AI's Innocence Lost: How Capital Will Force Us to Get Real About Value

We are living through the biggest freemium product demo in history, and it cannot last. AI capability is advancing at breakneck speed, but the capital being committed vastly outweighs the value being realised, and aspirations are not cashflows.

This Sifted piece argues that 2026 marks the end of AI's era of innocence, as Big AI is forced to monetise harder and the question of who pays, how much and for what can no longer be deferred. The piece examines how monetisation will likely come through control of attention and direction, through B2B stack wars between cloud providers, model providers and integrators, and through regulation that will determine not just safety but the division of power and profits across the economy.

Meanwhile, the aggregate returns still do not obviously cover what is being built upstream, and organisations are discovering that AI amplifies rather than fixes the structural problems already present in their systems.

Read here.