Transcription
The four big tech companies are spending hundreds of billions of dollars on data centers this year. They're spending twice what they spent last year, and it's roughly the size of some countries' entire economies.
But the thing is, we could stop today, and the world would actually be a better place. I'm being serious. We have more than enough advanced compute out there right now that we don't need to lift one more shovel. And that's because the models already exist. They already run. We have enough. Thank you very much. Now, please piss off.
So, in this video, I want to explain in simple terms what's wrong with these data centers and why you should feel very upset about it. As a former accountant, let me put my finance hat on for a minute, because this is really ugly. To fund this data center pissing contest, the hyperscalers are taking on massive amounts of debt. Around a trillion dollars and half of it over the next few years. And the investors are handing them this money expecting a massive payday. And they could not care less about the social, economic, or environmental implications. This is capitalism on steroids, and it's reminiscent of other massive build-outs that have occurred in history.
You see, in a normal business, you'd be looking at investing capital at roughly the same rate as your depreciation. That way, your invested capital base stays around the same number. In some years, it might go higher. In some years, it might go lower. But for an established business with investors and shareholders, there is an expectation that the company will behave in a reasonably consistent and rational way. But that's not what's happening in the big data center world. And that's because it's fueled by pure greed. By some of the worst humans on the planet who are going for broke, quite literally.
Do you remember that climate change thing? You know, when we were all worried about cooking the planet and causing trillions of dollars of economic damage, food security, all that sort of stuff? But yeah, but then AI came along and, you know, hold on, baby. I got to get me some cat videos. >> Ranking the craziest AI cat videos. And this is the problem. The technology that investors are putting their money into hasn't delivered any meaningful social benefits.
Data centers now draw around 6% of all electricity in the United States, and it's heading up towards 12% by the end of the decade. Households electricity prices across the country have jumped almost 10% in a single year. This investment is directly impacting ordinary people, and the upside is all going to the richest companies on the planet. The the cost goes to a family paying more to run their fridge, and that is a gross misallocation of investment capital.
This is the bit that most people have missed. Most of the AI work in a business is actually not that hard. Businesses are not trying to fold proteins or work out how to talk to dolphins. Most businesses are looking for basic basic stuff. Summarize this, classify that, pull that number, draft a reply, read the document. This is boring, high-volume, routine work that makes up the vast majority of enterprise AI use cases. And for that, you do not need a trillion parameter model sitting in a data center to do any of that. And that's because there's an enormous amount of energy wasted in moving data from your computer to a data center, processing it, and sending it back to you. Most people only will ever need a small capable model running on a device that they already own. The chips in those devices now ship with dedicated AI silicon as standard. And the small models that do the routine work near instant speed are up to 90% lower cost, and the data leaves never leaves your building.
You know, big tech promised all this amazing stuff, and none of it has arrived. Where are the drug discoveries or the massive technology advantages that address poverty or inequality or disability? We got none of that, just cat memes and job losses. That is seriously up. >> [screaming] >> Now imagine if the same amount of money that was spent building these data centers was poured into startup companies building innovative on device technology. Imagine the problems that we could have already solved with this that could really have materially improved the quality of people's lives.
I'm making this video because this cloud AI data center rollout is being treated as inevitable and governments are bending over backwards to please these tech giants. They're expediting approvals and bypassing all the usual community consultation. And that's because this rollout is being treated as inevitable and it's not. We need to push back. We need to ask questions. Nobody should be forced into making these decisions. The cloud AI model is not the only way that we can use this technology and none of us need be funding the most expensive bet in corporate history. We need to stop with the data centers right now. I mean, stop. Just tools down. Enough is enough. >> [laughter] >> THANKS FOR WATCHING. I'M DR. ERROL BRAUN. IF YOU ENJOY my content, please like and subscribe. Feel free to connect with me on LinkedIn and please [music] share your thoughts in the comments below. Thanks again and I'll see you in the next one.