29th April, 2026: AI has captured global attention through products people can see and use. Chatbots, image generators, coding assistants and workflow tools have become the public face of the sector. They are fast to launch, easy to demonstrate and designed to attract users quickly.
But Nicole Junkermann believes the most important shift in AI is happening elsewhere.
The next stage of artificial intelligence will be defined less by interfaces and more by infrastructure. As adoption accelerates, the central question is no longer simply what AI can do. It is whether the systems beneath it can support that demand at scale.
That means compute capacity, data architecture, energy supply, connectivity, cybersecurity and the industrial-grade platforms that allow AI to move from experimentation into everyday economic life.
The first wave of AI was driven by model breakthroughs and consumer curiosity. That phase rewarded novelty, speed and visibility. But every successful application creates new pressure lower down the stack.
More users mean more compute. More automation means heavier data demands. More enterprise deployment means greater requirements around reliability, compliance and integration with legacy systems. At the national level, rising AI adoption also raises questions about power generation, strategic resilience and digital sovereignty.
In short, demand at the top of the market is exposing constraints underneath it.
This is a familiar pattern in technology. Early excitement tends to centre on what is visible. Longer-term value often accrues to what becomes indispensable.
Apps can rise quickly, but they can also be replaced quickly. User preferences change, interfaces evolve, and new entrants emerge at speed. Infrastructure behaves differently.
Critical systems are expensive to build, difficult to replicate and deeply embedded once adopted. When a platform becomes part of how businesses operate or how industries function, switching costs rise and strategic value increases.
That is why the economics of AI are shifting. The sector will continue to produce successful applications, but the companies enabling compute, data flows, deployment and resilience will occupy a stronger position over time.
The glamour of AI may sit at the surface. The leverage increasingly sits below it.
This is one reason many sophisticated investors are looking beyond the application layer. Rather than asking only which consumer product will gain users fastest, they are asking which systems the wider market will depend on for years.
That broader thesis helps explain why investors such as Nicole Junkermann, through NJF Capital, have backed businesses operating close to foundational layers of major industries. The portfolio includes names such as Groq in AI compute, Owkin in AI-enabled healthcare, Rippling in enterprise operations, Revolut in financial infrastructure and SpaceX in next-generation connectivity and launch systems.
These businesses operate in different markets, but they share a common feature: they provide capabilities that others build upon.
AI will keep producing new products, new use cases and new headline moments. But as the sector matures, competitive advantage will increasingly belong to those solving harder structural problems.
Who can provide enough compute? Who can power growing demand? Who can integrate AI securely into complex organisations? Who can build trusted systems that scale globally?
Those questions will shape the next decade of AI more than the launch of any single app.
The public story of artificial intelligence may still be written through interfaces. The deeper commercial story will be written through infrastructure.
About Nicole Junkermann
Nicole Junkermann is an international investor focused on technology, artificial intelligence and life sciences. She is the founder of NJF Holdings, leading its venture arm NJF Capital, which backs early-stage companies in deep tech, healthcare and data-driven systems, and Gameday by NJF Holdings, focused on technology-led transformation in sport and media.
Visit Nicole’s Substack for regular updates.