The cost of using AI fell sharply in 2025, making advanced tools widely accessible. But according to the analysts, the real breakthrough will not come from virtual AI alone. Instead, growth will accelerate when AI is integrated into physical capital goods such as autonomous vehicles, drones and humanoid robots.
“Virtual AI has improved office workflows, but the impact on productivity is still limited,” says Thomas Thygesen, head of equity strategy at SEB. “The real revolution will happen when AI moves from software to hardware.”
Signs of this transition are already visible. Autonomous vehicle operators like Waymo and Apollo have begun scaling services, and Chinese manufacturers delivered thousands of humanoid robots last year. Still, these numbers remain small compared to global demand, meaning widespread economic effects will take time.
The stock market reflects this shift. While hyperscalers and model developers have seen flat or negative returns since late 2025, companies producing semiconductors and chip-making equipment have surged as the strong demand for AI hardware creates real physical bottlenecks.
A recent SEB Nordic Seminar survey supports this outlook. Most companies see AI as a long-term transformation rather than an immediate game-changer.
“We expect productivity growth to pick up, but it will likely take a few years,” says equity strategist Mads Bossen.
He says factories and infrastructure for AI-enabled products must be built before macro-level gains appear.
By 2030, embedded AI tools – from robots to autonomous mining equipment – are expected to become mainstream. Historically, such periods of accelerating productivity have coincided with strong equity markets, suggesting opportunities for investors focused on the hardware side of the AI value chain.