July 25, 2018 - 1:27pm EST by
2018 2019
Price: 18.85 EPS 2.59 3.80
Shares Out. (in M): 302 P/E 7.3 4.9
Market Cap (in $M): 5,700 P/FCF 7.5 5.0
Net Debt (in $M): 2,561 EBIT 1,178 1,672
TEV (in $M): 8,261 TEV/EBIT 7.0 4.9

Sign up for free guest access to view investment idea with a 45 days delay.


We backed up the truck for Graftech’s stock on its IPO day, and it is now a top-five position for our fund.  If all goes well, Graftech’s current price of $19 could put its valuation at 7.3x 2018 earnings, 5x 2019, and 4x 2020.  We do not expect things to go quite that well, because it seems likely that a recession within the next two years will reduce steel demand and therefore reduce electrode prices and profits.  But under any reasonably likely scenario, including a recession, Graftech’s stock looks far too cheap; its revenues and profits will rise from today’s levels even if market prices for its products fall by half.

Our interest was spurred by Puppyeh’s excellent VIC write-up just before the IPO.  This write-up supplements Puppyeh’s and does its best not to repeat that one.  It offers new insights based on our further research and analysis and new information gleaned over the intervening four months, including the company’s 1Q earnings release and call.  (The 2Q release is next week.)  Those insights fall into four areas: (1) the importance of needle coke to Graftech’s value, (2) the solidity of the long-term contracts, (3) Graftech’s opportunity to raise prices and margins significantly in the near future, which we have not seen anyone discuss yet, and (4) the low valuation and our theories (on top of Puppyeh’s) as to why the value opportunity exists.



The key to a Graftech investment is the average investor’s belief that the company’s current earnings power will fall significantly (by half or more) and relatively soon (within a few years).  Without that belief, the stock would not be trading at 1/3 or 1/4 of the S&P 500’s earnings multiple.  We believe earnings will eventually fall but will be “higher for longer” – high enough for long enough that Graftech will probably earn near its entire market cap before a significant fall.  Needle coke is an important part of our reasoning.  Although anyone who looks at Graftech’s IR materials will know the basic needle coke story, we believe management (and Puppyeh) have undersold how important it can be.

To recap some of the basics:  Petroleum needle coke is the key ingredient used to make almost all the carbon graphite electrodes used in electric arc furnace steel mills.  It takes roughly one ton of needle coke to produce one ton of electrodes.  Some electrodes are made from coal-based pitch needle coke, but they have lower performance specifications and get used for other purposes.  Only four companies can make high-quality petroleum needle coke.  Graftech is one of the four and is the only electrode maker who owns needle coke capacity.

Traditionally 98% of needle coke was used to make electrodes.  Recently a second use has emerged: it is the material used to make almost all the anodes in the lithium-ion batteries that power electric vehicles.  You all know in a general sense what is happening to the production volumes of electric vehicles.  They have gone from essentially nothing a few years ago to “small but noticeable” today and will keep growing at very high rates for the next ten years or more.  Needle coke demand from electric vehicles grew from 10,000 tons in 2015 to 60,000 tons in 2017, 9% of total demand, and is still growing at triple-digit rates.  Graftech’s IPO documents included this long-term forecast for needle coke demand:


It predicts electric vehicle demand will grow 10x between 2017 and 2025, which will cause total needle coke demand to double. 

We believe a 10x estimate is too low and potentially far too low.  In 2017, global sales of EVs (both full-electric and plug-in hybrids) were 1.4% of total auto and light-vehicle sales.  If we assume a 3% CAGR for total sales, then 10x growth in EVs implies that 2025 sales will be only 11% of total sales.  Given the flood of EV models being introduced over the next few years and EV’s obvious operational and economic advantages, we would be surprised if EV’s share of the total in 2025 is less than 20% and would not be surprised if it is 50%.  On those numbers, EVs may triple or quadruple the demand for needle coke, rather than merely doubling it.

Doubling is more than enough to help Graftech, because almost no spare needle coke supply exists, and significant amounts probably won’t come online for several years.  As with electrode plants, the planning, permitting, and building of a new needle coke plant will likely take years to complete.  The only new capacity plans we have found are those of the #1 player, Phillips 66, to debottleneck its plants to increase utilization by 15-20% by year-end.  At roughly 50% market share, Phillips 66 will add only 7-10% to effective industry capacity.  Needle coke prices have already sextupled in the last ~18 months, from $500/ton to $3,000/ton.  The likely price path going forward is higher.  Again, industry supply needs to rise by 50% in the next few years and at least 100% in eight years just to keep up with demand.

The needle coke situation helps Graftech in two ways, one direct and one indirect.  First, Graftech produces enough needle coke to supply ~75% of its electrode production.  On this 75%, Graftech has a large cost advantage over competitors who must buy their needle coke.  Market prices for electrodes have risen to capture this added cost and will keep rising if needle coke prices rise further.  The $2,500/ton price increase (so far) for needle coke is pure incremental profit for Graftech when it sells a ton of electrodes.  We would not hazard a guess as to how many new needle coke plants can be built in the next ten years, but given the coming explosion in demand those plants must meet, it seems reasonably possible that this boost to Graftech’s profitability will last a long time.

The indirect boost to Graftech may be even more important.  While needle coke prices have increased $2,500, electrode spot prices have increased roughly $17,500 – from $2,500 in 2016 to roughly $20,000 in 1Q18.  (Graftech management’s most recent estimate of spot prices was $17,000-23,000.)  If $2,500 of that increase comes from needle coke pricing, the remaining $15,000 must come from improvement in electrode industry supply/demand, independent of needle coke.  Obviously these super-cycle profits will not last; the relevant question is how long they will last.  Graftech is in the midst of capital improvement projects to debottleneck its existing plants and increase capacity by 21%.  At least one other producer has said it is doing some debottlenecking, but we doubt the entire industry can achieve 20% gains.  10% seems more likely, and some increase is necessary just to keep up with demand growth (roughly 3% per year from a combination of steel volume growth plus mix shift from blast furnaces to electric arc furnaces).  Graftech has the industry’s only idle capacity, the St. Mary’s plant, which it expects to restart in early 2019.  That will increase Graftech’s capacity by a further 11% and the industry’s by 2-3%.

The only potential source of major capacity expansion is new greenfield electrode plants.  Graftech management maintains that it would take at least five years for someone to plan, permit, and build a new plant.  From everything we have seen, this prediction is credible.  (The message board thread for Puppyeh’s write-up has some good discussion of this issue.)

Needle coke supply’s extreme tightness makes management’s prediction more credible.  There is no point in building a new electrode plant if you can’t procure any needle coke for it.  That insight was by far our largest takeaway from Graftech’s 1Q earnings call.  Although management predicts that they will restart St. Mary’s in early 2019, they would not commit to a schedule or, indeed, to a restart at all until they negotiate and sign the required contracts with needle coke suppliers.  Those negotiations are underway.  On the call, someone asked, "at the right price is there enough needle coke supply to expand St. Marys?  It's not an issue of supply, but it's an issue of price.  Is that fair?"  The CEO responded (emphasis added):

No. I don't think it's quite that simple. I'm not sure that -- it's difficult to be sure that at any price, you could get necessarily what you needed unless you have the relationship with the third parties. Look, there was -- everybody looks at their customer base and wants to be sure that they're doing deals with the folks that have good stability, longevity and are best-in-class. And we think we bring those attributes as a customer to the third-party needle coke suppliers, and as such, think that, that places us in a position where they're more inclined to do business with us than not.

To recap, for investing purposes, Graftech is almost as much a needle coke producer as it is an electrode producer, even though it sells no needle coke, and it will likely keep gushing profits thanks in part to electric vehicles, even though it sells nothing to electric vehicle makers.  Because it is in part a technology investment disguised as a heavy industry investment, we put on our technology-investor hats to worry about a risk that usually doesn’t apply to steelmaking: technological obsolescence.  What if someone develops a new battery that replaces lithium ion batteries in electric vehicles, or what if someone develops a better anode for lithium ion batteries made of something other than needle coke?  Such a shift would cause rapid declines in needle coke prices, electrode prices, and Graftech’s profits.  Thanks to a personal connection with a research scientist at one of America’s major laboratories who is solely focused on next-generation battery technology, as well as our own desk research, we were able to conclude with a reasonably high level of confidence that this risk is a minimum of five years away and is more likely ten years away.  On the new-battery side, several battery producers have already made announcements of expected next-generation batteries.  Even those announcements suggest production would be at least five years away, and our researcher friend believes those timing predictions are optimistic.  On the anode side, there is nothing new on the horizon, and no reason to expect a substitute will emerge.  The anodes are not a pain point for existing lithium-ion batteries in either their performance or their cost (which is a tiny piece of the total even with skyrocketing needle coke prices).



In late 2017, Graftech signed over 100 long-term fixed-price contracts with customers that lock up 77% of its capacity in 2018, 66% in 2019, 59% in 2020, 51% in 2021, and 49% in 2022 (taking into account the coming capacity additions from debottlenecking and the St. Mary’s restart.)  The average price is $9,700/ton, less than half today’s spot price, yet Graftech is making 70% gross margin on them.  The pricing has built-in inflation escalators.  Graftech is protected on the cost side because all of the contracts’ needle coke needs can be met by its own supply, and Graftech has fully hedged the price of all the petroleum byproducts needed to make the needle coke.  These contracts provide significant protection to ~60% of Graftech’s profits over the next five years.  We expect the company to sign further contracts covering the out-years as the existing contracts get fulfilled, which will extend the profit certainty.

Some investors have expressed concern that these types of contracts in cyclical industries do not survive bad industry downturns; the customers demand that the suppliers tear them up and start over, on the threat that either the customer goes bankrupt or the customer gets mad and will never do business with the supplier again (so the threat of a lawsuit is weak).  We know of such outcomes in other similar industries in the past.  However, we believe Graftech is better protected than some other suppliers were, for several reasons.

First, the contracts, along with being take-or-pay, include (in the words of Graftech’s S-1) “significant termination payments (typically, 50% to 70% of remaining contracted revenue) and, in certain cases, parent guarantees and collateral arrangements to manage our customer credit risk.”  From a practical standpoint, these protections are not at all iron-clad, but they are much better to have than not to have.

The second protection is the huge spread between the contract prices and current spot prices.  Spot prices would need to fall over 50% just to get back to parity and to fall materially more before a customer would begin to think about breaking the contract.  Given where prices have come from – the long-term average electrode price before the steel apocalypse of 2015-2016 was $4,500/ton – it is certainly possible that prices fall that far again, but it does not look likely in the next five years unless there is a bad recession.  In addition, if prices did fall that far, customers would have enjoyed (probably) years of paying $10,000/ton below spot prices.  They would be on shaky fairness/ethical ground to complain about paying $2,000 or even $5,000 above the spot market for the remainder of the contract; they would still come out ahead overall.  Practically speaking, that is a very different set-up than initially paying, say, $2,000 below spot and later paying $10,000 above spot.  The cost savings vs. spot that will be banked by customers prior to any spot price falls would carry a great deal of weight in any attempted renegotiations with Graftech or in front of a judge or jury during a breach of contract lawsuit.

The final protection is the electrodes’ small share of total steel production costs – less than 5% even after the recent price spike.  Dramatic swings in electrode prices don’t much change a steel mill’s total costs.  Therefore the mills care less about the pricing and care more about the quality and reliability of supply, they are not much disadvantaged versus competitors if they are paying more for electrodes, and a fixed-price electrode contract is not the commitment that’s going to drive anyone into bankruptcy.



Assuming no industry downturn, Graftech has the opportunity to almost double its earnings power over the next 18 months, relative to the already-strong profits it reported in 1Q.  The earnings increase will come not only from capacity additions, which management has made perfectly clear, but also from pricing increases, which management has not publicized.

The capacity increases are simple and have been touched on above.  Graftech’s debottlenecking products, scheduled to be completed by year-end, will increase capacity by 21%, from 172,000 tons to 208,000.  The St. Mary’s restart, planned for early 2019, will add another 11%, to 230,000 tons.  The total increase is 34%.

The less obvious profit growth is from price increases.  Graftech has three chunks of electrode capacity that have very different financial profiles:

  1. The long-term contracts, supplied with Graftech’s own needle coke.  These contracts provide known revenues and, given Graftech’s control over its input and production costs, predictable margins (70% gross margin). 
  2. Short-term contracts supplied with Graftech’s own needle coke.  Graftech can produce these electrodes just as cheaply as the long-term chunk but sell them at prevailing prices, which are currently much higher.
  3. Short term contracts for which Graftech must buy needle coke from third parties.  These electrodes can be priced high but also have a higher COGS and lower profit margin, because of the higher needle coke cost.

The relative sizes of these chunks will change over time as Graftech’s total capacity and amount committed to the long-term contracts change:

In 1Q, Graftech reported an average selling price of $10,124 per ton.  The long-term contract pricing averages $9,700 per ton.  From that we can infer that the short-term contract volume also sold for around $10,000 per ton, even though management was citing spot market pricing of $17,000-23,000 per ton as of 1Q.  We could guess why this discrepancy existed, and when we asked management, they confirmed it:  Although those contracts are shorter-term, they were negotiated in early 2017, before most of the increase in spot pricing occurred.  We asked whether that pricing could increase significantly once the old contracts roll off and new ones are signed, and they said yes.  Theoretically, then, to get to current spot pricing, 34% of Graftech’s 2019 sales volume could see prices double, which would grow gross profit on that piece by roughly 150%.

Adding together the volume increase and pricing increase, we could see earnings per share rise from $2.60 in 2018 to $4.70 in 2020, versus a consensus estimate of $3.62 (30% higher).  We do not expect earnings to get this high because we expect a recession to start by then, but we don’t need that level of earnings for the stock to be a bargain.  Because all of Graftech’s capacity is being sold near $10,000/ton, its realized prices would not fall even if spot market prices fall 50%, and its revenues and profits would still rise 34% thanks to the capacity increases.



At the time of Puppyeh’s write-up, Graftech’s investment bankers had set the initial expected IPO range at $21-24 per share.  Bankers usually low-ball the expected range, raise the range prior to the IPO, issue the stock at or slightly above the top of the raised range, and still leave room for the stock to trade higher on its first day (so that the IPO is “successful” and its investor-clients are happy).  We assumed the range would be raised to, say, $23-25, the IPO would price at $26, and the stock would begin trading at around $28.  Instead, the bankers lowered the range to $18-20, the IPO priced at $15, and the stock traded in the $14s throughout its first day.  At today’s price of $19, the stock still has almost 50% upside to get to what “should” have been the immediate post-IPO price.  If things go well, the current price puts the valuation as low as 4x 2020 earnings.  Given Graftech’s long-term contract protection, it is difficult to see how a cyclical downturn could lower earnings enough to move that multiple anywhere near the market average.

Why was the IPO so busted, and why is the stock still so cheap?  Puppyeh’s write-up has an excellent section addressing this topic, and we agree with everything it says.  We can add three other possible factors.

-The price increase opportunity is not widely known.

-It’s a stealth technology stock.  The importance of electric vehicle demand is probably underappreciated (it directly helps prices and margins and constrains potential new electrode supply), and the potential growth of electric vehicle demand is also probably underappreciated.  Most investors looking at Graftech were either heavy-industry specialists or generalists who could not integrate the electric vehicle piece of the analysis comfortably and quickly. 

-This type of stock is deeply out of favor.  We spent several years with our long book heavily weighted towards high-growth, high-multiple technology stocks.  (Well over half our total long exposure and all of our top five positions.)  Lately the market has been rewarding and valuing that kind of stock more and more, at the expense of low-multiple stocks for which there is any sort of fear about future earnings headwinds (either cyclical or secular).  Given our fondness for great technology businesses and our history of owning them, we should have some credibility in saying: the valuation gap between these two types of stocks has reached historic and ridiculous extremes, the highest since the 1999-2000 internet bubble.  The misvaluation applies to both sides, with high-growth stocks too expensive and “feared” stocks far too cheap.  We may be early, but we believe now is a good time to sell the high-multiple ones and buy low-multiple ones, and we have been doing so, in size, for the last several months.  For purposes of buying Graftech, one need not believe that high-growth stocks are overvalued; it is enough to conclude that stocks like Graftech are dirt cheap even assuming that the investor fears are directionally correct and that earnings will eventually decline.  We are willing to wait for these companies to generate profits for us the old-fashioned Buffett way, by generating cash, without needing a valuation multiple re-rate.  It will be a bonus if the market decides to rotate out of momentum and into deep-value, but that rotation seems inevitable, even though we can’t predict when and even though it might be years away.

I do not hold a position with the issuer such as employment, directorship, or consultancy.
I and/or others I advise hold a material investment in the issuer's securities.


  • Earnings grow and exceed forecasts due to capacity and pricing increases
  • Strong cash flow used for buybacks and special dividends
  • Cyclical downturn is later or shallower than feared
  • Graftech earnings remain robust despite an industry downturn
  • Valuation multiple rerating from a marketwide rotation out of high-growth / high-multiple stocks and into low-multiple / "fear" stocks
    show   sort by    
      Back to top