Expensive Energy A Big Worry For The Future Of AI
One of the forgotten risks of AI is the energy capacity situation in the United States.
Many people forget that AI will require immense energy with a hoard of energy-guzzling data centers to facilitate the next tech revolution.
Many consumers have come to realize how the cost of energy has skyrocketed lately and no doubt the interest rates cut next year might turbocharge commodity prices around the globe.
There is an increasingly real chance that Silicon Valley might not be able to afford AI simply because the costs of energy will deem the AI concept unworthy.
Green energy hasn’t developed as fast as many experts once thought and the United States is still very much dependent on fossil fuels to facilitate tech and business in general.
A pressing question that is popping up is whether the United States can deliver the energy capacity that AI chips demand.
The question is hard to dissect because the situation is always changing.
Numbers need to make sense just like how builders build when they think they can sell their houses and apartment for a profit to the end buyer.
The military conflict in Eastern Europe has forced German manufacturing to deindustrialize because producing without that cheap Russian energy is loss-worthy. AI could follow a similar pattern.
The data grid will become strained but by how much is the next most important matter.
A ChatGPT query, on average, requires almost 10 times as much electricity to process as a Google search does.
The rise of generative AI coincides with a heightening of other factors increasing energy demand, from the electrification of transportation and infrastructure to the on-shoring of US manufacturing. Adding yet another acute demand: AI systems need power all the time.
Critics of AI fanaticism point to potential wastefulness and this could end up morphing into a government regulatory quagmire like so many industries that are overburdened by government agency overreach.
If in the case, the energy demands spiral out of control with everyone going the AI route with every country building AI data centers, the exploding costs will mean that tech won’t be able to profit from AI as quick as it wants.
Many analysts are already raising the flag as to whether all these billions poured into AI investments will really pan out or not. AI isn’t free to produce but shares of it are priced as such.
Much of this hot money is migrating into companies that haven’t proven anything or never even turned a profit, look at OpenAI, it started out as a non-profit.
The issue I have is that generative AI is priced to have zero pushback of its revenue trajectory and I do believe that is wrong.
When there is a pullback, it will be deep and sharp even if not long.
I believe that would be a healthy event for AI because the stock shares of AI have gone parabolic when there isn’t much meaningful follow-through to the underlying business models.
On top of that, generative AI is programmed to be ultra-left-leaning on the social spectrum which could cause conflict down the road.
In short, ride up the momentum until the wave crashes, but watch out for the canary in the coal mine which will bring attention to a deep dip in AI shares.
