Gloria lives in a bubble, and made the mistake of thinking her extremely comfortable, highly secure bubble was the whole world. That's not surprising. Gloria only moves among other bubble people, from one gated bubble pad to the next, in her bubble transport system, where they don't talk about the turbo-fans and ICE V8's that power it all, or the staggering quantity of power it takes to climate control everything in her bubble world.

That's not new. We're ruled by such bubble folk, indulging their bubble concerns, pursing their moral panics, signaling their virtues, and carefully ignoring all else beyond the bubble.

What's new here is this: the consequences of this have reached the privileged students of our prestigious academic system. Suddenly it's not just the hoi polloi on the shit end of the stick. Johnny Winston-Blake IV is also having his future deleted by the bubble people. And he's mad about it.

That's pretty crazy. 100% US Grade A clown-world right there.

You can only think about eating, don't you.

It's all easy wins. Winning a war is hard. This is just chips, software and data.

The 120kW figure is indeed input power. Thrust is typically quoted in Newtons, not Watts. The input power is useful because it's a proxy for thrust and vastly easier to measure. Ultimately, none of that really matters, however: the real figure of merit for ion engines (all rockets, really) is Specific Impulse. When NASA claims these use 90% less mass for the same total impulse, they're saying it's about one order of magnitude more propellant-efficient than a chemical rocket.

Who cares. This stuff is still in its infancy, and new algorithms and new hardware is going to collapse all of this to commodity level value anyhow. A few years from now you'll buy a GTP-5.x/Mythos equivalent in a box for gaming console money.

If you buy a platform from a company that once infected their customer's PCs with a rootkit, you deserve whatever Sony choses to inflict on you.

Enjoy.

The argument will be that its not a bank, (I mean, who wants to trust those things?), and the money is merely travelling in a conveyance

I think they're doing well, given the givens. Panther Lake launched on time, and they're actually shipping in quantity. 18A is showing a typical yield ramp rate, getting to 60-75% (depending on who you read) yield now, and probably into 80-90% in 2026.

They're actually executing. Feel free to beat them over the head about past failed promises, if that makes you happy. 18A is a GAA + backside power design that this actually a bit ahead of the industry curve. They deserve some credit.

Musk understands he needs to jump start his fab: he doesn't care to take the time to build everything from scratch. This is a licensing deal that gets Musk into the fab business quickly. This is also good deal for Intel: it means Intel has brand new, well capitalized customer (see SpaceX IPO in 2026) that can help get 14A working. Intel captures all the Musk fab money, shutting out TSMC, Samsung, etc., and ends up with a working 14A node for its own markets. It also makes Intel look more credible to other 14A customers, being willing to collaborate on and sign deals with partners on fab tech.

While DRAM isn't exactly slow, SRAM access time is about one order of magnitude faster than DRAM. That's why it's used for CPU caches, TLBs, registers, etc. SRAM is more power efficient, as it doesn't need refresh. The downside is die area: an SRAM flip-flop bit much larger than a gate+cap DRAM bit.

SRAM has never been built at this scale, afaik. Cerebras was ahead of the curve here, building wafer scale SRAMs years ago. The penalties of DRAM (even with HBM) are now so severe that everyone is taking the gloves off and building mighty SRAMs. This has always been possible in theory, but the high cost never justified it.

The impact on semiconductor fab demand is significant. SRAM cells are larger than DRAM bits: more silicon die area for the same amount of RAM.

Also, the training vs. inference split Google is baking into actual hardware is a big deal: it's the reality that training and inference are very distinct things asserting itself, which has been obvious to anyone that hasn't been drinking excessive NVidia cool-aid: there is a future where costly, general purpose GPU-like devices aren't actually necessary for operating LLMs.

It's a click bait headline. This is a $10 billion option, not a $60 billion acquisition.

$10 billion is still on the insane side for a vscode extension. However, Elon has an AI platform, and that platform lacks the IDE integration that others have, so my guess is he's looking to plug that hole with money.

When you couple all that with the recent "Terafab" kickoff, it's clear Elon wants his whole AI compute stack under one roof; from the chips to the developer stack. He's building a vertically integrated AI platform.

He's doing all that because he's convinced solar powered space compute is the answer, and will make him billions. He's been right often enough that I'm not betting against it, but it's a big bet. He won't die a pauper either way, so why not?

Weasel words.

Cite your "evidence."

I dont see an issue with that, as UK schools also ban a lot of other things during "free time" (its not actually time without restriction), for example leaving the school grounds for most of the school body (when you get into sixth form, you gain more freedoms as you are deemed to be there voluntarily).