Summary
- Power Systems took £4.6 billion of orders in the half, more than 50% above the prior year, producing a 1.8 times book-to-bill ratio.
- Within power generation, order intake rose 55% as data centres bought both backup and prime-power solutions.
- Power Systems underlying revenue rose 28% to £2.604 billion; power-generation revenue rose 41%, including strong data-centre growth.
- The division’s underlying operating profit rose 72% to £528 million and its margin increased from 15.3% to 20.3%.
- Rolls-Royce expects power-generation original-equipment revenue to grow 25% through 2030 and plans a denser next-generation engine for 2028.
- The disclosure does not provide data-centre-only orders, revenue, profit, megawatts, customers or the split between backup and prime power.
Orders show commitment, not completed capacity
The first ledger is demand. Power Systems recorded £4.6 billion of order intake, more than 50% higher than a year earlier, while book-to-bill reached 1.8 times. For every pound of revenue recognised across the division, roughly £1.80 of new orders entered the book. Its total backlog rose 42% to £8.0 billion.
The data-centre signal sits one level below those totals. Rolls-Royce says power-generation order intake rose 55%, reflecting demand for backup and prime-power solutions for data centres. It does not say that all £4.6 billion came from data centres, or that the £8.0 billion backlog belongs to that end market. The division also serves other power-generation and industrial applications.
An order is a commercial commitment under specified terms. It is not recognised revenue, an installed engine, an energised campus or usable compute capacity. Cancellation provisions, production schedules, site readiness and commissioning still stand between the order book and operation. The 1.8-times ratio therefore establishes future workload and visibility; it does not establish how many megawatts are live.
Revenue proves that part of the queue has converted
The second ledger is delivery. Power Systems underlying revenue reached £2.604 billion, 28% higher year on year. Power-generation revenue rose faster, by 41%, and the company explicitly includes strong data-centre growth in that movement.
That disclosure matters because revenue requires more than market interest. Equipment or contracted work has progressed far enough to be recognised under the company’s accounting. The data-centre power thesis is no longer supported only by announced partnerships, enquiries or a future pipeline. Some demand has entered the income statement.
The denominator remains deliberately bounded. £2.604 billion is total Power Systems revenue, not data-centre revenue. The 41% is growth in the wider power-generation activity, not a disclosed growth rate for data centres alone. Rolls-Royce does not publish the starting revenue, customer list, unit count or geographic mix behind the data-centre contribution. The right conclusion is that data centres are a material disclosed driver, not that their exact share can be reverse-engineered.
Margin is the new information, but attribution has limits
The third ledger is value capture. Power Systems underlying operating profit rose 72% to £528 million. Its operating margin increased by five percentage points, from 15.3% to 20.3%. Profit grew considerably faster than revenue, showing operating leverage and a more favourable earnings mix.
Rolls-Royce attributes the improvement partly to power generation driven by data centres, but also to higher volume, mix and commercial optimisation. Those causes cannot be collapsed into one. A factory carrying more throughput can absorb fixed cost better. Product and service mix can move pricing and gross margin. Commercial optimisation can improve contract terms. Data-centre demand may support all three, but the result does not allocate the £528 million or the margin increase by end market.
This is still stronger evidence than an order headline. A supplier can win fashionable contracts that dilute margin. Here the division is reporting demand, delivery and a wider profit expansion at the same time. The next test is whether the 20%-plus margin persists as the backlog converts, particularly if customers demand faster delivery, bespoke site integration or price concessions.
Prime power changes the product from insurance to infrastructure
Backup generators and prime-power engines solve different problems. Backup equipment waits for a grid failure and runs intermittently. Prime power supplies the site’s main electricity for sustained operation. Rolls-Royce reports demand for both and highlights growing use of its Series 4000 gas reciprocating engines as prime power for data centres.
That distinction changes the economic role of the engine. A backup unit is an insurance asset around an existing grid connection. A prime-power fleet becomes part of the production system that allows servers to operate. It must meet a different duty cycle, fuel plan, maintenance regime, emissions context and availability standard.
The company does not say why every buyer selected prime power. It is reasonable to infer that some projects value an on-site route around slow or constrained grid access, but that inference is not a disclosed customer explanation. Neither does the result quantify how the 55% order growth divides between backup and prime power. The useful signal is narrower: data-centre customers are buying Rolls-Royce equipment for more than emergency standby, widening the addressable value per site.
Density sells time and land as well as electricity
Rolls-Royce plans a next-generation engine for 2028 with 20% higher power density and says full-system testing has started in the United States. Higher density means more output from a given equipment footprint. Around a data-centre campus, that can reduce the land, building and balance-of-plant burden attached to a target power level.
It may also increase the supplier’s value in schedules where electrical capacity, rather than server procurement, sets the opening date. A denser modular system can be staged with construction and compute deployment. That does not make it equivalent to a grid connection: fuel supply, air quality, permitting, noise, heat and maintenance remain site constraints.
The 20% figure is a product target, not a field result. The engine is planned for 2028 and testing does not establish commercial launch, reliability or customer acceptance. Its relevance today is strategic. Rolls-Royce is investing as though high-density on-site generation will remain a durable data-centre requirement rather than a short bridge.
The 2030 upgrade is guidance, not booked performance
Management now expects power-generation original-equipment revenue to grow by 25% through 2030, compared with its previous 20% mid-term expectation. The revision follows stronger orders and raises the divisional opportunity. It also extends the period over which execution must be judged.
The wording matters. This is an expectation for original-equipment revenue, not a statement that revenue has already increased by another 25%. It does not specify a data-centre-only base, guarantee annual compounding or convert the backlog into a fixed 2030 outcome. Production capacity, supply chains, project permitting, fuel economics and customer commissioning will determine how much demand becomes recognised revenue.
At group level, Rolls-Royce raised guidance to £4.7 billion-£4.9 billion of underlying operating profit and £3.8 billion-£4.0 billion of free cash flow. Power Systems contributes to that confidence, but defence and civil aerospace remain part of the group result. The data-centre theme should not be credited with every pound of the upgrade.
The next result must reconcile the three ledgers
The next evidence should connect orders, revenue and margin without pretending they are the same measure. Order intake and backlog should show whether demand persists after an exceptional buildout phase. Revenue should show whether factories and project teams convert that workload on schedule. Margin and cash should show whether speed and complexity preserve value rather than merely add volume.
The operational questions are now concrete. How much production capacity is added for Series 4000 demand? Does prime power become a larger disclosed share? Do delivery lead times shorten or lengthen? Can margin stay near 20% as orders convert? Does the 2028 engine pass system testing and enter customer programmes? None requires an invented data-centre revenue estimate.
Rolls-Royce has not published a clean data-centre segment. It has published something more useful than another market-size claim: a trace of the demand through orders, recognised revenue and divisional profit. The discipline is to follow that trace without turning the wider division into a proxy for an undisclosed customer market.