Inflation is going up dramatically right now, and may very well lead to rising interest rates in the future. But there's also a counterbalance to all our inflation -- technology. Improvements in technology can cause a lowering of prices. One example of this is what happened after Alphabet (GOOG 0.84%) (GOOGL 0.77%) acquired the artificial intelligence start-up Deep Mind seven years ago. The company gave its AI solution the task of how to cool its many data centers.
In this clip from "The 5," recorded on Oct. 11, Motley Fool contributor Danny Vena describes what happened next.
Danny Vena: One of my favorite stories about a company that used technology to bring down its costs comes from Google, Alphabet of all things, ticker GOOGL. That story is the company back several years ago acquired an AI company called DeepMind. This is one of the foremost AI companies available in the world. What they did was they took the AI that DeepMind was using and they set it to the task of cooling its data centers.
Now, if you think about Google and you think about the number and the massive footprint of data centers in its operation, you can see where even just saving a small amount of power in their data centers could result in big savings. They showed this graph where this is energy that we were using before we used machine learning to regulate the temperature and use different things, whether or not to use the fans, or to use the cooling, or to open dampers or close them, etc. You could actually see that when they turned the machine learning on, there was a significant drop in the amount of energy that they used.
They did that for a period of time, and then they turned the machine learning back off and it jumped right back up again. What they found was that it amounted to a 40% reduction in their amount of electricity needed for cooling their data centers, and that resulted in a 15% reduction in their overall power usage. That was a ridiculous amount of money saved just by taking an AI model and setting it to a complicated task that had, I think they said there were like 120 different variables that they could use when cooling their data centers. They were able to wring a lot of big chunk of change out of this. When you think about the fact that they only paid somewhere in the neighborhood of, if memory serves somewhere between $400 and $600 million for deep learning. Just the savings from their electricity usage alone between then and now probably paid for the company.
