The UK government’s decision to funnel nearly £9 billion into Microsoft’s cloud and AI services over five years has ignited a fierce debate over public money, market power, and the future of digital sovereignty. While the Strategic Partnership Arrangement 2024 (SPA24) promises simplified procurement and early access to Copilot AI, critics warn that opaque pricing, entrenched vendor lock-in, and the lack of independent oversight could leave taxpayers footing a bill that far exceeds the advertised value. This agreement, which replaces the earlier DTA21 arrangement, took effect on 1 November 2024 and covers Microsoft 365, Azure, Business Applications, and—for the first time—Microsoft Copilot, all under an aggregated procurement model designed to deliver “enhanced value” through scale.

Parliamentary disclosures have revealed approximately £1.9 billion of Microsoft licence spend routed through SPA24 in a recent financial year window, annualising to a multi-billion-pound commitment. Commentators have rounded the five-year total to about £9 billion, a figure that has become the focal point for scrutiny. At a time when every public pound is under the microscope, a procurement of this magnitude demands commensurate transparency and demonstrable savings—two things that, so far, remain elusive.

A Colossus in the Cloud Market

Microsoft’s financial might underscores why this deal is being scrutinised so intensely. For fiscal 2025, the company reported revenue of $281.7 billion and net income of $101.8 billion, with net margins comfortably in the mid-30s. Microsoft Cloud alone generated $168.9 billion, and Azure revenue growth hit 39% year-on-year in the most recent quarter. Large government contracts like SPA24 are not just purchases; they feed revenue lines that drive market valuations and strengthen competitive dynamics.

The UK’s Competition and Markets Authority (CMA) has already concluded that competition in the public cloud infrastructure services market “is not working well.” In its final report published on 31 July 2025, the CMA identified systemic barriers—including proprietary licensing, data egress charges, and technical incompatibilities—that suppress multi-cloud adoption. The regulator found that two hyperscalers, Microsoft and Amazon Web Services (AWS), dominate the UK market, and it specifically highlighted Microsoft’s licensing practices as a concern. The CMA has recommended using its new digital markets powers to designate both companies with Strategic Market Status (SMS), a move that could lead to pro-competition interventions. A decision on which SMS investigation to prioritise is expected in the first quarter of 2026.

This regulatory backdrop transforms routine procurement oversight into a national economic concern about market structure, supplier leverage, and the long-term costs of dependency.

Dissecting the Deal: Promised Benefits and Unverified Savings

SPA24’s proponents point to genuine advantages. Central purchasing through the Crown Commercial Service (CCS) can yield economies of scale, lowering unit prices for commodity items like Office licences and cloud compute. Operational standardisation across government departments can reduce integration friction, and bundling Copilot could accelerate AI adoption for back-office automation and data-driven decision support.

However, the mere existence of plausible benefits does not equate to proven value for money. The deal’s economics remain opaque in three critical areas:

  • Discount transparency: Public statements mention “discounted pricing,” but independent disclosure of baseline prices, marginal discounts on Copilot seats, or pricing elasticity as volumes shift is absent from publicly available documents.
  • Bundled high-cost items: Copilot licences are materially more expensive than base productivity seats. When bundled into multi-year arrangements, they can significantly raise total contract value and recurring costs.
  • Hidden transformation costs: Migration, retraining, identity and access redesign, and exit costs—such as data egress fees and portability work—can overwhelm nominal headline discounts if not included in total cost of ownership (TCO) modelling.

Without contract-level transparency, it is impossible for taxpayers to assess whether SPA24’s “enhanced value” is a real saving or simply a convenience premium. The National Audit Office or an independent commercial review must be allowed to audit the commercial mechanics and report publicly.

Copilot and AI: Strategic Opportunity or Lock-in Accelerator?

Generative AI features in Copilot can boost productivity on repetitive tasks, speed document drafting, and surface insights from large data stores. For customer-facing services and back-office processing, there are plausible use cases that might justify the licence cost. Microsoft’s deep integration of Copilot across Office and Teams also promises low friction for large user populations.

Yet the risks are substantial:

  • Ecosystem entrenchment: Designating Copilot as the default AI engine in a government MoU shifts future project dependencies toward Microsoft’s APIs, data models, and governance frameworks.
  • Data governance and sovereignty: As public datasets and workflows are processed by a vendor’s AI, exit and portability options become more complex—particularly where models or derivative outputs are concerned.
  • Comparative pricing opacity: Copilot seat pricing and usage-based AI costs can vary dramatically depending on contract terms. Without clear comparators, the public sector cannot know if it is paying a market rate or a premium tied to vendor lock-in.

A government-wide AI procurement policy that prioritises interoperability, clear data-sovereignty rules, and mandated exit and portability arrangements would reduce the risk of SPA24 becoming a one-way street. Such a policy should define acceptable data flows, model auditability standards, and independent benchmarking of productivity claims before large-scale consumption begins.

The Crown Commercial Service: Enabler or Preserver of the Status Quo?

The CCS operates as a trading fund and executive agency of the Cabinet Office. It applies levies and commissions on commercial agreements to recover costs and deliver a return, and it counts aggregated demand as a core part of its value proposition. While its annual reporting tracks commercial benefits and levy mechanisms in detail, it does not publish contract-by-contract commission receipts for strategic MoUs like SPA24.

This opacity creates a perception risk: if CCS’s business model rewards revenue scale rather than aggressive supplier competition, incentives can align to preserve stable large suppliers rather than drive down costs. Some commentators have calculated that even a small percentage levy on £9 billion represents a non-trivial sum—roughly £30 million across the contract term. The mechanics of CCS levies are published, but the precise levy schedule applied to SPA24 and the amount retained are not disclosed in a single accessible line item. Until such line-item transparency is provided, these fee estimates should be treated as reasonable but unverified.

Alternatives and Opportunity Costs

The scale of SPA24 invites the question: what else could £9 billion buy? A mere 10% improvement in procurement outcomes would free £900 million—enough to fund thousands of frontline staff or accelerate modernisation in health, transport, and education. This reframes procurement as a lever for public policy, not just a technical exercise.

Are realistic alternatives available? Microsoft’s advantages—scale, compliance certifications, deep integration with existing workflows, and a massive partner ecosystem—are concrete, and replacing that base is neither free nor guaranteed to succeed. International case studies highlight both potential savings and implementation risks:

  • French Gendarmerie (GendBuntu): The migration to a Debian/Ubuntu-based distribution and open-source toolchain demonstrated measurable savings and reduced licence dependency over many years, but it required sustained technical investment and local capability building.
  • Munich’s LiMux: Initially delivered savings and independence, but political changes, interoperability problems, and shifting priorities resulted in a partial rollback. The story underscores that open-source transitions succeed only when governance, skills, and political will are aligned.

Small-scale pilots, rigorous TCO analysis, and hybrid approaches mixing open-source and proprietary stacks can create credible contestability without exposing critical services to operational risk.

Lessons from Past UK Reform Efforts

The Gershon Review (2004) and the coalition government’s 2010 moratoria on non-essential IT spending both imposed discipline and central accountability on public-sector IT. Gershon emphasised procurement consolidation and efficiency; the 2010 moratoria produced immediate cash savings and tighter controls on large IT projects. Both efforts demonstrate that structural levers—procurement frameworks, central scrutiny, and enforced spend pauses—can deliver value, but only if they remain active and transparent governance continues. History shows the discipline can erode without ongoing political and institutional will.

Without active competitive testing, periodic pilot competitions, and mandatory benchmarking of new high-value items like Copilot, central deals risk becoming default choices by convenience rather than choices proven to deliver better value.

Practical Recommendations for Policymakers

  • Increase transparency immediately: Publish SPA24-specific pricing bands, discount schedules, and the methodology used to compute “enhanced value.” Independent reviewers must be allowed to audit the commercial mechanics and report publicly.
  • Mandate external verification: Any claim that SPA24 will save a certain percentage or deliver a specific benefit must be supported by independent ex-post evaluation from the National Audit Office or equivalent, with public reporting.
  • Create AI procurement guardrails: Introduce a government-wide AI policy covering interoperability, model audit trails, data provenance, portability, and exit rights. No large-scale AI procurement should proceed without clear outcome metrics.
  • Fund pilot programmes for alternatives: Allocate a modest but meaningful fund to run open-source and multi-cloud pilots across health, local government, and education, measured with rigorous TCO and service-continuity metrics.
  • Regularly test commercial arrangements: Require scheduled re-competition and benchmarking events (every 12–24 months) for core commodity items and AI services to retain negotiating leverage.
  • Strengthen supplier-agnostic capability: Invest in public-sector cloud and procurement skills so departments can challenge suppliers with meaningful commercial and architectural alternatives.

Conclusion: Pragmatic Partnership or Complacent Dependency?

SPA24 is a pragmatic response to the operational needs of modern public services: governments require scale, proven compliance, and access to leading-edge technologies, and Microsoft is a logical partner in many respects. But pragmatism must not calcify into complacency. Where the deal can deliver genuine value, that value should be measurable, independently verified, and re-contestable. Where risks exist—particularly around Copilot, licensing economics, and market concentration—they must be mitigated through transparency, competition, and active governance.

A five-year, multi-billion-pound procurement deserves more than a single MoU and occasional reassurances. It demands continuous transparency, refreshable competition, and an AI procurement framework that protects the public interest while enabling innovation. That is the only pathway to turning a large vendor agreement into sustained public value rather than an accelerating revenue stream for an already dominant supplier.