Shares Out. (in M):||302||P/E||8.7||6.3|
|Market Cap (in $M):||6,800||P/FCF||8.6||6.3|
|Net Debt (in $M):||1,715||EBIT||1,095||1,423|
Thesis summary: Writing up an IPO on VIC may seem a little unusual and perhaps I am temping fate, but Graftech (EAF will be the new code) seems such a compelling proposition that I am going to go ahead and do it anyway. If you purchase EAF at the mid-point of the IPO range (I believe the book closes next week so unsure on final pricing), I believe you are buying the only vertically-integrated producer of graphite electrodes (GE) at ~5x EV/EBITDA, 6x FCF and 6x P/E on my 2019E numbers - with very little if any upside baked in for higher spot prices. This is for a business where 2/3 of volumes have been pre-sold (for five years) at levels that imply ~65%+ EBITDA margins (and near 40% returns on invested capital); where you are by far the lowest cost producer in an industry operating at 100% utilization already, with both significant barriers to entry and a strong secular growth outlook; where you are the only name in the industry capable of meaningfully growing production capacity (due to raw material shortages); and where the selling shareholder has already set the tone re capital returns (ie, a lot will be forthcoming). I feel the 'right' price for a business like this is at least 10x FCF as a starter and see no reason why it will remain on pure commodity-cyclical valuations (where it is being sold in the IPO) given its new business profile. This re-rating alone would suggest 50% upside in a year, outside of excess capital returns. Keep in mind that other GE names of much less quality in India and China, trade at high single digit EBITDA multiples; while the Japanese comps - with 30pt lower EBITDA margins and much less visibility - trade at 5x EBITDA; meanwhile other specialty carbon names (which EAF could easily claim to be, over time), tend to trade on double digit EBITDA multiples. This opportunity exists because a) Brookfield has already made 10x on this investment; b) Brookfield is only selling 15% of the company (and so is motivated to leave some money on the table now for future offerings); and c) there is a historical taint on the industry given the deep cyclicality of the business in the past (which I think no longer really applies).
Graftech will be listing with a ~$7bn market cap and around $900mm in free float so should be liquid enough for most all funds. This will not be a long write-up as the case is pretty clear. There is an online roadshow available (through JPM/Citi/other bookrunners) with a lot of useful information. Graftech has been written up three times in the past, (two long, one short) but all three writeups were more idiosyncratic and before the recent restructuring of the industry, so are of limited value to the thesis today.
Quick background: Graftech makes graphite electrodes (GE). GE are used to produce steel in an electric arc furnace, the dominant method of steel production in the US (but not globally, where blast furnaces are still dominant). Graphite is used because it is the only material that has the chemical properties to adequately pass a current consistently at the temperatures needed to produce steel in an EAF (around 1800C) - hence there is no substitute for this product. Ultra-high performance (UHP) electrodes - Graftech's specialty - can take up to 6months to produce and yet are consumed in one single 8-10hour shift. Despite the recent massive rally in prices (GE prices went from $3k/t in 2016 to $10k/t now, at the contract level, while spot pricing in some markets ins $15-30k/t), GE still constitute a small portion of the value of the steel produced. For example, even at $10k per ton, the cost of GE needed to produce 1t of HRC steel from an EAF is around 2% of the value of the product. This is a very important consideration as it clearly means that steel producers value quality of product, consistency of supply, and performance of electrodes far more than the price paid.
For most of the past 10yrs GE was a typical volatile commodity product, beset by vicious pricing swings. This changed rapidly in the last 18mos or so due to a confluence of factors:
- tightening Chinese steel markets increasing global steelmaking utilization
- increasing environmental standards making new-build GE capacity in China very difficult
- increasing EAF (vs BOF) steelmaking demand
- removal of 20% of high-cost capacity through rationalization (including some done by Graftech)
- consolidation in the industry (no 2 and no 3 players merged)
- and, most importantly, the development of an alternate use-case for the key input cost into GE, needle coke (it is used as an anode material in LiB batteries)
This last point is the most important, because it essentially has created a new secondary market for needle coke - the only input for UHP GE. This end market, LiB batteries, is growing (in demand terms) by 30% (according to some analysts) and already constitutes 10% of the total demand for needle coke. Today, the GE market is operating at 100% utilization, but there is no way to source additional GE because it can only be made from needle coke (for high quality GE), which is itself at full utilization. New capacity takes a long time to bring online because needle coke is essentially a by-product of the refining process (made from a very small portion of the output slate), so other than de-bottlenecking, it is quite hard to bring new, dedicated capacity online. Graftech estimates it will take at least 5yrs to see meaningful needle coke capacity emerge outside of China, and perhaps longer than that to see real GE capacity emerge as well.
In the meantime, Graftech is the only vertically integrated GE producer globally because it owns Seadrift Coke - a captive source of 140kt of needle coke per year (versus 170kt annual GE production, rising to 200-230kt over next two years). There are only 5 global facilities that can produce the quality of needle coke at scale needed for UHP, and Graftech owns one of them (while >50% of the rest of ex-China capacity is owned by Philips 66). This is why spot prices for GE have risen so far, so fast - it remains a small component of the cost of production for steelmakers, it is completely essential, and it cannot be easily sourced in the current environment.
Cost push has been the primary driver of GE price expansion in the last year or so (even though GE prices have expanded far more) but the key point is because Graftech is vertically integrated on ~75% of its production, it is the sustainable low-cost producer. In 1Q'18, Graftech says their vertically-integrated total cash costs per ton ($2600/t) is lower than the cost of needle coke alone in the spot market (around $3000/t). Essentially then Graftech has at least ~$3500/t advantage versus marginal producers, hugely insulating it from the next downturn (whenever it comes, if it comes). Thus, for example, even at $7k/t - assuming needle coke prices didnt come down as well - Graftech would be making mid-40s% EBITDA margins (ie 5-10pts higher than comps today). Still, it is unlikely prices fall below current contract price levels (ie $10k/t) when all clients are happy to buy as much GE (five years out) at $10k/t if they could secure the supply.
The Changing Nature of the Industry: nothing speaks more to the 'new normal' in GE than Graftech's new pricing/contract structure. This industry used to be subject to huge annual or semi-annual earnings changes as contracts were negotiated with steelmakers on this basis; with the industry plagued by over-capacity for a long time, this meant there was very little visibility to earnings and cashflows out more than say one year (similar to many pure commodity industries). However Graftech, a few months ago, sold forward for five years (via take-or-pay contracts) around 2/3 of their production at around $10k/t - essentially locking in 60%+ EBITDA margins, I think - to willing buyers. To me, this underlines the much improved industry structure: steelmakers would simply be unwilling to commit in this way if they thought the current level of pricing was either unfair or unsustainable. I am not sure the market really understands how significant this was/is, but to me it clearly suggests a much healthier, more viable industry (and thus deserving of a non-commodity multiple) once the market gets its head around the consequences.
Valuation: As mentioned in the intro, I think on next year's numbers, Graftech is coming around 5x EV/EBITDA, 6x FCF and 6x P/E (with only 1.5x of leverage at end-this year, falling to 0.5x end next year even assuming they pay a 5% yield). That, to me, sounds very much like a discounted, top-of-cycle commodity multiple for a business where the market expects cash flows to fall significantly in the out-years. However, that won't happen for at least 5yrs, given the hedging profile of the company (they have hedged production costs on the integrated volumes, too), and instead I see a business where the end-product is likely to stay tight for the foreseeable future; where new capacity is not likely to affect delivered volumes for at least 5yrs; where you can own the low-cost producer with the only opportunity to organically grow in the industry; and, significantly, where you are completely aligned with majority owners (Brookfield) who clearly want to upstream cash (hence the large divs paid out pre-IPO and commitment towards a target leverage ratio, 1.5x, that implies huge shareholder returns).
Looking at a variety of steel/other commodity names, pure price-taker commodity businesses like steel cos trade at 4-5x EBITDA currently, while better (but still commodity) businesses with low visibility like BHP/RIO/etc, trade at 5-6x EBITDA and around 9-10x FCF. This is where (on a FCF basis) much lower-quality Japanese/Chinese/Indian graphite names trade as well - none of which are vertically integrated, have hedged 5yrs of sales, are listed in the US, have this size market cap and likely analyst coverage, nor commitment to large-scale capital returns. I would argue a similar multiple on FCF is justified at a bare minimum - and I am not even pricing in much upside for where prices currently trade (say, $20k/t), as I am essentially assuming just $10.8k/t average pricing next year (and $10.5k/t the year after). Even so, 10x FCF gets me to $36/share on FY19E numbers, excluding any excess capital returns, ie a likely 50% IRR from here. At that level I think the stock would be at 8x EBITDA going on 7x 2020E, still reasonable in my mind.
It is also worth noting that when Graftech was previously listed, as bad as this industry was historically, the average EV/EBITDA multiple was 10-12x over the last 8yrs (and the business was much more levered). So there could really be much more upside, even if it takes more time to get there.
Why is it so cheap? Ok, the obligatory 'what am I missing' section. There are a few potential reasons, I think most make sense and I'm comfortable with the thought process, but please feel free to disagree in the comments:
- Brookfield is a price insensitive seller: Brookfield bought this for $800mm in 2015, and, while they may have invested some more capital during the early dark days, they are effectively clearing nearly a 10x on this ($7bn equity valuation + $2bn and change in divs). They don't need to sell this at full value right now, especially because...
- Brookfield is only selling 15%: I think they want the deal to go well as they have a lot of stock to sell later; at the same time I think they recognize this is too cheap to sell all their position so they are motivated to see it trade better as well. In the meantime they are highly motivated to keep paying huge special divs...which helps us as minorities
- The taint of history: pretty clear really, there will be a ton of non-believers who think this is the same old crappy industry (cf, MU and the memory industry), who aren't willing to pay for the structural improvement or Graftech's new-found contractual visibility. OK, i get it, but I am also comfortable thinking this will come over time, especially as shareholder returns accumulate
- No comps: this is a pretty niche industry and there aren't any US-listed comps, while the global comps are few (a couple of Japanese, a couple Indian) and not the simplest (they have other ancillary businesses). I can see a number of managers throwing it in the 'China risk' bucket and approach later on once it has listed/gets coverage
I think that is really about it. I think this is about as good as a mid-cap IPO can get, happy to talk more about the risks in the comments.
They print massive quarterly FCF and EBITDA margins and return huge gobs of capital via dividends/buybacks