It really doesn't. You're purely relying on radiation fins to carry heat away, which are incredibly inefficient.
> The radiator surface area problem also scales uncomfortably. At 838 watts per square meter, rejecting 1 megawatt of waste heat requires roughly 1,200 square meters of radiator. Deploying that much surface area on a satellite is a structural engineering challenge that gets harder with every order of magnitude. The ISS solar arrays span about 2,500 square meters total.
So even a 2MW data centre in space requires a cooling array rivalling the international space station.
Starcloud launched a single H100 in November and they were unable to run it 24/7 due to heat buildup.
Even with novel solutions to make heat transfer to the fins more efficient, like phase-change liquids, the limiting factor is that the vacuum of space is a tremendous insulator.
It's also not so easy. The for profit entity essentially has to buy all the previous entity's assets and assume liabilities. Since the assets are considered charitable trusts, the proceeds from the sale then need to go into a new charitable foundation. Regulators also need to approve that the assets were fairly valued and the entire process was free of conflicts of interest. Albeit complicated, the process is pretty straightforward for hospitals. But in OpenAI's case it seems more like they tried to jump through every legal loophole they found.
You do it to yourself, you do, and that's why it really hurts.
> Importantly, the company did not mandate AI use (though it did offer enterprise subscriptions to commercially available AI tools). On their own initiative workers did more because AI made “doing more” feel possible, accessible, and in many cases intrinsically rewarding.
This isn't a good idea regardless of why it's being deprecated.
If it's no longer being maintained then put a depreciation warning and let it break on its own. Changing a deprecated feature just means you could maintain it but don't want to.
Alternatively if you want to aggressively push people to migrate to the new version, have a clear development roadmap and force a hard error at the end of the depreciation window so you know in advance how long you can expect it to work and can document your code accordingly.
This wishy-washy half-broken behaviour doesn't help anyone
Western employment has survived because automating and outsourcing labour has pushed people to take up knowledge/services work.
If AI is somewhat successful at automating knowledge work, what feasible job could exist that doesn't require your mind or body?
Services like healthcare and plumbing aren't going away of course, but there's not enough demand to support an economy on those jobs.
In my opinion the whole economy needs a rethink regarding what our actual goals are as a society, and maybe AI will force that conversation to happen, but I'm sceptical if it'll be a well-considered consensus.
> If AI is somewhat successful at automating knowledge work, what feasible job could exist that doesn't require your mind or body?
Very few office jobs in bigcos, but it does mean that you'll not need as much capital as before to start a business and compete with existing incumbents.
In a world where AI automates jobs, arguably any business services you could offer would likely have no customers because those potential customers can automate the services you offer themselves just like AI automates the employment skills employees offer. Not sure why people keep bringing up entrepreneurship as if that couldn't be automated.
> In a world where AI automates jobs, arguably any business services you could offer would likely have no customers because those potential customers can automate the services you offer themselves just like AI automates the employment skills employees offer
Take the POS market for example. It's relatively trivial to setup your POS system in dbase/foxpro/MS-access. Yet almost every shop uses lightspeed/toast/square because they don't want to take on the additional burden of maintaining it and want support. They also hire web developers/designers to make their website even though framer/square space are super easy to use.
With AI I think what's happening is that more competitors are challenging these incumbents leading to downward pressure on pricing. Which is great for customers!
> Not sure why people keep bringing up entrepreneurship as if that couldn't be automated.
Ah, how would you automate entrepreneurship? This isn't the matrix where AI will suddenly wake up one day and start an entrepreneurship journey.
> With AI I think what's happening is that more competitors are challenging these incumbents leading to downward pressure on pricing. Which is great for customers!
If AI automates jobs, POS system vendor is using an AI to run the business. You can do that yourself and remove the middle man.
re: "Ah, how would you automate entrepreneurship?"
I meant that every non-physical business need that would be filled by a local business can now be filled by the same AI you have already use to automate half the jobs away.
There is this fantasy where employers can replace workers with AI but businesses cannot replace their software vendors with AI
There are plenty of non-US index funds, like EU, UK, Japan and others. There are also indices that track smaller companies rather than just the S&P500 or Nasdaq.
Or diversify in both directions - small US, and big and small international funds.
This is often made worse as a result of hiring outside consultants. Firstly they don't have the institutional knowledge you have when starting a project, but they also aren't incentivised to properly document and hand over their knowledge at the end since that means less future work.
This is why a lot of government projects take so long, they don't see the value in keeping an in-house team of trained experts (see the difference in train line contruction costs in the UK compared to Spain), until you realised how good they were but you can't hire them back.
Back when Anderson Consulting hadn't yet disgraced itself, a corp I worked for hired them for a cost-cutting project. "Find ways we can cut costs."
Blew my mind that 22-year olds, fresh out of good-brand universities (their "qualification"), were doing the research on how to cost-cut. Chesterton's fences all over the place were violated. It was sad watching the slow-moving disaster.
That’s a great point. In my city, they started paving side streets with with in-house staff. They have dedicated funded budget, government can borrow cheaply, and they don’t need to price risk and margin.
Over 5 years, they spend 30-40% less depending on utilization of equipment. (We had a two slow years due to pandemic and weather)
For the work that’s funded by state aid or grants, they use contractors as its a variable workload.
I am currently on a team of consultants. Ironically, most of us have more institutional knowledge than the client due to internal churn. Seems like every few years they try to cut our utilization in favor of some off-shore company that's "cheaper", the project blows up, then we have to jump in and save some middle manager's job.
Except it's getting so difficult to find the companies producing the more durable alternative, so everyone is forced to buy the flimsy piece that falls apart
It is not that hard, if you do the minimum effort to educate yourself. For example 20 years ago I struggled to find motorcycle gear in Eastern Europe, it was very hard and stuff was extremely expensive for the salaries in this region. I bought initially cheap stuff that broke fast, then the next generation I knew what to buy and I have now equipment that is over 10 years old that I am using with great pleasure. It is similar in most cases I have to buy something, but it takes some effort to look for options.
Assuming the rides are comparable, the article has a table which includes price/km (weird) of Lyft: $7.99, Uber: $8.36, and Waymo: $11.22. On that data, Waymo is roughly 40% higher, so way more than just a tip.
in my limited experience, you're not usually tipping a percent but a flat dollar amount of like $2-5 per ride, so $3 on an $8 ride basically removes the price difference between lyft/uber and waymo
> The radiator surface area problem also scales uncomfortably. At 838 watts per square meter, rejecting 1 megawatt of waste heat requires roughly 1,200 square meters of radiator. Deploying that much surface area on a satellite is a structural engineering challenge that gets harder with every order of magnitude. The ISS solar arrays span about 2,500 square meters total.
So even a 2MW data centre in space requires a cooling array rivalling the international space station. Starcloud launched a single H100 in November and they were unable to run it 24/7 due to heat buildup.
Even with novel solutions to make heat transfer to the fins more efficient, like phase-change liquids, the limiting factor is that the vacuum of space is a tremendous insulator.
https://thecoolingreport.com/intel/starcloud-orbital-data-ce...
https://satnews.com/2026/03/17/the-physics-wall-orbiting-dat...
reply