Enterprises debate build-or-buy AI Factories
Firms are investing in in-house capabilities to protect differentiation beyond generic API models.
Global enterprises are accelerating investments in artificial intelligence as spending is projected to reach $632 billion by 2028, sharpening a key strategic dilemma: whether to build proprietary AI capabilities or buy standardised solutions.
The debate is intensifying as more companies pursue “AI factories,” a model that goes beyond adopting tools and treats AI as an operational backbone tied to competitive advantage.
In practice, most enterprises are not training large models from scratch due to cost and complexity. Instead, they are using foundational models while developing differentiated layers internally.
“The most common use, and where we see our enterprises in the region making the most progress are things like fine tuning, adapting or extending an existing model such that the output and the contribution is novel and protectable, something that creates a strategic advantage,” said Abhishek Srivastava, Principal at Arthur D. Little Southeast Asia.
That approach is shaping how executives frame capital allocation. “The most important thing for AI investment decisions is which parts to build and which parts to buy,” Srivastava said. The risk, he added, is that dependence on standardised access limits differentiation: “If you were to use generic AI models through APIs alone, it is difficult for enterprises to create competitive advantage.”
As a result, companies are “waking up and investing more and more into developing a clear competitive advantage compared to others in this space by really building capabilities in house,” he said, while emphasising that “not everything is built in house” and that many capabilities are “developed on top of existing foundational models.”
Where AI delivers enterprise value today is still concentrated in productivity and cost efficiency, particularly through automation and removal of manual steps. But firms are also pursuing growth by redesigning workflows and developing new offerings powered by AI.
