zlacker

[parent] [thread] 33 comments
1. nwelli+(OP)[view] [source] 2026-02-03 03:34:32
Yeah it does not make a whole lot of sense as the useful lifespan of the gpus in 4-6 years. Sooo what happens when you need to upgrade or repair?
replies(7): >>Lalaba+Z2 >>gricar+n3 >>tgsovl+94 >>superb+v4 >>mandee+l9 >>rlt+Uc >>pantal+Tp
2. Lalaba+Z2[view] [source] 2026-02-03 03:59:42
>>nwelli+(OP)
This is a question that analysts don't even ask on earnings calls for companies with lowly earthbound datacenters full of the same GPUs.

The stock moves based on the same promise that's already unchecked without this new "in space" suffix:

We'll build datacenters using money we don't have yet, fill them with GPUs we haven't secured or even sourced, power them with infrastructure that can't be built in the promised time, and profit on their inference time over an ever-increasing (on paper) lifespan.

replies(1): >>acchow+14
3. gricar+n3[view] [source] 2026-02-03 04:03:34
>>nwelli+(OP)
not to mention that radiation hardening of chips has a big impact on cost and performance
replies(1): >>parine+Ui
◧◩
4. acchow+14[view] [source] [discussion] 2026-02-03 04:09:34
>>Lalaba+Z2
> This is a question that analysts don't even ask

On the contrary, data centers continue to pop up deploying thousands of GPUs specifically because the numbers work out.

The H100 launched at $30k GPU and rented for $2.50/hr. It's been 3 years since launch, the rent price is still around $2.50.

During these 3 years, it has brought in $65k in revenue.

replies(3): >>hdjrud+T5 >>fauige+lx >>kd913+691
5. tgsovl+94[view] [source] 2026-02-03 04:12:17
>>nwelli+(OP)
> Sooo what happens when you need to upgrade or repair?

The satellite deorbits and you launch the next one.

replies(2): >>youngt+ZI >>jeroje+xc1
6. superb+v4[view] [source] 2026-02-03 04:15:30
>>nwelli+(OP)
With zero energy cost it will run until it stops working or runs out of fuel, which I'm guessing is between 5-7 years.
replies(1): >>tactic+Gg
◧◩◪
7. hdjrud+T5[view] [source] [discussion] 2026-02-03 04:27:51
>>acchow+14
They can run these things at 100% utilization for 3 years straight? And not burn them out? That's impressive.
replies(2): >>vlovic+Hd >>imtrin+tt
8. mandee+l9[view] [source] 2026-02-03 04:58:24
>>nwelli+(OP)
> the useful lifespan of the gpus in 4-6 years. Sooo what happens when you need to upgrade or repair?

Average life of starlink satellite is around 4-5 years

replies(2): >>lesost+Pt >>imglor+V71
9. rlt+Uc[view] [source] 2026-02-03 05:37:25
>>nwelli+(OP)
A "fully and rapidly reusable" Starship would bring the cost of launch down orders of magnitude, perhaps to a level where it makes sense to send up satellites to repair/refuel other satellites.
◧◩◪◨
10. vlovic+Hd[view] [source] [discussion] 2026-02-03 05:43:30
>>hdjrud+T5
Not really. GPUs are stateless so your bounded lifetime regardless of how much you use them is the lifetime of the shitties capacitor on there (essentially). Modulo a design defect or manufacturing defect, I’d expect a usable lifetime of at least 10 years, well beyond the manufacturer’s desire to support the drivers for it (ie the sw should “fail” first).
replies(1): >>mike_h+7D
◧◩
11. tactic+Gg[view] [source] [discussion] 2026-02-03 06:10:25
>>superb+v4
5 to 7 months given they want 100kw Per ton and magical mystery sauce shielding is going to do shit all.
◧◩
12. parine+Ui[view] [source] [discussion] 2026-02-03 06:31:30
>>gricar+n3
You could immersion cool them and get radiation resistance as a bonus.
replies(1): >>bigblu+ln
◧◩◪
13. bigblu+ln[view] [source] [discussion] 2026-02-03 07:09:51
>>parine+Ui
Yes, because launching then immersed in something that will greatly increase the launch weight will help...
14. pantal+Tp[view] [source] 2026-02-03 07:30:59
>>nwelli+(OP)
Same that happens with Starlink satellites that are obsolete or exhausted their fuel - they burn up in the atmosphere.
◧◩◪◨
15. imtrin+tt[view] [source] [discussion] 2026-02-03 08:01:46
>>hdjrud+T5
I don't see anything impressive here?
◧◩
16. lesost+Pt[view] [source] [discussion] 2026-02-03 08:04:06
>>mandee+l9
damn. at this point its not even about a pretense for progress, just a fetish for a very dirty space
replies(2): >>nolok+zz >>MPSimm+971
◧◩◪
17. fauige+lx[view] [source] [discussion] 2026-02-03 08:30:54
>>acchow+14
Beyond GPUs themselves, you also have other costs such as data centers, servers and networking, electricity, staff and interest payments.

I think building and operating data center infrastructure is a high risk, low margin business.

◧◩◪
18. nolok+zz[view] [source] [discussion] 2026-02-03 08:48:18
>>lesost+Pt
It's essentially a military network (which is why other power sphere want their own) and a way to feed money into spacex
replies(1): >>infini+YC5
◧◩◪◨⬒
19. mike_h+7D[view] [source] [discussion] 2026-02-03 09:15:43
>>vlovic+Hd
The silicon itself does wear out. Dopant migration or something, I'm not an expert. Three years is probably too low but they do die. GPUs dying during training runs was a major engineering problem that had to be tackled to build LLMs.
replies(1): >>Majrom+Yg1
◧◩
20. youngt+ZI[view] [source] [discussion] 2026-02-03 09:58:32
>>tgsovl+94
Such a waste of resources
◧◩◪
21. MPSimm+971[view] [source] [discussion] 2026-02-03 12:55:38
>>lesost+Pt
They re-enter and burn up entirely. Old starlinks don't stay in space.
replies(1): >>youngt+1o1
◧◩
22. imglor+V71[view] [source] [discussion] 2026-02-03 13:01:21
>>mandee+l9
Starlink yes, at 480 km LEO. But the article says "put AI satellites into deep space". Also if you think about it, LEO orbits have dark periods so not great.

A better orbit might be Sun Synchronous (SSO) which is around 705 km, still not "deep space" but reachable for maintenance or short life deorbit if that's the plan. https://science.nasa.gov/earth/earth-observatory/catalog-of-...

And of course there are the LaGrange points which have no reason to deorbit, just keep using the old ones and adding newer.

◧◩◪
23. kd913+691[view] [source] [discussion] 2026-02-03 13:10:52
>>acchow+14
They worked out because there was an excess of energy and water to handle it.

We will see how the maths works out given there is 19 GW shortage of power. 7 year lead time for Siemens power turbines, 3-5 years for transformers.

Raw commodities are shooting up, not enough education to cover nuclear and SMEs and the RoI is already underwater.

replies(1): >>Sketch+En1
◧◩
24. jeroje+xc1[view] [source] [discussion] 2026-02-03 13:32:01
>>tgsovl+94
so, instead of recycling as many components as possible (a lot of these GPU have valuable resources inside) you simply burn them up.

I'm guessing the next argument in the chain will be that we can mine materials from asteroids and such?

◧◩◪◨⬒⬓
25. Majrom+Yg1[view] [source] [discussion] 2026-02-03 13:57:08
>>mike_h+7D
> GPUs dying during training runs was a major engineering problem that had to be tackled to build LLMs.

The scale there is a little bit different. If you're training an LLM with 10,000 tightly-coupled GPUs where one failure could kill the entire job, then your mean time to failure drops by that factor of 10,000. What is a trivial risk in a single-GPU home setup would become a daily occurrence at that scale.

◧◩◪◨
26. Sketch+En1[view] [source] [discussion] 2026-02-03 14:32:57
>>kd913+691
My cynical take is that it'll works out just fine for the data centers, but the neighbouring communities won't care for the constant rolling blackouts.
replies(1): >>kd913+no1
◧◩◪◨
27. youngt+1o1[view] [source] [discussion] 2026-02-03 14:34:47
>>MPSimm+971
So they pollute the upper atmosphere instead!
replies(1): >>youngt+WN4
◧◩◪◨⬒
28. kd913+no1[view] [source] [discussion] 2026-02-03 14:36:49
>>Sketch+En1
Okay but even in that case the hardware suffers significant under utilisation which massively hits RoI. (I think I read they only achieve 30% utilisation in this scenario)
replies(1): >>Sketch+Uz1
◧◩◪◨⬒⬓
29. Sketch+Uz1[view] [source] [discussion] 2026-02-03 15:29:21
>>kd913+no1
Why would that be the case if we assume the grid prioritizes the data centers?
replies(1): >>kd913+JB1
◧◩◪◨⬒⬓⬔
30. kd913+JB1[view] [source] [discussion] 2026-02-03 15:37:57
>>Sketch+Uz1
That is not a correct assumption. https://ig.ft.com/ai-power/

Reports in North Virginia and Texas are stating existing data centres are being capped 30% to prevent residential brownouts.

replies(1): >>Sketch+8E1
◧◩◪◨⬒⬓⬔⧯
31. Sketch+8E1[view] [source] [discussion] 2026-02-03 15:46:31
>>kd913+JB1
That article appears to be stuck behind a paywall, so I can't speak to it.

That's good for now, but considering the federal push to prevent states from creating AI regulations, and the overall technological oligopoly we have going on, I wonder if, in the near future, their energy requirements might get prioritized. Again, cynical. Possibly making up scenarios. I'm just concerned when more and more centers pop up in communities with less protections.

◧◩◪◨⬒
32. youngt+WN4[view] [source] [discussion] 2026-02-04 11:37:12
>>youngt+1o1
Down voting doesn't make that statement any less true…
◧◩◪◨
33. infini+YC5[view] [source] [discussion] 2026-02-04 16:22:19
>>nolok+zz
Correct, https://en.wikipedia.org/wiki/Golden_Dome_(missile_defense_s...
replies(1): >>bigyab+SW5
◧◩◪◨⬒
34. bigyab+SW5[view] [source] [discussion] 2026-02-04 17:48:58
>>infini+YC5
Stop linking this same Wikipedia page if you're not going to expound it with further details or evidence. I'm holding you accountable for following HN guidelines here.
[go to top]