If you attack a party for something well in advance of an election, it loses its effectiveness before election day. People get desensitized to the message and it no longer impacts their decision to vote or who to vote for. If you noticed during the last election, different promises were made at different times. They waited to promise full legalization of cannabis until the last minute for a reason. There are also alignments of promises or points emphasized during primaries based on which state primary or primaries are up next.

deleted by creator

It takes time to build out capacity to increase HDD production, but when operating margins are approaching 30%, somebody is going to want cash in. If the new reality is that every company wants to harvest and retain as much data as possible at all times in case it has potential as training data at some point in the future, then additional capacity will eventually be built to tap into that.

I know it sucks if you need storage now, but I think the smart decision is to hold off until the market stabilizes. You might even get lucky; the bubble might burst and the market will be flooded with cheap excess HDDs.

Ford’s $30,000 EV truck is supposed to be here fairly soon as well, although it looks like that will only be available as a crew cab, sadly.

Homemade ranch with fresh herbs, garlic, buttermilk and fish sauce is amazing.

You wouldn’t, but that’s what governments are effectively asking be done, lending validity to the analogy.

The article is about AWS data centers not AI compute specifically. That’s the cloud infrastructure for a lot of the web. The comparison to agriculture is about scale, not the importance of the Internet vs. agriculture. Using 0.011% of the water as the farms in Kansas to power 25% of the Internet is nothing.

The average for farm land is around 500,000 gallons per acre per year. Using the same amount of water as 5,000 acres of farmland to run >25% of the nation’s cloud resources isn’t enormous. Not that they couldn’t do better.

It is still far from commercially viable yet.

Solid-state sodium is still in the laboratory stage. People assumed Donut was claiming to have developed a solid-state sodium battery due to their “no lithium” statement, but they never specifically claimed they were using sodium.

All solid-state lithium is a bit further along. Korea has pilot plants producing full-sized EV batteries that are being used for testing before they do the final scale up to production. Chinese manufacturers are also basically at the same stage. Those will likely be available in production EVs by 2030.

I don’t want it to happen, but I don’t see any real scientific reason why machine learning won’t eventually subsume jobs at a faster rate than new jobs for humans are created. That might take 15 or 30 years to really hit, but it’s certainly something that needs to be planned for extensively.

Gurning.