A prominent AI researcher makes the following statement: 'The primary goal of our work should be to maximize the reasoning capabilities of large language models through inference-time scaling techniques. The associated computational costs are a secondary issue that future hardware advancements will inevitably solve.'

Critically evaluate this statement. In your response, discuss the validity of this perspective and explain why focusing on the efficiency of these scaling methods is, or is not, a crucial research direction for the field.