GRAFTECH INTERNATIONAL LTD (EAF) stock intrinsic value fair value estimate & owner earnings analysis

GrafTech makes graphite electrodes, the consumable rods that conduct electricity into electric-arc furnaces that melt scrap steel. It runs three electrode plants in Monterrey (Mexico), Calais (France) and Pamplona (Spain) with about 178,000 metric tons of capacity, and it owns the Seadrift, Texas plant that makes petroleum needle coke, the critical raw material most competitors must buy. Sales in 2025 were $504 million: about 54% in the Americas, 39% in Europe, the Middle East and Africa, and 6% in Asia Pacific. This business swings with electrode prices. After a needle-coke shortage in 2017-2018, GrafTech signed three-to-five-year contracts at roughly $9,700-$10,000 per ton. That produced more than $900 million of cash in 2018 and $350-450 million a year through 2022. Those contracts have run out. The company now sells at spot prices: its average price was about $4,100 per ton in 2025, down 13% from 2024. Cheap capacity in China and India keeps pushing prices down. Management itself calls industry pricing "unsustainably low" and says competitors cut prices even harder in late 2025. Sales volume rose 6% to 109,000 tons and the plants ran at 63% of capacity, but owner earnings have been negative for three straight years: about -$3.48 a share in 2023, -$4.68 in 2024 and -$4.83 in 2025. Operating cash flow was another -$84 million in the first half of 2026. How we measure earnings: we start with operating cash flow, take out working-capital swings and stock pay, and subtract capital spending. We treat all capital spending as upkeep because capacity has not grown in ten years. A 2023 goodwill write-off of $171 million is not cash and is ignored. Before interest, the business turned out about -$32 million of cash in both 2024 and 2025. That is roughly break-even at the operating level, and the roughly $90 million a year of interest is what turns it into a large loss. The debt decides the outcome for shareholders. At the end of 2025 GrafTech owed $1.125 billion: a $175 million first-lien term loan, $498 million of 4.625% second-lien notes, $446 million of 9.875% second-lien notes, and about $6 million of leftover 2028 notes. Almost all of it comes due in December 2029. In the first half of 2026 it drew the $100 million delayed-draw term loan to fund losses, which pushed debt to about $1.23 billion against $145 million of cash. The term loans cost about 9.9% and the 9.875% notes cost well above what long-term money costs. Even after crediting the cheap 4.625% notes, the debt is worth more to its holders than its face value. And a company this leveraged would almost certainly pay more than today's blended rate to refinance in 2029. Counting that, we treat what shareholders owe ahead of them as about $1.29 billion. Liquidity was $340 million at the end of 2025, and at 2025-2026 burn rates the cushion runs out around the 2029 maturity unless prices recover first. The revolving credit line also has a 4.0x first-lien leverage covenant that limits how much more the company can borrow. Base case: the business is worth about $1.4 billion, owes about $1.29 billion, which leaves about $4 a share. We assume the trough lasts through 2027, with cash before interest of about -$45 million in 2026 and -$20 million in 2027. After that, prices recover partway, and by 2030 the business earns about $100 million a year before interest and holds that level. We think that recovery is supported by three facts. The industry has had price cycles before; the 2015-2016 collapse was followed by a recovery. Today's prices are below what management says competitors can sustain. And since the end of 2023 GrafTech has cut its cash production cost per ton by 31%, so a price of about $5,000-$5,200 per ton, below what it earned on the spot market in 2023, would produce that $100 million. We do not credit market-share gains, battery-anode needle-coke demand, or the 5-10% volume growth management expects for 2026. None of these is contracted, though 65% of 2026 volume was already in the order book in February. The plants and Seadrift do not wear out on any schedule, so we let that level continue indefinitely. Even so, almost all of the business value is used up by the debt. That is why the base value is so small and so sensitive: every $10 million a year of lasting cash flow is worth about $5 a share. Bear case: the business is worth about $300 million, owes about $1.29 billion, which leaves about -$38 a share. Prices stay near today's levels through the 2029 maturities because Chinese and Indian capacity stays in the market and the U.S. and EU anti-dumping duties the company relies on are cut back at a periodic review, a risk the company flags itself. On top of that comes one steel downturn of the kind seen in 2008-2009 or 2015-2016, when electric-arc steel output and electrode prices fell together. The business settles at about $30 million a year before interest. On that path the company owes far more than it is worth, by roughly $1 billion, and the 2029 debt cannot be refinanced as it stands. Creditors would take most or all of the company in a restructuring. In practice the shares would be worth little or nothing; the negative figure shows how far short the business falls of covering its debt. Bull case: the business is worth about $3.2 billion, owes about $1.19 billion, which leaves about $76 a share. Prices recover to about $6,000 per ton by 2028, the spot level reached in 2022. The shift of volume toward U.S. customers keeps going, utilization rises toward 75-80%, and the lower cost base holds. Cash before interest reaches about $220 million a year by 2029, below 2021-2022 levels because no premium contracts are assumed. The 2029 debt then refinances on ordinary terms without issuing shares. Previous figures: our earlier estimate assumed a large share issue in the base case and did not subtract the debt the way it is done here, so those figures are not directly comparable. What has changed in the business since then: the first-half 2026 results confirmed that the cash burn is continuing, and the company has now used its $100 million delayed-draw loan to fund it. How sure are we of the earnings level? Not very. For four years the business has not earned enough before interest to cover its interest, and the 2018-2022 boom came from contracts that cannot be repeated. The worst plausible outcome is a 2029 restructuring in which creditors take the company, and given the size of the debt the equity would not survive that. What we are most likely to be wrong about: the recovered price (each $500 a ton is roughly $55-60 million of cash a year, or about $25 a share), how long the trough lasts, and the terms of the 2029 refinancing. Nothing is paid out while waiting. There is no dividend, and dilution is a real risk. As things stand, the shares are a call option on an industry recovery arriving before 2029, and price alone does not justify a margin-of-safety price. If cash flow turns positive for two straight quarters and the 2029 debt is pushed out without handing most of the company to lenders, the equity would be worth roughly $12-16 on the base recovery, with real upside if prices return toward 2022 levels. From $7 under those conditions, the base recovery alone implies roughly doubling over an assumed holding period of about four years, and the bull path implies several times that. That is why the margin-of-safety price is $7, applying only once those conditions are met. Items we could not fully reconcile: stock pay was not separately reported before 2019 and was treated as zero for those years. The working-capital build in the first half of 2026 could not be separated from the operating loss in the data available, so we used the full-year 2025 run rate rather than that half-year. GrafTech also owes its former private-equity owner, Brookfield, 85% of certain tax savings under a tax receivable agreement. This only costs money once the company is profitable enough to use its tax losses, and we did not deduct it, which slightly flatters the bull case.