AI took centre stage at the opening of London Tech Week, where UK Prime Minister Keir Starmer and NVIDIA CEO Jensen Huang shared a stage to outline what comes next. What began with policy announcements quickly shifted into a focused conversation about how AI is moving from niche capability to critical infrastructure.
The message was direct: AI is no longer a specialised tool—it’s becoming the foundation for how industries operate. From compute gaps to skills access, the session offered a clear view of what still needs to be built for this transition to happen at scale.
AI Is Infrastructure Now
Huang’s core argument was straightforward: “AI is both a technology but it is also an infrastructure.” Why? “Because it affects so many industries simultaneously… healthcare and government and education and manufacturing and financial services.”
He compared AI to past foundational systems like electricity and internet connectivity—things every sector eventually built on. “We're going to have a new infrastructure, and this new infrastructure is called artificial intelligence.”
Starmer echoed this framing, noting that “every department in government” is now being asked how it can use AI to transform delivery, not just improve efficiency.
Prompting Is the New Skillset
One of Huang’s most resonant points was about accessibility: “The new programming language is called human.” Most people don’t know Python, he said. “Everybody, as you know, knows human.”
His broader point: programming AI is increasingly about describing intent. “You say: you are an incredible poet, you are deeply steeped in Shakespeare... and it generates a wonderful poem.” If you don’t like it? “You could say, I feel like you could do even better—and it does.”
Starmer brought it back to education, arguing that “you can't aspire to do something if you don't really know what it is.” Embedding AI exposure in schools is about more than skills—it’s about broadening ambition.
The Compute Bottleneck
Despite the UK’s strengths in research and investment, Huang highlighted a critical missing piece: “It is surprising—this is the largest AI ecosystem in the world without its own infrastructure.”
His point was that the UK has the ingredients—talent, capital, university research, startups—but lacks the compute to match. “You can’t do machine learning without a machine,” he said, adding that other scientific fields all have core instruments: “If you were a particle physicist, you need a linear accelerator… if you're in the world of AI, you need a machine.”
That’s why NVIDIA is setting up a lab in the UK and partnering to expand access to local compute capacity.
Our Takeaways from the session
- AI is infrastructure. It’s not a tool to adopt—it’s a layer to build on.
- Prompting is a skill. The interface is human language, and fluency matters.
- Compute is strategic. Innovation depends on having the capacity to build.