ai PM market analysis — 2026-10-15

The contest in artificial intelligence is shifting from raw model capability toward control of ecosystems and the infrastructure beneath them. UK competition analysis points to a risk that a small group of incumbents could dominate critical inputs such as compute, data, expertise and funding, even as model software itself becomes more open.

Economic research on open foundation models suggests firms are willing to accept weaker near-term monetisation in exchange for developer adoption, downstream customisation and a broader ecosystem that can pay off later. Stanford HAI’s work on responsible openness reinforces that case, arguing wider release can improve scrutiny and participation, though only where release practices include real safeguards rather than openness for its own sake.

Smaller, more specialised models also widen the field by lowering deployment demands and supporting more on-device use, which cuts against the idea that scale alone decides the market. Taken together, the sourced material describes a market where openness can broaden participation at the model layer while the underlying inputs that make any model useful remain concentrated among a few well-resourced players.

That split keeps the overall read cautious. Openness may widen who can build on top of foundation models, but it does not by itself resolve the access questions around compute, data and funding that the competition analysis treats as the more durable constraint.