Tether CEO Paolo Ardoino envisions a future where AI model training becomes more efficient, reducing the need for millions of GPUs. He emphasizes that relying solely on brute force computing is inefficient. Instead, improved models will lower training costs, with a focus on data acquisition. Additionally, reasoning is expected to move towards local or edge devices, which will help avoid excessive spending on computing power.

PANews reported on January 27 that Tether CEO Paolo Ardoino tweeted that in the future, training powerful models will no longer require millions of GPUs. This approach of relying purely on brute force computing is both inefficient and inelegant. Better models will significantly reduce training costs, but data acquisition is still key. Reasoning will shift to local or edge devices, avoiding wasting hundreds of billions of dollars on pure computing power.

Tether CEO Predicts Shift in AI Model Training Towards Efficiency and Edge Computing