CEO Nansen Alex Svanevik states that the AI industry bubble may burst as Chinese-origin language models become more efficient and capable of running on simpler hardware. He also highlights the falling prices of H100 and H200 GPU rentals as an indication of structural changes in demand for computing.

In recent years, the dominant assumption has been that the greater the need for artificial intelligence, the higher the value of computing providers. But now, model capabilities are increasing precisely as processing costs begin to decrease, creating a paradox for the supporting sector.

Many believe that AI advancements always require increasingly expensive computing resources. In fact, software innovations are beginning to reduce dependence on the most advanced devices.

Developers are increasingly focused on pursuing processing efficiency rather than simply expanding capacity. The priority is shifting from acquiring more chips to maximizing results from existing resources.

The emergence of more resource-efficient models, the increasing global semiconductor production capacity, and the availability of alternatives outside the dominant ecosystem are driving a new balance between demand and supply of computing.

The industry is moving from a scarcity phase to an optimization phase. Market valuation is no longer solely based on hardware capacity but on the ability to deliver performance at lower costs.

If efficiency continues to improve faster than demand growth, large investments in infrastructure may face lower-than-expected returns.

High-value scarcity persists until efficiency arrives.

The emerging story is that AI competition is shifting from a race to own the largest computing capacity to a race to utilize computing most efficiently.

#AI #ArtificialIntelligence