one thing to point out: with a planet that is 20-30% larger than the earth the rocket equation makes it almost it impossible to get out of the atmosphere with chemical rocket technology. So even if you could visit, we could probably only have drop ships or rovers. coming back out of the gravity well without space elevators or stuff would be very hard. It also means that life on those planets would be hard for them to establish a space faring society.
This isn’t accurate. 20-30% larger diameter with a constant density would mean about 20-30% greater escape velocity (mass is radius cubed but gravity is inverse radius squared, etc). So instead of 11km/s escape, it’d be 13-14km/s. And instead of ~9km/s to reach LEO, it’d be about 11km/s. This is not at all a showstopper. Even in the 60s and 70s, we built rockets with enough stages to go land on the moon and return to Earth, which is a round-trip delta-v of like 14-15km/s, and the Soviets did this mostly with low Isp hypergolic propellants, not even hydrolox let alone an exotic propellant like nuclear thermal hydrogen or electric thrusters. We often just launch stuff directly to GTO or GSO instead of bothering with electric propulsion, so even today’s rockets (many of which also have high enough thrust to weight ratio, like an Atlas V 551) you could do it (and then probably rely more heavily on electric thrusters). Even with higher gravity losses.
It’d be more expensive, sure, but the idea that it’d be almost impossible is just not true at all. It’s a very common thing that even a lot of fairly smart people often say, though.
A larger earth with the same average density of Earth that has 1.3 Gs of surface gravity would need around twice the fuel to escape into space and a system that could handle that extra fuel and longer burn without self-destructing.
The fuel itself would weigh more, and thus require more fuel to lift. The fuel to lift that fuel would also weigh more and would also require more fuel, etc, and the radius of gravity to escape would also be larger even though it would also be mediated somewhat by the inverse square rule.
I actually wouldn't be surprised if you needed more than twice the rocket fuel to leave a 1.3 G super earth that is 1.3 times it's diameter.
There are a bunch of exoplanets in that size range, and we could probably contrive chemical rockets with over 20km/s delta-v if necessary. Nuclear rockets 30km/s.
How many in Goldilocks zones with stable orbits for billions of years? With non-toxic soils and atmospheres? With water at the surface and in significant amounts?
Now, how many can humanity reach in the life time of frozen embryos?
And if it's more than 30% bigger then escape velocity is likely beyond the practical limits of science.
That is a totally different topic. We’re talking about a hypothetical, where Earth is 20-30% larger, not whether an identical copy of Earth but 20-30% larger actually exists nearby. We know such roughly similar exoplanets are common, though. In that size range and in habitable zone.
And again, 30% greater escape velocity just isn’t a challenge. Existing rockets could be used to reach LEO, like Atlas 551 (well, probably want 552 to reduce the gravity losses). And then you can use electric thrusters (which don’t have a hard Isp limit) to reach wherever.
Heck, we could probably reach LEO even if escape velocity was twice what it is. Even 3 times, if you include more exotic propulsion methods that we’ve done some testing for (nuclear thermal, etc). With staging, etc, it’s just a matter of a worse and worse mass fraction, not really whether it’s possible or not. And once you get to low orbit, electric propulsion allows much better scaling as you can adjust Isp higher (although reduces thrust and means it takes longer to do your mission).
It certainly wouldn’t be easy, but a massive rocket could get up high enough in a suborbital trajectory where it would meet up with a skyhook or some other vehicle specifically designed to fully boost a capsule the rest of the way into orbit.
That all depends on already having infrastructure in orbit though. I assume the difficulty of such a thing would pale in comparison to the challenge of arriving at that planet in the first place.
A bunch of people are commenting that this could lead to grid-death spiral where people opt out of the grid because they can just use solar + batteries. Maybe, but I doubt in Vermont. Vermont is really hard for solar because of tree cover and clouds (according to my conversations with a solar installer). Secondly, in this setup you have buy your own power plant (solar + batteries), which likely for most folks, is a pretty big expense. Green Mountain power subsidizing the cost of buying the Tesla power walls is a big deal for me because I use this setup for a cabin / second home and want to keep my costs relatively low. I can't do solar because I on the side of a mountain and lots of trees.
I couldn't read the article because of paywall issues. I have a Tesla powerwall that I pay $50 / month to lease from green mountain power. I have piece of mind for power in a rural area - the power company get peaker benefits. Seems like a great deal. I wish we had this in Massachusetts. Mad river valley area.
Ex AWS. AWS had one zone specific products. If those were in the affected AZ that data is lost. AWS has multiple multi-AZ products. S3 and EFS are good examples. If you loose an AZ for an extended periodic of time, the dataplane of those services will migrate shards around until you get back to your durability goals.
if you use these things to generate design docs / text, it should be good news if it is actually better at prose as advertised. Some people like sol for prose better the anthropic models.
This is somewhat orthogonal to the article, but the whole bubble on AI data centers seems to presume that the need for compute is so massive that it far exceeds the expected optimizations we would expect with at scale inference (PIM, ASICs, etc). I would expect that there is a set of optimizations like this one (or variations) that would someone negate the buildout. But it's not really discussed.
There's been a ton of optimizations already, it hasn't remotely reduced demand even temporarily. More efficiency just makes the compute have even higher ROI per $ and watt spent.
With sufficient optimisation, there ought to be a tipping point beyond which local inference is good enough. And, sure, datacentre compute will still be needed for training but one of the biggest current uses will begin to taper off.
The question really is how soon we reach that tipping point, and whether it's before or after the current bubble runs out of steam for some other reason.
>there ought to be a tipping point beyond which local inference is good enough
There's no such ought really. Even at current levels you'd need like a 100x gain from here to approach current top proprietary models (probably a lot more for say Mythos or Mythos 2), and it's not like they are stoppng to improve. This is before we even account that you'd just be running 1 agent then, and not a swarm like you'd be able to in the cloud or that you can do only so much compression before you are losing out
It's not just inference, some things done in data centers like simulations, testing, are complementary to inference.
And in these types of hardware, the time between a successful prototype and a fully deployed product is pretty long. Maybe they count on that to know when to stop?
Some other commentators said this was AI slop. I don't have enough expertise to say if it really is, but it sure has the "claudish" cadence of AI generated text that makes it hard to take it seriously. The authoritative and strong statements, the use of phrasing with colons and dashes, the use of bold, etc.
To me it read like a combination of human written and AI slop, as if the author had Claude write it and then rewrote portions of it, or the author wrote an initial version and told Claude to "punch it up, without rewriting the whole thing entirely".
Regardless of who or what actually wrote it the piece was much longer than it needed to be (though I do think it highlights a very real problem).
I really wanted to stay in the google chromebook / googlebook echo system. But the hardware was expensive for what you get. Apple announced the macbook neo and I picked one up. Great hardware. can run light weight mac software. I don't run much beyond chrome and wahoo SYSTM (bike trainer app). It's really solid hardware and cost $600 or so.
I use gemini extensively (and claude). But - do I need this integrated in my laptop? Don't quite see it. And it's hard to beat Apple on hardware now.
this post will probably never be read but.. I was on the team that was trying to make the marriage of S3 and EFS work a year ago. it's a pretty hard problem. At one point we proposed this solution (which seems like a caching layer) but it got shot down for a more complex system that would have attempted to rebuild EFS on faster S3 blob storage. I left before this engineering monstrosity made significant progress, and it clearly died at some point.
Looks like they went back to a simpler solution they could deliver but with some obvious warts. good to see something get launched but the sausage making her was brutal.
The reality is that if you read https://www.allthingsdistributed.com/2026/04/s3-files-and-th..., it sounds like the great minds at S3 figured out that a caching layer was the way to go. We (EFS) fucking proposed that years ago. But we had to deal with Seattle and the S3 braintrust who didn't want to do that. I know we wrote a PRFAQ that was close to this concept probably four years ago. The political story is that EFS was taking over by S3 and the EFS folks didn't have the agency or political backing to build a more workable solution. So we wasted a shit ton of time tackling something that was never going to work and many of the tenured EFS engineers left.
100% agree to this sentiment. Although Amazon/AWS seems to be overall innovative, amount of ideas and passion killed between the same meeting rooms the article describes as "innovating driving, heated conversations" are immense.
Obviously not the same, but at home I am running a Raspberry Pi with s3fs mounting my personal S3 bucket. I am exposing the same directory with /etc/exports (NFS). Which also allows me to use filesystem-caching as a bonus on the client side.
On the other hand, I should probably move out from S3 and use R2 or something...
How could they release something that doesn't support atomic rename and has no prospect of supporting atomic rename? Lots of workloads will crash and burn on this layer.
reply