SpaceX, Nvidia designed a space-optimized Vera Rubin NVL72 (orbit launch in Q4) | Hacker News Reader

SpaceX, Nvidia designed a space-optimized Vera Rubin NVL72 (orbit launch in Q4)

twitter.com

8 pointstosh16 comments

but this is all a coordinated scam to pump up stock prices

(end sarcasm)

That's true of many press releases, yes.

no, the whole concept of orbital datacenters

I continue to fail to see how this design solves the issues that can only be solved by breaking the laws of physics in cooling and shielding.

Is it possible that your understanding of the problems is not quite on par with the best engineers in the world?

Nothing that the best materials scientist in the world can do will ever break the laws of thermodynamics so I’d say it’s much more likely that Elon is getting played like countless times evidenced by the many times he’s talked out of his ass about things he didn’t understand.

I'm assume you would have said the thing about an orbital shell of 10k satellites with laser links for global internet. Or a fully reusable 2nd stage made of steel.

No, microsatellites have been a thing long before SpaceX developed them. Maybe you don't understand physics, or can you explain exactly how SpaceX plans to deploy a solar array of sufficient size to power a data center with the necessary radiators to have meaningful compute?

I'm not going to educate you, just google it. It's not some new physics. Starlink aren't microsats. There's never been a large laser-linked shell. Everything you say is kind of wrong.

I googled it extensively way before this conversation and have researched it for a while. There's no economic sense to the size of the cooling arrays needed and the shielding involved and that's simple matter of how black-body radiation works.

Anyone who told you that any laws of physics are being broken is lying to you.

What's the surface area of radiators that this thing is going to need? The cost of compute in space is determined by the energy that must be gathered and dissipated. What about micrometeorites? There's nothing economically viable or meaningful about running a data center in space.

Lol, go ahead and short SpaceX, great opportunity for you if you really believe this.

Precisely the reason I would never invest in SpaceX is because Musk is pathological liar that's full of shit and their investors don't live on a world where stock pricing is based on fundamentals.

There's literally a slide for those questions on the website. https://www.spacex.com/spacexai/starmind

160 square meters, which if ran in optimum conditions would reject between 100 to 150 kilowatts.

Micrometeorites is not a significant concern. There aren't that many of them and 100% reliability is not required. If one is hit then you end of life it or de-rate it (i.e. shut down some of the processors or underclock them) appropriate to the lost solar energy or lost radiating power.

Server clusters already replace racks every 5 years or so and dispose of them. (It used to be ever 3 years.) How is this any different?

Don’t post his trash here, please.

Reply on news.ycombinator.com