When I wrote my MBA dissertation, I focused on business models. That was the moment when I truly understood how important they are. A strong technology, a promising product idea or even a clear market need is not enough by itself. Value is created and captured through the business model: through the way an organisation defines its customers, delivers value, monetises that value and builds the capabilities required to scale.
Since then, I have tended to look at new technologies through a business model lens. Whenever I see an emerging technology or a promising innovation idea, my first question is not only: What can this do? It is also: What kind of business could be built around it?
Over the past years, AI has become my professional field. And the more I worked with AI strategy, enterprise adoption and AI-enabled innovation, the clearer it became to me that we are facing a growing gap. AI technology is advancing extremely fast – but the methods we use to think about AI-enabled business models are not evolving at the same speed.
That is why I started to explore this topic more systematically. My recently published paper is the first step in this effort. It examines why frontier AI may require a new way of thinking about business model innovation. The next step will be to develop a more practical methodology that companies, startups and innovation teams can actually use.
Frontier AI is changing the starting point
A few years ago, most business conversations about AI were about automation, prediction and efficiency. AI could help companies forecast demand, classify risks, recommend products, detect anomalies or optimise processes.
These use cases remain highly valuable. But frontier AI is expanding the scope of what AI systems can do.
Today’s most advanced models can work across text, images, code, data and increasingly complex reasoning tasks. They can generate content, summarise knowledge, support decision-making, simulate scenarios, plan multi-step actions and interact with external tools. In other words, they are beginning to look less like isolated software components and more like reusable cognitive capabilities.
This matters because the business implications are very different.
Classical AI was often designed for a specific task in a specific workflow. Frontier AI can be reused, adapted and recombined across many contexts. It can support customer service, research, software development, education, marketing, management decision support, product design and scientific discovery – sometimes using the same underlying capability layer.
That changes the business model question.
The real question is not only what the model can do
Much of the public discussion around AI focuses on model performance: which model is faster, cheaper, more multimodal or more capable. These are important questions, but they are not sufficient from a business perspective.
The deeper question is this:
How can these new AI capabilities be translated into sustainable value creation and value capture?
History shows that powerful technologies rarely create economic value automatically. Their real impact emerges when organisations redesign processes, products, services, roles, revenue mechanisms and ecosystems around them.
This is especially true for AI. Many companies are already experimenting with generative AI, but not all of them can clearly demonstrate measurable business value. In many cases, the issue is not the technology itself. The issue is the absence of a clear business model logic: What is the value proposition? Who captures the value? How is the solution embedded into workflows? How is trust created? How is the offering scaled? How is it monetised?
In AI, adoption is not the same as value creation.
From cheap prediction to cheap cognitive task execution
One of the most useful earlier ways to understand AI was to describe it as a technology that reduces the cost of prediction. This framing works well for many classical AI applications: forecasting, scoring, classification, recommendation and optimisation.
Frontier AI extends this logic.
It does not only make prediction cheaper. It may also reduce the cost of executing a broader range of cognitive tasks: writing, analysing, comparing, explaining, planning, coding, simulating, coordinating and, in some bounded contexts, acting through tools.
This does not mean human judgment disappears. On the contrary, human goal-setting, accountability, ethical judgment and contextual understanding become even more important. But the boundary between what humans do and what machines can support is shifting.
That is why frontier AI may change not only individual workflows, but the economics of knowledge work and innovation itself.
From AI use cases to AI-enabled innovation systems
For many organisations, the first wave of AI thinking has been use-case driven: identify a business problem, apply AI, measure the benefit. This remains a practical starting point.
But frontier AI invites a broader question: What if AI becomes part of the infrastructure of innovation?
Companies may increasingly use AI not only to automate existing tasks, but to test ideas, simulate outcomes, explore strategic options and accelerate learning cycles.
Some of the most exciting opportunities may emerge beyond today’s familiar chatbot and productivity use cases. Frontier AI could support the training of physical AI systems and robots in simulated environments before they operate in the real world. It may help companies model customer behaviour, group dynamics or market responses before launching a campaign or entering a new segment. It could enable new forms of synthetic experimentation for testing marketing programmes, pricing strategies or customer journeys.
In science and healthcare, AI-driven simulation could accelerate the search for new drugs, treatments and research hypotheses. In industrial contexts, it could help design, test and optimise systems before they are physically built. In strategy, it could support scenario exploration at a speed and scale that was previously impossible.
These examples point to a broader shift: AI is becoming not only a tool for automation, but a new infrastructure for experimentation.
Where new business models may emerge
This creates new opportunity spaces across the AI value chain.
Some companies will build or provide the underlying infrastructure: compute, cloud platforms, data layers and governance tools. Others will create foundation model platforms or orchestration layers that help AI systems interact with tools, workflows and enterprise systems.
Many of the most attractive business opportunities may appear closer to the customer: enterprise copilots, domain-specific AI applications, grounded knowledge systems, agentic workflow solutions and vertical AI products designed for specific industries.
There will also be a major role for transformation partners who help organisations redesign processes, governance models and operating models around AI. In many cases, the differentiator will not be access to the model itself, but the ability to combine AI with domain knowledge, trusted data, workflow integration, compliance and change management.
In other words, the winners will not necessarily be those who own the largest model. They may be those who build the most valuable business model around AI-enabled capabilities.
Governance becomes part of the value proposition
Frontier AI also introduces new risks. When the same AI capabilities are reused across many applications, errors and weaknesses can propagate across systems. Questions of reliability, auditability, accountability, data rights and human oversight become central.
This means governance is not just a compliance issue. In frontier AI, governance can become part of the business model itself.
Customers will not only ask whether an AI solution is powerful. They will ask whether it is trustworthy, explainable, secure, aligned with their processes and safe to deploy in real business environments.
For many AI businesses, trust will be a source of competitive advantage.
The next step: business model innovation for frontier AI
Frontier AI is not just another wave of software innovation. It changes how we think about capabilities, work, experimentation and value creation.
That is why we need better methods for designing AI-enabled business models. Existing tools remain useful, but they were not built for a world where AI acts as a reusable cognitive capability across products, workflows, organisations and ecosystems.
My recently published paper is the beginning of this work. It sets out the conceptual foundation for understanding why frontier AI challenges existing business model thinking. The next step is to move from conceptual analysis to a practical methodology for designing, evaluating and iterating business models enabled by frontier AI.
Because the next wave of AI advantage will not come from using powerful models alone.
It will come from designing the right business models around them.
My published paper: https://infota.org/DOI/z6rCG9.html