Waste heat from data centers can boost air temperatures in downwind neighborhoods by as much as 4 degrees Fahrenheit, researchers at Arizona State University report in a new study conducted in the Phoenix metro area, the hottest in the U.S.
YouTube requires enormous amounts of data center hardware. Do you have any idea how much video data they injest, transcode, and distribute in even just an hour or a minute? It’s not very economical to run actually and they had major problems even breaking even.
Most large Lemmy instances run on cloud services nowadays. So they rely on a datacenter somewhere. I am well aware you can self-host - I have done so for multiple things including AI models and tools - but that’s a minority of Lemmy users. Most are using public instances which are hosted on servers in the cloud. In fact if you read the deployment guide for Lemmy it is meant to be deployed using cloud native technology.
You also don’t need a whole Stargate to host one users AI usage. You don’t even need a whole server. Typically AI servers process requests from multiple users simultaneously. The actual marginal cost of each user is relatively small, hence why people can and do run local AI models.
YouTube requires enormous amounts of data center hardware. Do you have any idea how much video data they injest, transcode, and distribute in even just an hour or a minute? It’s not very economical to run actually and they had major problems even breaking even.
Most large Lemmy instances run on cloud services nowadays. So they rely on a datacenter somewhere. I am well aware you can self-host - I have done so for multiple things including AI models and tools - but that’s a minority of Lemmy users. Most are using public instances which are hosted on servers in the cloud. In fact if you read the deployment guide for Lemmy it is meant to be deployed using cloud native technology.
You also don’t need a whole Stargate to host one users AI usage. You don’t even need a whole server. Typically AI servers process requests from multiple users simultaneously. The actual marginal cost of each user is relatively small, hence why people can and do run local AI models.