Recently I heard a story about forklifts that explains something government consistently gets wrong with technology.
A seasoned engineering executive, advising a sprawling industrial firm that had grown through acquisition, shared an observation with me. Each business unit runs independently - buying its own equipment, software, and yes, its own assortment of forklifts. The waste isn't occurring in the big cost centres. It's happening in every small one.
The same story plays out across drills, batteries, and oil specifications. Different suppliers, different maintenance contracts, different parts - despite sites doing essentially identical work. The opportunities for consolidation aren't exotic. Common parts, shared batteries, standardised oil are simple to articulate, but notoriously difficult to implement, because they require something more elusive than innovation. Strategy.
This is precisely where government technology is struggling.
You don't need to understand container orchestration to grasp this pattern. Departments are independently building and maintaining hundreds of similar technology platforms, and those platforms almost never get retired. The same infrastructure, procured separately, configured differently, maintained in isolation, accumulating year after year. Multiple departments piloting the same AI code assistants. Duplicated machine learning experiments solving identical problems, sometimes just office floors apart. Separate cloud platforms doing essentially the same work.
Sprawl is easy to dress up in the language of agility and transformation. Each platform launch is celebrated as essential at a departmental level. But it's rare to hear anyone asking the forklift question. Why are we maintaining such a vast estate when we could standardise?
Effective leaders understand that organisations create differentiated value by ruthlessly commoditising everything beneath it. You don't hand-craft every component, you standardise the boring bits so you can focus resources on things that genuinely require innovation.
Instead, government is spending time, money, and capability on infrastructure that should be a commodity. Every bespoke platform is time not spent on services, on user needs, on citizen outcomes. Every duplicated AI experiment is budget not available for deployment at scale. The opportunity cost is staggering, and compounding every single day.
One root cause is incentives. Funding arrives in departmental silos, and programme-based accounting rewards delivery within your own organisation, not reuse of someone else's work. Senior Responsible Owners are measured on their programme's success, not on whole-of-government efficiency. Add the genuine political need to demonstrate progress quickly, and you get the current situation with a thousand parallel efforts, each locally rational but globally wasteful.
Government is starting to acknowledge this. The Blueprint for Modern Digital Government called for whole-of-public-sector agreements and reformation of the Government Digital Service signals renewed intent around cross-government working. These lay administrative groundwork for a different approach. But groundwork isn't strategy.
The work ahead isn't glamorous. Standardised platforms that departments use rather than build. Shared AI infrastructure that teams deploy models on rather than spinning up from scratch. Common procurement vehicles delivering enterprise software at true scale. Centralised expertise that advises rather than letting every team learn the same expensive lessons independently.
This challenge isn't fundamentally technical, which is why it cannot be foisted upon technologists alone. Open source provides a well-refined, battle-tested approach. When you standardise on open technologies, you're not locked into a single vendor's roadmap or pricing structure. You're choosing infrastructure designed to be shared, scrutinised, and maintained collectively - freeing departments to focus on work that genuinely differentiates them.
Success needs to be measured not by how many platforms have been launched, but by how few are needed. The team that says "we're using the existing platform" should be praised as much as - or more than - the team that builds something new. Without that friction, nothing changes.
Sonia Patel, the new Government CTO faces a simple but difficult question about preparedness to make economies of scale the priority, even when it conflicts with departmental preference? But the answer doesn't rest with one office. Permanent Secretaries control spending. Treasury sets the rules. Without their active support, no amount of technical leadership will shift the pattern.
The transformation won't come from innovation labs or AI pilots. It will come from the unglamorous work of consolidation. Reducing hundreds of platforms to dozens, standardising AI infrastructure so experiments can scale, creating the boring, reliable, commodity foundation that lets government focus resources on genuinely differentiated services - without locking into exploitative contracts with proprietary vendors.
The engineering executive I spoke with had unbounded clarity about the vision. Differentiate where it matters, commoditise everywhere else. The lesson translates perfectly to government technology.
The question for government is whether the system can reward this shift. The alternative is already visible, with AI pilots stalling at proof-of-concept, productivity promises deferred, and a widening gap between digital rhetoric and actual delivery. The cost goes beyond wasted money, wasting credibility simultaneously.
Standardise on the same forklifts. Make the parts work everywhere. Then focus on what government is actually meant to be building.
Red Hat is hosting the discussion table 'Operationalising AI' at the 2026 Government Transformation Summit. Find out more below.