if I may add, you can't really launch a station three times the size of ISS with a single rocket so there will be multiple launches. Just the launch costs alone could likely finance multiple similarly sized server rooms on land.
It should be smaller than ISS. IIRC, ISS has three solar arrays and two radiators. DGX is about the size of a fridge. Let’s add two more fridges worth of infra for solar and radiator. Maybe another fridge for comms. But these three fridges replace all the modules of ISS. It’s still gonna be about 6 (maybe up to 10?) tons of mass to lift. But Starship can do 100-150 tons to LEO so might be doable.
What is the power budget for that DGX. The power budget for ISS is 75–90 kW. If your DGX fridge needs more power, it will need similar capacity of solar and radiators.
> In 2027, the unified approach will almost certainly be taken to new levels as NVIDIA's “Kyber” system will launch, with 576 GPUs in a single rack requiring a whopping 600 kW, equivalent to delivering enough power for 500 US homes into the space of a filing cabinet.
That fridge needs 600 kW of power. That will require 6x more solar panel space than the ISS and 6x more radiators. It's not about the volume of the object but rather its power and heat budgets.
If that fridge can reject all of the heat from the rack in space, it will work much better on earth and all the data centers would be using that fridge instead of their cooling towers and AC.
DGX is only 10,2 kW. But also only 8 H200s. Kyber seems a bit more power efficient (per-GPU) but requires much more power for a single unit. With that power requirement it doesn’t seem like it will fly.
10 kW is something that could be household load (a backup generator for a home when there's a power outage).
If one could put a DGX in my basement without issue (there's an idea - would you trust me with a DGX rack in my basement for six months for winter heating? https://www.cnbc.com/2026/01/27/data-centers-ai-district-hea... ), what is the value of shipping it into space? Granted, I couldn't afford a DGX in my basement... but it's not one DGX that you're putting in a datacenter, but rather racks upon racks of aisles upon aisles.
Agreed. I also prefer conformity over sporadic use of new features going against an already set of standards in a codebase. it's overall less cognitive load on whoever is reading it imho.
My bar for super-rough is Servo, which doesn't have password autofill… and doesn't render the Orion page right.
Orion is less rough, but the color scheme doesn't work, and it doesn't have an omnibar (as in: type in the address bar, enter, and it shows search results).
if you are a little bit familiar with graphics you go: duh, things appear smaller with increasing distance. if you are not tho, it's a great intro to perspective projection. I love how accessibly educative his videos are.
I always found it odd that perspective had to be "discovered" by artists, but a little digging online turned up this interesting, detailed look at its history.
It's a lot less about being discovered, or invented, and a lot more about the idea of using it at all. The Renaissance was a massive change in culture. Before that, art was a tool used in rituals or storytelling rather than something to be enjoyed on its own. There was more emphasis on reproducing things as they actually were than how they looked from a particular vantage point.
Artists are still struggling with the fact that human perception arises from binocular vision. Two distinct retinal inputs are integrated by distributed neural processes into a single, coherent 3D experience. This integration is neither a simple planar stitching nor a direct representation of the world, but an active construction shaped by neural computation and subjective awareness.
It is quite likely that artists in earlier periods struggled with this as well, and were less concerned with adhering strictly to a photographic or geometrically exact perspective, as we are. The adoption of the camera obscura probably influenced things a lot.
Even ignoring binocular vision it's very unintuitive to "draw what you see" because of this. Our brain usually interprets our environment as objects, 3d shapes, and things. Turning that off and trying to grab a literal image from it is difficult
When I was a little kid trying to do 3D graphics on my Spectrum I couldn't find any books with the algorithm for how it worked. I remember my artistic friend and I sitting down with reams of graph paper trying to figure out how to do it. It's so simple and obvious after you learn, but until you do I felt like a caveman.
it is indeed a shame. if you are doing anything remotely new and novel, which is essential if you want to make a difference in an increasingly competitive field, LLMs confidently leave you with non-working solutions, or sometimes worse they set you on the wrong path.
I had similar worries in the past about indexable forums being replaced by discord servers. the current situation is even worse.
reply