No this is more like having someone standing over your cubicle, and every other person's cubicle, taking notes of every time you go to the bathroom or get a drink of water or stretch.
My mother used to work as an environmental consultant in her career as a chemical engineer; and one of the larger projects she helped direct was a water treatment facility for a pit-mine-turned-lake.
it turns out that all of the things leaching into the water made it quite acidic and relatively toxic!
so, y'know, probably good to check what kind of mine it was before going for a swim!
Well, the context here is Germany (from the headline). And this isn't a country that overly embraces libertarian / darwinism.
Over here, the seas that are public for swimming are usually checked. Not just for acidic things from a mine. But in summer, when it's hot, for cyanobacters. And then they warn (e.g. on radio, in local TV news like "Hessenschau", on the web. And if there are organized beaches, then even with posters.
We have few actually dangerous ones. After a search, I only found the "Silbersee" in Bayern. That one has sulfuric fumes. And that reminded me that some vulcanic "Maar" sees in the Eifel (area next to Belgium) might be dangerous to to sulfur and CO2 since they are still a bit vulcanic active.
Ex-seas from mining companies that are still toxic are fenced off and not accessible.
Maybe if you're a US reader you don't know about the "fencing culture" difference between USA and Europe. In the USA, basically everything is behind a fence. I drove e.g. in Texas, New Mexico, Maryland: even the corn fields were behind barbed wire or other fences. This is totally not the case in Germany. Here by default things are accessible. Fields, Meadows, Forests ... by default no fences. Even if e.g. the forrest is private property. So if you, as a german, come to an area where the lake is actually fenced around, then this speaks loudly to you and your complications sensor.
> Over here, the seas that are public for swimming are usually checked
An interesting 'Germanism': German 'see' translates to English 'lake', German 'meer' translates to English 'sea'. It is even more interesting when comparing German to Dutch:
German 'see' -> Dutch 'meer'
German 'meer' -> Dutch 'zee' (pronounced almost like German 'see')
at current prices it's above $1000/kg to orbit, and we're just conveniently ignoring all of the operational constraints of a completely unmanned datacenter upon which you can perform no maintenance.
What do you do when a micrometeorite causes pinprick leaks in your radiator loop?
is LEO sufficiently shielded from cosmic and solar radiation?
nobody is building real datacenters in space at $1000/kg - at that price the cost is comparable to ground-based and thus no big point. The explosion will start at $100/kg when as i mentioned the base cost would beat ground-based almost 10x, and thus would allow for the additional costs (with total cost still beating ground-based) of all the additional issues/concern you and the others usually mention in this context.
4 years ago we had the ChatGPT moment and notice how civilizational change has been accelerating since then. In a few years we're gong to have the seemingly profound Starship moment. Back at the time it was obvious, i may be said it is even here, that Starship low price will cause explosion of amount of launched payloads and a technological revolution as a result, yet it wasn't clear what kind of payloads it would really be. And now we have the AI - Starship and AI are basically ideally dove-tail each other. Two revolutionary technologies accelerating each other - Strarship lowering deployment cost and removing other obstacles for AI while AI providing guaranteed massive scale launch market for Starship - that will be a thing to watch (or participate if you're lucky :)
As per another of my comments, Alphabet, applying learning curves arbitrarily far into the future, recon it will take SpaceX launching 370,000 tons to LEO to make the costs come down enough to be worth it: https://arxiv.org/pdf/2511.19468
That's not even $100/kg, that's $200/kg.
Even my bull case puts that ($200/kg) 10 years off, which is so far away it lacks relevance just because compute and AI models move so much faster than that timescale; my bear case says that's about 45 years off.
Always account for how over-optimistic Musk's public timelines are. Starship has only just a few days ago managed its first circular orbit, but when he first talked about it:
The basic game plan is like we're going to send a mission to Mars with every Mars opportunity from 2018 onwards. So and they occur approximately every 26 months. So you know, we're establishing cargo flights to Mars that people can count on for cargo.
SpaceX is building spaceport with 10 pads or something like this. At 1 Starship/pad per day that 1000 tons/day to LEO - i.e. 1 year.
The Starship is making great progress. Even if it takes 10 years to once-a-day reusability, the need will be even greater by then.
> compute and AI models move so much faster than that timescale
tech moves faster with each iteration, and datacenters and networks only grow with it. The ground-based datacenters has and will continue to grow in cost. While space-based costs will be cheaper and cheaper. Not the first time we have 2 such curves, and such curves dynamics usually leads to paradigm shift.
> SpaceX is building spaceport with 10 pads or something like this. At 1 Starship/pad per day that 1000 tons/day to LEO - i.e. 1 year.
When you making forecasts that assume success, they're generally over-optimistic.
So, here, if they launch 1 Starship per pad per day from 10 pads. And also if they double the current capacity per launch because the demonstrated launch capacity right now has "only" been 44 tons. (44 would be impressive! It's just that Musk's pronouncements are nowhere near what gets delivered).
> The Starship is making great progress. Even if it takes 10 years to once-a-day reusability, the need will be even greater by then.
Reality check: we're 14 launches in and development is so slow that we're more than 10 years past when he said he'd be sending things to Mars within two years and it's only just now managed to get a circular orbit.
Getting there from here requires funding at the intermediate cost, not the final cost. That means they have to spend order-of 300-400 billion dollars, which means they have to raise that many billion dollars in capital, which is around 4x times the 86 billion they actually raised in their IPO (not trillions: market cap is what you get if you could sell all the shares, and they didn't do that).
This is, inflation-adjusted, around 1.5 times the Apollo mission. Since this whole thing began, the development process has been slower and more expensive than the actual Saturn V launch program.
Right now, SpaceX needs more financing rounds to cover their existing spending commitments. Some of this has nothing to do with going to space, thanks to them having merged with xAI and the AI spending spree.
> tech moves faster with each iteration, and datacenters and networks only grow with it. The ground-based datacenters has and will continue to grow in cost. While space-based costs will be cheaper and cheaper. Not the first time we have 2 such curves, and such curves dynamics usually leads to paradigm shift.
False. Any argument that "datacenters and networks only grow with it" applies regardless of where those datacenters are placed, ground or space.
They get more expensive in proportion to how much compute you have, and less in proportion to the tech of the processors themselves; but place is irrelevant to the compute tech, price of location only shifts with regard to the transport.
Why this matters:
Applying Koomey's law for 10 years makes any give unit of compute 14 times more energy efficient, so what was once a 1kW data centre GPU becomes a 70 watt laptop integrated graphics processor, and what was once limited to the 70 watt power envelope of a laptop becomes the default performance of a bargain basement phone.
Also, the models seem to still be getting more efficient at fixed parameter count, so model performance which can be had on your phone today would have needed hundreds of gigabytes and a beefy GPU just a few years back.
Either by themselves, let alone both, means it's quite plausible that all the data centre investment, both on the ground and in space, is worthless within a decade.
More applications need the ability to log in as a "read-only" mode, so you can more safely grant access to tools like this.
I can imagine something like Dots being utterly invaluable for running a traditional brick and mortar business, streamlining all of the admin work, but until they're really safe and well integrated we'll have to wait...
I mean it sucks but it kinda seems like their work training new models has vastly outpaced the ability to comprehensively secure those systems.
I don't think it's because they're incompetent or lazy, but that the nature of the problem is that security vulnerabilities seem to scale superlinearly with complexity but model training scales linearly or logarithmically with complexity.
But the organization isn't allocating the resources accordingly, which would likely be economically unsustainable for the competitive environment they're in. Put simply, if the security team was more than twice the size of the research team, they might have a chance at keeping up.
I didn’t like the article because all of the missing defense-in-depth necessary to prevent the intrusion is firmly in the category of traditional cyber security. I could sympathise with an argument about organisational pressures and lack of time/people leading to an insecure sandbox, but that doesn’t seem to be the argument they are making. People have been securing systems that can run untrusted code for a very long time, and whilst mistakes happen, there are not really many novel ones.
The problem of “how to stop the model from attempting to be a hacker” is probably a more interesting and novel one, but it does not need to be solved to prevent the incident.
There's no magic involved, just as-yet-unexplained phenomena.
Everything in the chain of causality has some explanation or mechanism, understanding it step by step dissolves the illusion of "emergence".
reply