Ask HN: Is the downside of investing in AI real?
- Every time AI is used, it costs the AI providers money in processing power, electricity, and cooling. To capture market share, they are currently charging less than the actual cost of providing the service, and if prices were to rise, using it would become less profitable.
- Many AI projects yield no results because companies struggle to adapt their workflows, or because they fail to properly measure costs and benefits, despite the technology being highly capable.
- AI equipment and chips become nearly obsolete in just 3 to 4 years, whereas infrastructure takes 10 to 20 years to pay off; furthermore, maintaining this infrastructure can cost more than the profits it is expected to generate.
Therefore, is it wise to carefully evaluate whether integrating AI truly makes sense for companies? In doing so, would their AI investments be more deliberate and efficient, generating both economic and productivity benefits?