Strategist warns AI investors are backing the wrong future

Joachim Klement said Wednesday that investors are focusing on the wrong future by prioritizing large language models housed in massive data centers [1, 2].

This perspective challenges the current market trajectory, where billions of dollars are flowing into the physical infrastructure required to sustain giant AI models. If the industry shifts toward smaller, local models, the projected growth for data center operators could face a significant disruption.

Klement, a managing director at Panmure Liberum, said the matter during an appearance on CNBC Television [1, 2]. He said investors are "investing in the wrong future" by focusing on large language models running in massive data centers ^[1].

The current AI boom is largely built on the assumption that larger models require ever-increasing amounts of computing power and centralized storage. This has led to a surge in the construction of expansive data centers to handle the processing loads of these systems.

However, Klement said that the emergence of smaller AI models, which can run locally on individual devices rather than in the cloud, could upend this trend [1, 2]. Localized AI would reduce the reliance on centralized infrastructure, potentially slowing the demand for the massive facilities currently being built [1, 2].

By concentrating capital on large-scale language models, Klement said the investment community may be missing an emerging opportunity [1, 2]. The shift toward local AI would represent a fundamental change in how artificial intelligence is deployed and monetized across the global economy.