The AI Race Has a Jet Engine Problem.

Imagine the investment is approved. The land is secured. The chips are ordered. The launch date is on the board’s calendar. Then someone asks a question that should have been answered months ago: when does the electricity arrive?

The AI boom has produced an unlikely industrial scene. Reuters reported on 9 November that developers and utilities are searching for second-hand gas turbines, reviving mothballed equipment and converting retired jet engines to generate electricity. Its reporting describes new US turbine deliveries averaging around six years, with some discussions focused on 2030–31. Those figures describe the market reported by Reuters, not a universal lead time for every supplier or technology. [1]

The image is striking: machinery from aviation’s past helping to power computing’s future. For executives, it raises a harder question. Does your company control its delivery date—or has that control quietly moved into someone else’s factory?

A real order behind the headline

On 16 April 2026, PROENERGY announced a contract for 13 PE6000 aeroderivative turbine-generator sets for AI infrastructure developer Crusoe. At 50 MW each, the order represents 650 MW of generating capacity intended for upcoming hyperscale data-centre projects. [2]

This is a specific equipment commitment, rather than evidence that the capacity is already operating. Nor does the announcement establish that these particular units are second-hand. The distinction matters: the wider hunt for used machinery and this announced manufactured-equipment order illustrate different responses to the same pressure.

Our reading of the order is straightforward. Power availability has become important enough to shape infrastructure procurement early. A technology strategy now needs an industrial delivery strategy alongside it.

The bottleneck moves. The business case moves with it.

Securing a generator does not, by itself, make a data centre ready. Goldman Sachs’ October 2026 analysis describes an interconnected set of constraints: transmission, grid connections, permitting, equipment, water considerations and skilled labour. It also notes that onsite generation can face its own environmental and air-permitting requirements. [3]

The IEA’s 2026 assessment adds a useful caution: given turbine shortages, onsite generation is not necessarily a faster route to market at scale. [4] Buying your way around one queue may place you in another.

For a project team, that changes the meaning of progress. A building can look nearly complete while the system needed to operate it remains unfinished. A procurement report can show purchase orders issued while critical manufacturing slots remain uncertain.

For the sponsor, the consequence is financial: capital can be committed long before the asset can earn. For the customer, it is a service promise at risk. For the contractor, it is an interface problem that can become a dispute over responsibility.

Three decisions that belong in the boardroom

1. Approve a dependency plan with the investment

Before endorsing the launch date, ask the team to identify every external commitment that could prevent operations. For each, name an accountable owner, the evidence supporting its date, the latest decision point and a credible fallback.

A useful board question is: “Which date in this business case is still somebody else’s estimate?” That question exposes uncertainty more effectively than another overall percentage-complete chart.

2. Evaluate the cost of being late alongside the purchase price

A cheaper package may create a worse business outcome if it arrives after the revenue window. Equally, paying a premium for early delivery can destroy value when other dependencies are not ready.

Compare options using the same assumptions: realistic operating dates, contribution earned or deferred, temporary operating costs, financing implications and the probability of successful integration. Test a downside case. Keep revenue separate from profit, and avoid counting the same delay impact twice.

The commercial team should also establish what the proposed delivery commitment actually covers: factory completion, shipment, installation or successful acceptance. Those are different milestones.

3. Make recovery options operationally credible

A substitute package is only a recovery plan when engineers, operators and suppliers can demonstrate that it fits the intended duty. Check performance, service support, spares, interfaces and acceptance testing before treating a shorter quotation as a recovered schedule.

The same discipline applies to phased openings. Launching a smaller first phase may protect value, but only if that phase can operate safely and independently, serve a real customer need and avoid expensive rework later.

The opportunity for Saudi Arabia and the Gulf

For regional developers and suppliers, the useful question is where they can reduce uncertainty between an announced investment and an operating asset.

Our view is that opportunities deserve investigation in equipment integration, commissioning, maintenance capability, procurement planning and programme coordination. These are potential avenues for business development—not claims that every project has an open contract or an assured return.

For bidders, the implication is practical: demonstrate how your proposal makes the customer’s operating date more credible. Show manufacturing commitments, interface ownership, test milestones and service readiness. A promise of “fast delivery” is less persuasive than a sequence the customer can verify.

For PMOs, report readiness to operate alongside construction progress. For sponsors, release major commitments against evidence that critical dependencies are converging. For advisers, challenge the assumptions connecting the programme to the business case.

Your turn: the launch-date dilemma

The question to take into your next meeting

The jet engine is the memorable part of this story. The management lesson is what happens when ambitious commitments outrun the physical systems required to deliver them.

Ask your team to name the single external dependency most likely to move your operating date. Then ask for the evidence that it will be ready.

Your strategy is only as fast as the dependency you forgot to manage.