The UK government’s new deal with Google Cloud aims to train 100,000 civil servants in modern tech by 2030, but behind the big number is a deeper story. It’s less about ambition and more about catching up. For years, outdated systems have bogged down public services, and this move hints at a government trying—at last—to cut loose from the legacy tech that’s slowed it for decades.
Announced during the Google Cloud Summit in London, the pact falls under the government’s wider effort to ensure that one in ten public officials becomes a “tech expert.” But questions remain about how much real power this initiative holds. Technology secretary Peter Kyle described the deal as a way to strengthen the public sector’s hand when negotiating with major tech vendors, suggesting that, until now, fragmented procurement had left departments isolated and without leverage.
The lack of detail around the agreement leaves gaps. The government did not specify what exactly Google Cloud will deliver, or how the partnership will support departments still wrestling with entrenched platforms. For instance, the Home Office continues to operate an outdated asylum case system nearly 25 years old, despite launching a new platform years ago. That disconnect reflects a pattern: bold announcements about digital transformation followed by slow, uneven execution.
Alongside Google Cloud, the government says DeepMind, Google’s AI division, will collaborate with in-house experts to deploy emerging technologies and support scientific research. However, the challenge lies not in the ambition but in delivery. Past audits have warned of missed opportunities totaling billions in productivity losses due to underused or poorly integrated systems.
It’s a big headline, sure. But real change? That’s slower, messier, and hinges on things that don’t fit neatly into a press release. Departments still speak different languages, procurement remains stuck in another era, and legacy systems cling on like rust. If those pieces don’t move together, it won’t matter how advanced the AI is. Come 2030, we could still be asking why powerful tools were handed over to a system that couldn’t use them.
