UPM: A Refinery That Slipped, a Paper Mill That Bleeds, and a Land Bank That Just Became an AI Asset
UPM: A Refinery That Slipped, a Paper Mill That Bleeds, and a Land Bank That Just Became an AI Asset
UPM's third quarter was not a disaster, but it was a quarter where the transformation story and the operating reality pulled in opposite directions. Sales came in at EUR 2.202 billion, 1% below last year; comparable EBIT fell 8% to EUR 131 million, a 5.9% margin. Sequentially the bridge was ugly and simple: variable costs rose by more than EUR 100 million and delivery volumes fell. Prices rose in most businesses but covered only part of the cost increase — and in Communication Papers they did not cover it at all.
What saved the print was cash. Operating cash flow of EUR 420 million in the quarter, net debt down EUR 320 million sequentially, leverage at 2.09x. Nine-month comparable EBIT was still up 10% to EUR 601 million. So the year is not broken — but the direction of travel inside the quarter was, and management's own language shifted with it: “the upper end of the guidance range at the moment looks less likely” — Tapio Korpeinen, Chief Financial Officer · 2026-10-28.
That phrase is worth flagging, because it is a rerun. In April 2025 the CFO used almost identical hedging: “arriving at the upper part of the range is more challenging as the world looks like right now” — Tapio Korpeinen, CFO · 2025-04-24. When the same soft-downgrade formula recurs, it stops reading as news and starts reading as a habit.
The Leuna slip is the real news
The one hard, unhedged change this quarter was at the Leuna biochemicals refinery. Two of the three core processes are running commercially; industrial sugars and lignin-derived chemicals are at meaningful volumes. But the third — sugars-to-glycols — has stumbled. The CEO's description was carefully underwritten: “we have identified small technological challenges that require some modifications” — Massimo Reynaudo, Chief Executive Officer · 2026-10-28. Integrated production now lands in Q2 2027, roughly six months late, with nominal capacity — and therefore the renewable functional fillers and glycols business case — only in early 2028, and EBIT-positive operations gated behind that.
Contrast that with July, when the same project was described as “more or less in line with our schedule... that points toward reaching breakeven in 2027” — Massimo Reynaudo, CEO · 2026-07-23. The 2027 EBIT drag is now guided to be broadly similar to 2026 rather than improving. A six-month slip on a first-of-a-kind plant is forgivable engineering. Presenting it as a detail, on the same call that the CFO hands off the energy business ahead of his departure, is the kind of thing investors remember. Notably, management says confirmed contracts and the qualification pipeline exceed multiple times annual capacity — the demand side of Leuna is not the problem.
Communication Papers: 30% overcapacity and no closures
The structural wound is old and getting older. Communication Papers posted an EBIT loss of EUR 8 million; European demand fell 7% year-on-year; input costs rose sharply; and the industry's answer was nothing. As the CEO put it, “there have been no meaningful capacity closures in Europe in 2026” — Massimo Reynaudo, Chief Executive Officer · 2026-10-28 in a market with roughly 30% overcapacity. The planned Sappi joint venture is the fix — and the fix just got harder. In July the regulatory process was “proceeding as planned” — Massimo Reynaudo, CEO · 2026-07-23 with a final resolution expected by year-end. In October, the European Commission restated serious concerns, and UPM has now opened discussions on remedies with a decision expected at year-end or shortly after.
In early October, the European Commission restated its serious concerns about the joint venture. We believe in the strength of our arguments, and we have been addressing the concerns raised. At the same time, we have opened discussions on potential remedies.
Remedies talk is not failure, but it is a different conversation from the one held three months ago — and any divestment remedy in graphic paper lands in a market nobody wants to buy.
Which makes the Q4 bridge worth reading closely: roughly EUR 60 million of energy refunds booked in December, about EUR 50 million of forest revaluation, against roughly EUR 40 million more maintenance. That is a net positive swing of perhaps EUR 70 million driven by regulatory timing and asset revaluation — not by paper volumes. The operating business does not fix itself in Q4; the calendar does.
The accidental AI trade
The genuinely new idea on this call — and the reason it is worth reading at all — is what UPM has quietly discovered it owns. Finland's electricity demand is inflecting, and UPM is the second-largest generator in the country with about 12 TWh of CO2-free output, the second-largest landowner, and roughly 1 GW of data center ready land that can be powered in under two years, with another 3 GW available over three years, plus a ~1 GW wind/solar/battery pipeline ready to build and long term PPAs and balancing services to sell alongside.
The strategic logic is sharper than the average "we have land" pitch, and the CFO framed the scarcity precisely: “what will become scarce is electricity supply that is not weather-dependent” — Tapio Korpeinen, Chief Financial Officer · 2026-10-28. That is the correct product in an AI buildout — firm, dispatchable, carbon-free power plus permitting and land, sold to counterparties investing billions. It also ties UPM into the global AI data center complex, a theme that was a monster 360-day winner across dozens of names before fading over the last 90 days — the tape has already voted on the demand side and is now debating the digestion. UPM sells the scarce input rather than the compute, which is the more defensible seat.
The offsetting ingredient is cost. increased variable cost was a top-three keyword this quarter, sitting alongside global High fuel prices pressure — UPM's energy leverage is not one-directional. Cheap power is an input to the mills and a product for the data centers; the two hedge each other imperfectly, and in Q3 the input side won.
What management stopped emphasising
Adhesive Materials got the most airtime, and the tonal shift matters: European label shipments fell 20% quarter-on-quarter and 5% year-on-year, yet unit margins held and a new India factory was announced. The explanation is destocking — described again as temporary, and partly a reversal of Q2 inventory building around the Middle East conflict. This is a company trope, not a discovery; the same word spiked in the company's keyword history back in 2023. Fibers returned to positive EBIT with 8 weeks of curtailment at Kaukas and 5 more at Pietarsaari to come, protecting margin at the cost of volume. And WISA Group, the plywood spin, lists on 2 November.
Leverage, meanwhile, is a promise in its third year of repetition. In July 2025 the CFO said the aim was to get back below 2x — “the aim is to come back then down to that level of leverage” — Tapio Juhani Korpeinen, CFO · 2025-07-24 — and here we are at 2.09x, with the same sentence and the same target. The cash flow is doing its part. The portfolio is not yet doing its part.
Net: a mid-cycle industrial with one genuinely scarce asset, one badly-timed regulatory negotiation, and one engineering overrun — priced as a paper company, but increasingly not one.