Intel had no problem selling StrongARM and its Intel-developed successor Xscale. It powered a whole generation of PDAs, palmtops, etc. They sold off their ARM IP to concentrate on x86. We got to see how well that worked out with Atom.
and they are not yet charging for the "tokens" what they need to charge to become profitable
We recently got access to Claude Enterprise and found how expensive it is. We were given $45 a month of budget. Everyone in the team blew through that in 2 days. And considering this is still being "subsidized" I honestly don't see what's the future for "AI Coding".
The whole AI revolution is predicated that today's models are clumsy, error-prone, and EXPENSIVE...just like the early computers, but advancements in technology will make them more reliable and cheaper and that in 10 years, doing the same work will be a fraction of today's cost with greater accuracy. The problem is accuracy is improving VERY slowly. Yeah, on benchmarks they cheat on, you can show whatever you like, but Claude still errors daily for me and has pretty consistently for the last 2 years. If there are positive changes, they're definitely subtle. I'll assume there are some.
But the dirty secret is that each new model is a lot more expensive than the last. There's no sign of the cost coming down. There's really no indicator that will happen. All indicators are that costs are going up. At the moment, the cost is being subsidized by VCs. Once investors pull their funding, Claude is going to have to pay for all the electricity they consume...so is DeepSeek and the lagging models. Also, our tolerance for failure is going to go down.
It's impossible to predict the future, but I see no route to the costs going down. There are no hints at the algorithms getting tangibly cheaper....yeah, some advancements have helped a few percent here and there, but we need factors, not percentages, like 5x efficiency gains, not 5%. The chips are already fully optimized. There's no know reason to assume they will get tangibly faster. We can't shrink them much more like we used to 20 years ago. There's no hope electricity will get drastically cheaper.
However, what I predict will happen? Those AI models will get a LOT better at breaking into your shitty agentic-engineered, vibe-coded monstrosities...so any money you thought you saved by having AI write it? I hope you don't mind data loss and ransomware!
No idea why you think the OP's post was "racist."
At best, it's completely irrelevant, but sounds like race-baiting to me. Sorry, we're not that dumb. That's like me saying "I'm NOT saying your wife is a fat drunk who smells like rancid lunchmeat"...wait...why are you offended? I was saying she's NOT those things!!!
It's a trade off: you get abundant free energy to run the server, with extreme constraints on cooling because your server is running in the most perfect Thermos bottle ever.
Others are taking the opposite tack: undersea data centers for abundant free cooling at the expense of having to get the power down to your servers.
If had to bet on which one is more practial, I'd go with undersea servers. Build them off the coast of Chile, run cables out from batery-backed solar plants in the Atacama desert.
I have been playing with routing tables. I have a few virtual machines around the globe that have full BGP routing tables and that gives me great IP -> ASN details. That means someone comes into a server and that might get redirected to as1221.my2026.example.com but that is a virtual web server with a few hundred thousands others. Everything nice and fast until there is a problem. Once that hits, that ASN gets redirected to the slow scrape box.
Also there is ipasn.net created at APNic. host -t txt 8.8.8.8.ipasn.net gives ownership of the block and other details involving things like its Route PKI status. It is part of their lab so I'm not sure how much traffic it can cope with.
For what its worth, people TALk a lot of sh$t here but our murder and crime rates are actuallly pretty low.
US intentional homicide rate is 5.763 per 100,000 population. That's considerably higher than the wild west known as the Philippines at 4.348, and Liberia at 3.087. It's more than five times Scotland at 1.038, Germany at 0.911 and Australia at 0.854. It's over ten times Northern Ireland at 0.521, and twenty times Japan at 0.229. US murder and crime rates are not low.
As a visitor, I've had random Americans yell death threats at me on the street. Never had that happen living or working as a foreigner in China. Do you have any actual experience or are you just parroting the domestic propaganda you've been brought up on?
It's a stupid "joke" about supposedly getting hungry again not long after eating Chinese food.
It checksums filesystem structures, but not file contents. It's also designed in a way that makes it fragment very quickly (both the filesystem structures and file data), so it performs very poorly on media with non-trivial seek times or slow random reads. It feels like a bit of a one-trick pony.
So the much the same as those with 5 figure incomes who drive rather than take public transit to avoid the homeless people.
Convince a woman to reproduce with you and you'll know most people don't drive to avoid the homeless, they do so because it's unaffordable to raise kids anywhere served by public transportation on a 5 figure income (really even 6 figure). They have to go far out in the burbs and again...while you might be able to get to and from work in 3h via public transportation, you have kids and loved ones waiting for you...I know given your history you might not be able to relate to that...but...yeah, fuck off.
People are driving because it's a direct route to their home and public transportation doubles the commute time, which they don't have surplus time to begin with...because they're decent people who love their wives and kids. Most people enjoy taking public transportation and as someone who takes it all the time, the homeless are not a huge problem in the last 3 cities I've lived in. They're not going to be your top annoyance commuting.
AI centers demand two things, power and water.
And so while they're going to use gas to power this site, and god knows about the water, the Grande Baleine dam in Quebec is just sitting there with... power and water.
Seems like a no-brainer to me. It's also closer to the US and Canadian population mass, so lower latencies. Running some fibre down the existing corridor does not seem like a possible impediment either.
Running fiber inside the Guard Cables of the High Voltage Transmission towers is a practice that dates back to the mid 90's.
Having said that, the location of a Datacenter is dictated as much by water, energy and fiber, as it is dictated by financial incentives to "git'em-jubs". Discounts on the land, tax rebates and such. Probably, the politicians with the dam tought that the dam alone would be enough to sawy Meta, meanwhile, the politicians in alberta were more acapable to achieve their goal
The spacecraft reacts against the Earth, the source of the magnetic field it's working within. The Earth is just so much more massive than the spacecraft that it doesn't affect it very much. It's much like when you jump, you're reacting against the Earth by pushing against it directly. It reacts against the Earth via electromagnetic forces rather than physically pushing against it.
A corncob pipe and a button nose and two eyes made out of coal (Frosty the Snowman)? Or an attitude and a carrot for a nose (Frosty the Yobbo)?
The actual original arcade version of Ninja Gaiden, or the completely different game on the NES? Because the punishingly hard NES version that a lot of people remember only really had the name in common with the arcade original.
Considering this study is about brain aging and not overall intelligence you seem to have gone completely off the rails on an anti immigrant tangent.
Perhaps if you kept up on your Duolingo to keep your brain more limber, you might not have hallucinated an anti-immigrant rant. But hey...you can make baseless accusations if you like...does it make you feel powerful?...did you put me in my place?
For large values of one, one equals two, for small values of two.
