Vossloh sees 2026 sales up to €1.66 billion, EBITDA margin to 14.5%

Vossloh reported FY2025 revenue of €1.34bn (+11% y/y) which slightly missed the €1.35bn analyst consensus; EBITDA was €179.4m (13.4% margin), EBIT before PPA €119.6m (+13.7%), net income €79.9m and EPS €3.24. Growth was driven by the first-time consolidation of Sateba and major infrastructure projects in Algeria and China, with strong performance in Core Components, Customized Modules and Lifecycle Solutions. Guidance for 2026 is upbeat: sales €1.56–1.66bn, EBITDA €215–230m (13.5–14.5% margin) and EBIT €118.5–131.0m, implying meaningful year-over-year expansion on the top and EBITDA lines.

Analysis

The most durable read-through is thematic: continued large-scale GPU demand (AI/EV compute) amplifies pricing power upstream (GPU makers, wafer/substrate, and server OEMs) while creating a two-speed market for ODMs—those with validated, local manufacturing footprints and direct OEM relationships will capture share and margin, smaller niche suppliers will be squeezed on price and lead times. For rail/infrastructure, M&A-driven modularization is a structural advantage because it converts bespoke project revenue into higher-margin, repeatable product sales; however, local-content rules and JV partners in China/Algeria shift competitive dynamics toward players with on-the-ground production and IP transfer capabilities.

Primary tail-risks differ by horizon: over quarters, AI capex inventory adjustments can erase demand growth within 1-3 quarters; over 6-24 months, emerging-market project execution (permitting, FX, receivables) and input-cost swings (steel, freight) are the biggest margin killers. Integration execution—both for a rail-group absorbing a bolt-on and a server OEM integrating new GPU platforms—can produce lumpy earnings and create windows for activist/arb activity if targets miss milestones.

Near-term catalysts to monitor are order flow and inventory metrics from large OEMs (Tesla/SpaceX supply channels, server OEM book-to-bill) in the next 1-2 quarters, and project milestone confirmations (contract signatures, equipment deliveries) in Algeria/China over the next 6-18 months. Watch commodity price moves and regional currency volatility as high-probability triggers that can quickly reprice margins and force knee-jerk guidance revisions.

The consensus risk is two-sided: investors may be underweight EM execution risk in infrastructure names while over-extrapolating linearity in AI GPU demand. That argues for asymmetric exposures (optioned upside into compute demand; hedged, conviction-sized equity exposure into infrastructure names with downside protection).

Key Decisions for Investors

• Long NVDA (tactical, 3–6 months): buy a 3–6 month out-of-the-money call spread sized to 1–2% NAV with strike width calibrated to capture a 15–30% move (entry on next meaningful pullback). Rationale: captures continued OEM compute demand with defined downside (premium paid) while benefiting from order-flow upside; stop-loss = premium*1x, take-profit = 2.5–3x premium.
• Long Vossloh (ETR:VOS, strategic, 12–18 months): initiate a 2–3% NAV long with a protective 12-month put ~25% OTM (or buy-to-protect downside). Rationale: M&A-driven margin levers and backlog conversion offer re-rating potential; hedge costs limit downside if EM projects or integration disappoint.
• Long SMCI (tactical, 3–6 months): buy small-sized call options (~0.5–1% NAV equivalent) to play server OEM re-rating from sustained GPU demand, size as a levered equity proxy rather than outright equity due to execution volatility. Risk: sensitive to near-term inventory swings—keep time decay within 6 months.
• Cross-asset pair (6–12 months): long NVDA / short a Euro industrials ETF (or broad EM infrastructure proxy) sized 1:1 by dollar exposure to isolate AI compute upside from cyclical infrastructure/execution risk. Rationale: isolates secular AI exposure while hedging macro/EM project execution and FX risk; rebalance monthly against inventory and milestone flow.