Industrials, Capital Goods & Transport

What happened

France’s draft 2027 budget proposes raising a tax on the most profitable motorway and large-airport operators from 4.6% to as much as 12.2%, potentially generating an additional €800m annually across the operators affected [1]. The proposal comes as three of VINCI’s historic French motorway concessions approach expiry: Escota in February 2032, Cofiroute in June 2034 and ASF in April 2036 [2]. In 2025, VINCI Autoroutes generated €2.64bn in free cash flow, accounting for approximately 38% of the group’s total [3].

What people are missing

Placing emphasis on the 38% is crucial, as VINCI Autoroutes brought in only 9% of the group’s revenue that same year [3]. This gap can be explained by the motorway being already built and cheap to run: the road has been paid for, drivers arrive without needing to be won as customers, and staff and maintenance costs are small compared with what is collected in tolls. In 2025, VINCI Autoroutes’ EBITDA margin was 71%, compared with 9.5% at Energy Solutions, the group’s largest business by revenue [3]. The motorways are therefore a relatively small part of what VINCI sells and a large part of what it keeps. That distinction changes what the expiry dates mean: judged by revenue, VINCI loses roughly a tenth of its sales when these concessions end, but judged by cash, it loses more than a third of what the entire group generates [3]. VINCI has reduced its reliance on these motorways for earnings: their share of group EBITDA fell from roughly 59% in 2006 to about 35% in 2025 [2]. That 35% is lower than the 38% free cash flow share because EBITDA is measured before investment, tax and interest. Even so, the businesses growing in their place are not there yet in cash terms: in 2025 VINCI Airports generated €1.25bn of free cash flow and Energy Solutions €1.20bn, each under half of Autoroutes’ €2.64bn, while VINCI Highways, the overseas roads arm, generated €86m [3].

Why it matters

The proposed tax increase could put pressure on VINCI’s near-term earnings, while the expiry of its historic motorway concessions presents a longer-term challenge to cash flow generation. Although Airports, Energy Solutions and international highways are contributing to the group’s diversification, their combined 2025 free cash flow of approximately €2.53bn remained below VINCI Autoroutes’ €2.64bn [3]. For investors, the key question is whether these businesses can grow enough to replace the cash flows from the expiring concessions.

What we are watching next

At VINCI’s full-year results in February 2027, we will be watching whether the company states what the higher tax would cost, since its 2026 guidance was explicitly based on no change in taxation [3]. Over the longer run, the figure to follow is the combined free cash flow of Airports, Energy Solutions and Highways. In 2025, the three together generated €2.53bn (own calculation), still short of Autoroutes’ €2.64bn despite earning several times its revenue [3].

Sources

1. JeChange — Budget 2027: Autoroutes, Taxe de 12,2% (2026)
2. VINCI — 2026 Half-Year Results Presentation (2026)
3. VINCI — Full-Year 2025 Results (5 February 2026)

What happened

Siemens reported record orders of €27.9bn in Q3 FY2026, up 14% on a comparable basis from €24.7bn a year earlier [1]. The increase was led by Smart Infrastructure, where data-centre orders recorded triple-digit growth over the first nine months of FY2026 [1]. At the same time, Siemens’ total order backlog reached a record €132bn, alongside quarterly revenue of €20.8bn, up 8% on a comparable basis [1]. This growth coincides with a new partnership Siemens formed with Rittal in March 2026 to jointly develop power-distribution and cooling infrastructure for AI data centres [2].

What people are missing

This jump in record orders alongside data-centre demand poses an interesting question: while it implies demand for Siemens’ equipment is strong, it doesn’t reveal whether Siemens can turn the order book into delivered, revenue-generating projects at the same speed — a gap underlined by orders growing nearly twice as fast as revenue (14% vs 8%).

AI chips require far more power than previous hardware generations. Because a chip does no mechanical work, the electrical power it draws is released as heat — power in equals heat out. AI workloads also require chips to “talk” to each other at extremely high speeds, which only happens when they sit physically close together, so chips are packed tightly onto a rack. As more chips are packed in, the rack’s total power draw, and therefore the heat it releases, rises in proportion. Siemens itself states that AI data-centre power densities now regularly exceed 100 kW per rack — roughly ten times the 8–12 kW traditional racks were built for — and could exceed 1 MW by 2030, a jump existing cooling and power-distribution infrastructure was never designed to handle [2][3].

Air cannot carry this heat away fast enough once rack density passes a certain point, which is why data centres are now shifting to liquid cooling with Rittal’s help [2]: a liquid-cooled cold plate pressed onto a chip removes heat roughly 300 times faster than air blown across it [3]. That switch cannot simply be scaled up overnight — an order can be signed in a day, but the liquid-cooling equipment still needs to be designed, tested, certified and manufactured before it can be installed, which takes far longer. That the partnership was needed at all, rather than Siemens and Rittal simply producing more of their existing air-cooling equipment, shows this is a genuine engineering challenge rather than a question of selling more units [2] — a strong signal that converting this slice of the backlog into revenue will take longer, or be less predictable, than Siemens’ more established product lines.

Why it matters

This uncertainty extends beyond Siemens’ own delivery capability: the hyperscalers and data-centre operators placing these orders are themselves committing vast sums of capex to AI infrastructure without yet proving it will generate a sufficient return. Siemens acknowledged this directly on its Q3 FY2026 earnings call, flagging potential volatility in data-centre order intake due to the lumpiness of large contracts, which could affect revenue visibility going forward [4]. For Siemens, the key question is how quickly its growing data-centre backlog turns into sales, since delivery constraints could make revenue growth less predictable [1]. Beyond that, stronger orders only translate into stronger earnings if Siemens can deliver them profitably, rather than simply building a larger backlog.

What we are watching next

At Siemens’ next results, we will be watching how quickly its record backlog converts into revenue and whether Smart Infrastructure can hold its margins as data-centre orders are delivered, plus any further disclosure on the size and timing of those orders.

Sources

1. Siemens — Q3 FY2026 Earnings Release (6 August 2026)
2. Siemens and Rittal — Joint Press Release: “Siemens and Rittal Partner to Design Data Center Infrastructure of the Future” (17 March 2026)
3. Network World, Sean Michael Kerner — “Why AI Rack Densities Make Liquid Cooling Nonnegotiable” (25 March 2026)
4. Siemens — Q3 FY2026 Earnings Call Transcript (6 August 2026)