Procurement notices are dull documents that occasionally say something a minister will not. The one the British Home Office published on October 30 runs to a date in December 2029 and puts a number on a relationship the department has never had to defend in public.
The notice sets out a renewal of the department’s public cloud contract with Amazon Web Services at a total of £1.05 billion, or about $1.42 billion. The term runs from December 2027 to December 2029 and can be extended by two further years. Excluding value added tax, the contract is valued at £881.6 million. A formal contract notice is expected on January 4, 2027.
The distance between the notice and the term it describes is the first thing worth noticing: fourteen months, at a value published before anyone has bid. A department that expects to open the competition in January 2027 has already told the market which supplier it expects to award the contract to, and how much it expects to pay. That is not how a competitive tender is supposed to begin. It is how an incumbent with an entrenched position signals that the competition will be, at most, a formality.
The spending history explains the confidence. BidSkim, which tracks public procurement, calculates that the Home Office has spent £444.6 million with AWS since 2020, and that the new contract implies annual spending more than doubling. The comparison depends on the base year, because a contract that began modestly and grew as more services moved to the cloud has a recent run rate well above its six-year average.
Set out plainly, the six-year total works out to roughly £74 million a year on average. The new term, priced before tax, works out to roughly £440 million a year. Even allowing for a contract that expanded steadily, the step from the historical average to the proposed annual rate is a multiple, not a margin. The department is not renewing a service it already buys. It is buying a larger one.
What that larger service consists of is left vague in the notice, and the vagueness is standard. Public cloud contracts are written to cover whatever the buyer needs over the term, because the alternative is to run a new competition every time a workload moves. The department’s systems touch immigration, policing and identity, which means the workloads are the kind that cannot easily be moved once they are built, and the cost of moving them is measured in years rather than months.
That is the lock-in argument in its plainest form, and it does not require anyone to behave badly. A department that builds on one provider’s managed services, its identity system, its databases and its security tooling has made a series of small decisions that individually saved money and collectively made the exit expensive. Egress fees are the visible part of the bill. The larger part is the staff who know one console, the auditors who have approved one set of controls, and the applications that call proprietary services by name.
The counterargument, which the department has not had to make in public, is that the alternative is worse. Running the same systems in a departmental data center would require capital, staff and a security posture that the department has spent a decade dismantling. Buying from two providers would double the integration cost without removing the dependency on either. The logic is not wrong. It is simply expensive, and the expense arrives on a schedule set by the seller.
The size of the number relative to the department’s own visibility is the political risk. A contract worth more than a billion pounds is a line item a parliamentary committee will ask about, and the answers available to the department are limited by the notice itself: it discloses a total, a term and an option to extend, and nothing about unit prices, service levels or exit terms. Those are the details that determine whether the renewal is good value, and they are the details that will be negotiated after the market has been told there is no real alternative.
The January 2027 date gives the department room to change its mind, and gives rivals room to complain. A contract notice is a legal step, and a competition that names an intended supplier while publishing a value can be challenged if the process that follows is not genuine. A prior information notice exists to let the market prepare. It does not require the market to be told the outcome in advance.
The broader question is whether the British state can keep buying computing this way. Every large department now runs on a handful of providers, and each renewal is larger than the last because the estate grows and the exit costs rise. The pattern is stable until it is not: a provider outage, a price increase, a change in the security relationship, and a department discovers that the contract it signed to save money is the contract that determines what it can do.
For now the arithmetic is simple. A department that has spent £444.6 million since 2020 is preparing to commit £881.6 million before tax over two years, with an option for two more. The formal notice arrives in January 2027, the term begins in December of that year, and the question a committee will eventually ask is the one the notice cannot answer: what, exactly, does the department get for a sum that is larger than everything it has spent on the same service so far.
