This reminds me of when I worked at Moto on the original cellphone. There were many skeptics at the company, but management made a major investment that worked out well, until Nokia took over leadership in the new technology. In the 1980s, no one I knew imagined a future dominated by smartphones. We did imagine widespread cellphone use and GPS guided tractors.

It is not uncommon for tech companies to bet everything on a new uncertain technology. Swim or die. New tech often has unexpected applications. And there can be major changes in which companies lead new tech industries. All of which makes projections and spreadsheets somewhat useless. Instead, focus on if the new tech is on the bleeding edge (not ready for deployment or no consumer demand), and what alternatives there are for investment (swim or die). AI is clearly ready for deployment with substantial consumer demand. What else would you have Google invest in that has better prospects than AI?

Tech companies do tend to overbuild and push features beyond what consumers want. Maybe that is what is happening today with AI data centers. But the AI chips are not fragile. They have a limited number of compute cycles. If the chips are not used, they will still be good to go when the demand picks up. I'm sure Google can do the ROI math that is common with new engineering projects. Like cellphones, AI is here to stay. Maybe some other company will replace today's leaders.