Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Anthropic is expanding to Colossus2. Will use GB200

twitter.com

Post reply on HN
Print view View original

351–360 of 367 posts

Re: Anthropic is expanding to Colossus2. Will use GB200

#351

Post by _heimdall »

bvcp wrote: post #348
_heimdall wrote: post #206

Earlier quoted context omitted.

The physics of data centers in space will be extremely difficult and expensive to pull off in any meaningful timeline. I fully expect "space AI" to be about as realistic as the flying cars and hover boards we've been promised since televisions were black and white.

its only a expensive to viable alternatives, is there a viable alternative for 1 terrawatt of compute on earth? how could you get approvals for such a project

The alternative would be just not building it.

I'm not sold that a data center in space is even possible unless and until we develop a drastically different solution for heat management.

Re: Anthropic is expanding to Colossus2. Will use GB200

#352

Post by _heimdall »

irq-1 wrote: post #329
jtr1 wrote: post #295

Earlier quoted context omitted.

You may want to update your priors [1][2][3] 1. https://www.carbonbrief.org/guest-post-how-solar-panels-and-... 2. https://www.fastcompany.com/91500104/google-minnesota-data-c... 3. https://www.rootsanalysis.com/sodium-ion-battery-market

https://www.fastcompany.com/91500104/google-minnesota-data-c... > The new plant in Minnesota will be big enough to deliver 300 megawatts of power and store an enormous 30 gigawatt-hours of energy, making it the largest battery by capacity that’s been announced so far. By comparison, that’s more storage than all of the battery projects built in the U.S. in 2024 added together.

That's to power a single data center though, how would that scale?

If I'm doing the math right, Minnesota used 65.7 TWh of power last year so to store 3 days worth of power just for that one city we would need a battery 18x larger than the one mentioned here.

I can't imagine we could ever scale such storage capacity for all energy use, let alone all the wind and solar required to fill it.

Re: Anthropic is expanding to Colossus2. Will use GB200

#353

Post by vel0city »

jpadkins wrote: post #298
sulam wrote: post #294

Earlier quoted context omitted.

Tl;dr — but it’s too heavy You’re making some decent points here, but you’re either forgetting or ignoring the major thing that people usually neglect to mention when they want to make a case for this (crazy) idea — weight. Unless SpaceX is going to completely redesign hardware such that it is optimized entirely for its mass, it requires many ( many ) launches to even get a small set of racks into space. I don’t norm…

Have you calculated the CO2 output of Terran datacenters run on natural gas vs. space datacenters run on solar? The launch CO2 usage is one time, the datacenter energy is for the life of the equipment.

If solar+batteries are so cheap and viable to power giant 100 megawatt datacenters in space, why not do the same on Earth instead of using natural gas generators?

Re: Anthropic is expanding to Colossus2. Will use GB200

#354

Post by jpadkins »

vel0city wrote: post #353
jpadkins wrote: post #298

Earlier quoted context omitted.

Have you calculated the CO2 output of Terran datacenters run on natural gas vs. space datacenters run on solar? The launch CO2 usage is one time, the datacenter energy is for the life of the equipment.

If solar+batteries are so cheap and viable to power giant 100 megawatt datacenters in space, why not do the same on Earth instead of using natural gas generators?

Solar panels on surface get ~70% less energy than in space once you factor in atmosphere absorption, day/night, weather, etc. Solar + battery on the surface is not economically viable versus nat gas generators (excluding externalities)

Re: Anthropic is expanding to Colossus2. Will use GB200

#355

Post by rafram »

47282847 wrote: post #345
rafram wrote: post #252

Earlier quoted context omitted.

It does, though.

Yeah. Well. I guess as a European I want the legal system to be robust in terms of protecting the citizen’s interests against corporations. I can see how you can also want robustness in terms of protecting corporate interests against citizens, so, yes, the statement remains technically true in terms of robustness. In terms of justice, again as a European, I think of Justitia, the Roman Godess, who judges based on “id…

None of those have anything to do with robustness.

Re: Anthropic is expanding to Colossus2. Will use GB200

#356

Post by 47282847 »

rafram wrote: post #355
47282847 wrote: post #345

Earlier quoted context omitted.

Yeah. Well. I guess as a European I want the legal system to be robust in terms of protecting the citizen’s interests against corporations. I can see how you can also want robustness in terms of protecting corporate interests against citizens, so, yes, the statement remains technically true in terms of robustness. In terms of justice, again as a European, I think of Justitia, the Roman Godess, who judges based on “id…

None of those have anything to do with robustness.

Huh?

Cambridge Dictionary: “the quality of being strong, and healthy or unlikely to break or fail”

I provided two dimensions in which I consider the US justice system to have failed. I can provide more. You of course can prioritize different dimensions, but that doesn’t invalidate that I am also talking about robustness. Robustness is not an absolute, but always an adjective - related to a specific aspect.

Re: Anthropic is expanding to Colossus2. Will use GB200

#357

Post by einsteinx2 »

kibibu wrote: post #349
einsteinx2 wrote: post #310

Earlier quoted context omitted.

I keep hearing this statement, and it always makes me wonder if people have actually used Grok… I have a Claude Max plan I use for coding, but I also have a Grok Lite plan I use for web search type tasks (similar to Perplexity) because I like how the Grok harness handles searches and I don’t need a SOTA model for that use case. I’d never pay $30/mo for a full SuperGrok account but to me it’s worth the $10/mo for Lite…

Did you miss all the Mechahitler, woke mind virus, white genocide, Musk could beat Tyson, stuff? Its plausibly not in the API and only on the twitter bot, but I see no reason to trust x.ai given this history of obvious manipulation.

I don’t really use Twitter so I’ve never used it via the bot, I’ve only ever used it via the web app.

I bounced back and forth between Grok and Perplexity for web search type tasks and at least for the moment am preferring Grok mostly because it seems to perform more searches and check more results per query vs Perplexity and their $10/mo plan covers my usage vs $20 for Perplexity Pro.

However I’m not married to any LLM service and will switch to another one the moment I get better results from it.

At least in my usage I haven’t noticed any obvious bias, but I don’t really search for politically related stuff so maybe I just haven’t seen it.

Re: Anthropic is expanding to Colossus2. Will use GB200

#358

Post by vel0city »

jpadkins wrote: post #354
vel0city wrote: post #353

Earlier quoted context omitted.

If solar+batteries are so cheap and viable to power giant 100 megawatt datacenters in space, why not do the same on Earth instead of using natural gas generators?

Solar panels on surface get ~70% less energy than in space once you factor in atmosphere absorption, day/night, weather, etc. Solar + battery on the surface is not economically viable versus nat gas generators (excluding externalities)

Sounds like the solution is to actually price the externalities of running on-site natural gas generation and suddenly we don't actually need to go to orbit.

Re: Anthropic is expanding to Colossus2. Will use GB200

#360

Post by mitjam »

urams wrote: post #20

Very possible Elon is doing this to make give Anthropic better chances against OAI while he attempts to reshape xAI. Also possible he sees infra as the future of xAI if he really believes in the value of space compute. Hard to see this any of this as anything other than a bearish sign for Grok though.

Interesting that nobody mentioned SAR in this thread, yet.

SAR benefits from space compute by reducing the strain on downlinks (pre-process, detect in space). A large-ish (think 6k sat) constellation of much bigger sats (2-2,5 ton) than current Starlink sats can run inference on board, distribute load to nearby sats, and enable full round-the-clock surveillance of the whole world.

The nominal compute capacity is in ballpark range of a modern AI data center, but it's only about 20% used on average due to duty cycles. An indestructible, global eye.

The big Starship launch vehicle is perfect platform to bring them into orbit, can maybe bring 40 at a time into orbit, so 150 launches for 6k. Maybe even fewer sats are needed, depends mostly on electrical efficiency of the components.

Post reply on HN