The datacenters in space argument is just so irrelevant. Earth based nuclear power can already scale up a lot more than what we've pushed. Nuclear Fusion is on the cusp. If Elon was (actually) smart, he'd shift his whole strategy into developing small modular reactors, not betting on the most dependent variable of the whole equation. Now compare with data centers in space, firstly you'd need gigantic solar arrays. Talking like the size of football fields. Only NASA has ever done something close to this and they are mounted to the most expensive engineering feat in history (ISS), and they're not even half the area of a football field. Then on top of that, space is not just cold, its the absence of a medium to conduct heat. What have we seen with earth based datacenters? They need so much fuckin water that they literally pollute towns worth of drinking water. That medium simply doesn't exist in space. You'd need advanced materials engineering on the order of something Humans have never built to create ultra efficient heat transfer materials and even then you'd need some order of football field sized radiators to expend that heat into surrounding space. The last piece of the puzzle is a competent orbital launch vehicle. Of course, Elon might just assume his starship will be ready in time. But if we take NASA's HLS proposal evaluation, Starship has a big chance of not being ready for anything by 2028. That is why NASA, after being so bullish on Starship for HLS back in 2021-2022, have recently started looking for accelerated proposals including from other vendors. \- So yeah. You're right but I just had to write down the stupidity of Elon here from an engineering standpoint. He as much as anyone should know space-based compute is idiotic considering the many earth-based solutions we have. You end up building a data center in space that costs 10x or more in setup, engineering, and launch costs.

---

They're now the size of blueberries 🫐

---

Brain the size of grapes, to compensate

---

Depending on the size of the loan, they might be able to just refinance and roll the debt over.

---

Gonna try on a couple 2/20 Mstr $25ps for size, hope they fit at least a little

---

Conduction what? There's nothing to conduct to, it's fucking vacuum. No atmosphere in space = the only way to dissipate heat is to turn it into infrared radiation. 1 kW needs ~ 3m^2 of radiator surface area. That means you need 3 million m^2 of radiatiors for a single gigawatt of datacenter capacity (this is roughly 3/4 of the size of Rhode Island) Forget even the cost of building radiatiors like that, just the lift cost alone exceeds the cost of every single space launch in human history combined by like a couple orders of magnitude A space data center is the most moronic cyno-brained regarded idea anyone has ever come up with

---

1 kW needs ~ 3m2 of radiator surface area. That means you need 3 million m2 of radiatiors for a single gigawatt of datacenter capacity (this is roughly 3/4 of the size of Rhode Island) Forget even the cost of building radiatiors like that, just the lift cost alone exceeds the cost of every single space launch in human history combined by like a couple orders of magnitude

---

A space data center is the most moronic cyno-brained regarded idea anyone has ever come up with No atmosphere in space = the only way to dissipate heat is to turn it into infrared radiation. 1 kW needs ~ 3m^2 of radiator surface area. That means you need 3 million m^2 of radiatiors for a single gigawatt of datacenter capacity (this is roughly 3/4 of the size of Rhode Island) Forget even the cost of building radiatiors like that, just the lift cost alone exceeds the cost of every single space launch in human history combined by like a couple orders of magnitude

---

The baseline temp of space is like 2.7 kelvin, or like negative 455 fahrenheit / -271C. The only residual temps comes from the microwave background radiation created at the big bang... which has been redshifted from gamma rays all the way to microwave spectrum from the expansion of the universe. In theory and assuming spaceX can get the thermal issues under control... going that low in temperature would allow for like a 40% improvement in compute efficiency for basically free. It will be a mega huge deal when chips hit the limits of small size, in 10 or 20 years... and the servers can basically utilize super conducting materials for even more efficiency. google 'landauer limit of computation'... which is a physical law derived upper limit of compute efficiency that is depended on temperature. Musk is banking on that reality of nature for space based servers and its probably gonna pay off big, just like falcon did 15 years ago.

---

Size big: here come the losses! Size small: here's your peanut sized gains!

---