When Fear Becomes a Moat: Is the AI Industry Fighting Its Own Commoditisation ?

The recent debate around restrictions on Anthropic’s advanced AI models immediately created the expected questions. Is artificial intelligence becoming too powerful? Should governments control access? Are frontier models becoming geopolitical assets?

Those questions matter. But another strategic question may be even more important: what happens if the greatest threat to frontier AI companies is not that their models become dangerous, but that their models become ordinary?

Technology markets have a recurring pattern: creating the breakthrough does not always mean capturing most of the value created by that breakthrough.

The companies building foundational infrastructure are not always the companies owning the economic relationship. Telecom operators built the networks that enabled the internet economy, but platforms captured much of the value generated on top of those networks.

Hardware companies enabled digital ecosystems, while software and service layers often captured the customer relationship.

The strategic question is rarely only “who invented the technology?” It is “who controls the bottleneck?”

Artificial intelligence may now be approaching this moment.

The uncomfortable question behind frontier AI valuations

Today, frontier AI companies represent some of the most strategically important organisations in technology. But their long-term economic position in the AI value chain remains unresolved.

Training frontier models requires enormous investment. Compute costs remain significant. Competition is accelerating. Open models continue improving. Enterprises increasingly want flexibility rather than dependency on a single model provider.

This creates an uncomfortable question: if intelligence becomes increasingly abundant, where does scarcity remain?

Because markets do not reward availability. They reward control.

The battle for the AI value chain

The AI economy is not one market. It is a stack of competing strategic layers.

Chip companies want compute to remain the scarce resource. Cloud providers want infrastructure to become the control point. Frontier AI companies want intelligence itself to remain the strategic layer. Enterprise software companies want models to become replaceable engines powering workflows they control.

Everyone is moving toward the same objective: owning the bottleneck.

This is why the current AI race is not only technical. It is economic.

The strategic value of being considered dangerous

This is where the Anthropic debate becomes more interesting. A restriction driven by security concerns can carry two different messages. The immediate interpretation is: “this technology requires oversight.” But the strategic interpretation is also: “this technology is too important to be treated as ordinary software.”

That distinction matters.

Commodities compete on availability and price. Strategic infrastructure competes on trust, dependency and control.

Being perceived as critical can reinforce strategic positioning.

The race to avoid becoming invisible

The nightmare scenario for frontier AI companies is not that nobody uses their models. It is actually the opposite.

Everyone uses AI, but the model disappears underneath applications, platforms and enterprise systems. The model becomes the engine. Someone else owns the dashboard. Someone else owns the workflow. Someone else owns the customer.

In that scenario, creating intelligence does not automatically mean capturing the value created by intelligence.

The first AI race was about capability: who could build the most advanced model?

The second AI race may be about economic control: who captures the value created by those models?

The future competition may not simply be OpenAI versus Anthropic versus Google. It may involve every layer of the AI ecosystem competing to become the unavoidable layer.

The defining question of the next decade may not be: who builds the smartest intelligence?

It may be: who prevents intelligence from becoming a commodity?

Because in technology markets, creating the revolution is only the first battle. Owning the value created by the revolution is the second.