On 3/16/26 15:19, kanze wrote: > To put things into perspective : > Global Electricity Breakdown > Industry 42% ~13000 TWh > Steel, chemicals, heavy manufacturing > Buildings (Residential) 25% ~8000 Fridges, > lighting, heating, TV > Buildings (Commercial) 22% ~7000 HVAC, > elevators, street lighting > Transportation 5% ~1600 > Data Centers & AI 4% ~1200 > > The Commercial buildings can be separated in categories : > Office Economy 30%, Retail & Service 20%, Healthcare 12%, Education 10%, Food > & Hospitality 13%, Warehouse/Other 15% > > Replacing half of the office economy with AI, while slightly optimizing the > rest, would result in a net energetic gain. And I'm talking about this > commercial category only.
We were already in a state where the office economy had drastically been reduced. During the pandemic everyone had switched to remote work, we found that in-person work for a large portion of "desk jobs" was simply not needed. However, it turns out that all it takes is incentives from local municipalities concerned about lunch hour traffic in downtown areas to push companies to get folks to the office. Now there is something to be said about each individual keeping their home a bit more cool instead of the office, but the point is that these decisions are not dominated by what is a greater good, it's driven by capital. Look at the waste of electricity on crap like crypto shit coins. Your electricity use can be objectively wasteful but as long as someone can speculate on it, we will gladly allow it to propagate. > > It will happen, and not happen, at the same time: Jevons Paradox. > > Now I hear you on the 'mundane reality' of tech, but look at the corporate > side: if you offered these companies a more efficient chip, they’d beg for > it. They’re stuck in a loop of aggressive, short-term tactics to lock in > users, but that’s not a sustainable path. They might be chasing 'cat video' > traffic now, but in the near future, the sheer cost of power will force them > to pivot toward radical energy efficiency. So let me get this straight. You think that the power prices will increase to a point that will price out mundane direct consumer use cases? So we're in agreement then that energy prices will raise now? Have you considered the impact that would have on industrial electricity use? At the point that rates increase so much that the mundane use gets priced out who knows how many nth order impacts we'd get from that. As we've seen from a number of recent geopolitical events the boat of the global supply chain is not nearly as resistant to shaking as we thought. If the cost of all process intensive chemicals and welding 50x'd we'd be in for a very rough time. Seems like a lose-lose, either we see prices skyrocket and disrupt actual useful uses of electricity, or we end up wasting huge amounts of power on bullshit mundane AI usage. If seems that you're hopeful that capital incentives will eventually align with what is better for society as a whole. I think this is really quite rare, and seemingly even more rare within the world of tech. AI is already not in that camp, clearly. The incidental outcome of this actually working out in our benefit seems less likely then the possibility that all these utopian future pipe dreams are made up lies (or delusions) created by Nvidia bag holders. ------------------------------------------ 9fans: 9fans Permalink: https://9fans.topicbox.com/groups/9fans/Tacc477007dd06ec0-Me96699711eabdf5886edc595 Delivery options: https://9fans.topicbox.com/groups/9fans/subscription
