so, we have trillions to invest in AI all of a sudden, but for decades, never had money available to put into chip foundries, and power capacity generation.

This circle jerk of an Ai bubble will decimate trillions from the markets. Compounded by the already massive disruption to the workforce, and spiking unemployment.

Curious what happens to all these companies/governments that "Adopted Ai" and the Ai company implodes... those will be some interesting times for sure as companies scramble for solutions.

$ is king... if this thing is taken out of commission for a few days/weeks for repairs every time it encounters a storm... that down time starts to eat into the profits. And by profits i mean they'd still be making money, just not as much as if they had been operating a standard cargo ship.

That is on top of the fact that a standard cargo ship runs 15-24knotts.. and can reach capacities of 300,000 tonnes (almost 60x).

So if you operated one of the behemoths... you can do twice the trips, carrying 60x the cargo... the benefit this offers being smaller is that it can operate in more ports/docks... they don't need to be as deep.

No doubt they'll find a market for green washed shipping options. Don't envision Maersk doing this for their fleet.

There is a company in one of the EU countries that has been operating sail enhanced shipping - with a parachute/kite style sail, and/or "wind wings" if you're interesting in this... been in the works for a while... and pretty interesting concepts out there.

This makes zero sense... proximity is irrelevant and more a hindrance for high value digital processing facilities/military targets... between increasing exposure to attack by proximity alone, you decrease ability to defend from an actual attack - it's an island! You complicate maintenance, staffing, repair, costs....weather/storms...

As others have hinted, the location on the equator makes it a decent listening post, and potentially jamming/interfering of Satelite systems in the hemisphere.

Aluminum hat says- Electronic warfare center.

From my understanding, the number of layers wasn't the issue ever, the thermal issues were. When you stack, you start compound the heat issues. With the tech these guys are talking about, the performance starts to degrade at 50c. Any real world scenario using these chips will require highly specialized cooling solutions.

So while great, they managed to stack vertically higher than others before... is the size/density benefit offset by the cooling requirements? Or is this one of those theoretical wins that will have a real world application with another 20 years development and massive investments in materials science.

think their surface tablets maxed out at 13... but they did a white board/touchscreen style one that maxed out at 85 inches.... To me, it feels like they should let the small shops cater to that niche market, and they can provide easily adaptable tech (i.e software that will work with a touchscreen easily) for those shops to build devices and use msft tech... there was no shortage already of specialized manufacturers creating these super specialized devices, and there was never going to be enough of a market for MSFT to make the expenditure worth it, or for apple.

It really boggles the mind what makes it through the decision making filter and approved... then they need to cut head count because of 'efficiency" while greenlighting more stupidity like this... If anyone knows any books/research papers on this, feel free to share.... really curious what makes organizations usually known for smart decision making do stupid stuff like this...

*In this case, it only makes sense in the- deny your competition the market strategic feint... to draw them into devoting resources to try and take over that piece of the pie. i.e. Msft creates device lineup to cater to "creative types" (apples core market)... so Apple needs to respond (naturally)... Apple devotes a disproportional amount of time, money, and resources to counter msfts token investment and looses a crap ton of money on the venture.

an 18 inch iPad?

Was there any person out there that wanted a tablet that is 18 inches? If you know who, can you go beat them with a rubber hose.

The weight will be insane, there is no avoiding it... it's just a matter how terrible it will be, not a question of IF.

The bulkiness will also make it useless as a tablet. Too cumbersome holding it.

Aside from a few niche areas, this will struggle to find a problem, where it is an optimal solution. Other than sucking money from fan boy wallets.

it's up to your risk matrix.... but i don't see how any risk calculus would allow security patches to sit for this long. And pretty sure it's in violation of the actual rules when it comes to maintain these systems... fed systems have rules they need to abide by.

I've usually advised deferring major updates by a few weeks so kinks can be worked out, but critical security were rolled out ASAP (after a backup was done WITH A TESTED BACKUP/RESTORE procedure) in a gradual roll out. we didn't spends months waiting for security patches, and we grouped systems by mission criticality/exposure/etc... And i think that the backup/restore is usually the ignored bit... and why people are hesitant to run updates too quickly... at these scales, this should have been an automated task, and the more i think about it, the more i feel this was a case of some tech worker no approving the patch for deployment by accident.

Or they deployed, tested, and it broke the systems for some reason... and they couldn't take them offline/further restrict access for another reason... so they were stuck maintaining an unpatched SharePoint deployment vs rebuilding it from scratch with all the patches.

In the case of the Apollo program, it was reliable... not secure. I'm sure if you went back in time and looked at how the chips operated, how the actual programming language behaved, how the machine code was actually compiled, and how the compiler behaved... you would be able to find loads of vulnerabilities in that chain today, they didn't even think/worry about security. They didn't have to worry about someone remoting into NASA and ransoming the moon lander. They did have to worry about ensuring that the computers were reliable, consistent and accurate... as basic (compared to today) as they were.

On a side note - if you're keen on this stuff... look into the track record of patches introducing additional vulnerabilities in the same vein of what was patched, completely new types of vulnerabilities introduced, or breaking previous patches... that's a fun subject to get into.

1- wait... weren't government entities supposed to get first crack at patches?

2- And how in the fuck do you go unpatching a security vulnerability for so long?
(if you say "ohh there aren't any proof of concept out in the wild, so we don't need to worry about it" ... you're an idiot that needs to go back to school... the moment a patch is released, it gets reverse engineered to find out what was being patched... then that gets targeted... this usually happens in under 36 hours... where you go from patch being released to threat actors actively exploiting the vulnerability... this is why it is critical to patch asap. Not to mention state actors don't release f'n POC to vendors or others.)

3- How the fuck are you still running on-prem sharepoint in 2025?

would love to have a beer (or case of beer) with the person responsible for their architecture and operations.

*to say ohhh no..Microsoft...stupid bad Microsoft... as if all vendors don't have fucken vulnerabilities... they all have issues you need to patch ASAP... atleast msft is relatively decent at pushing out patches on a regular cadence so makes planning easier.... fucken security appliance vendor cunts (Fortifucks! SonicWall, etc) love to push their patches/vulnerability press releases out fridays to avoid the stock hits...

so, for US people/executives/CEOs, they need to see in person what the rest of the world has known for years? yes there is propaganda... but you can still see past that...

Are the teams that run these companies really stuck in such thick and dark silos that they are unaware of anything their competition is doing?

The ego and hubris needed to think that the lead gained after WWII was going to last forever without actual investment into what made it possible, and instead sucking all value out of that lead until it became a husk.

investment in energy, tech, schools, health, training, domestic policies to advance industries... Not as if they were being warned about this for the last 40+ years and chose this path.

technically....

expect that "liquidity" will evaporate when you find out that the exchanges that run this shit were all gambling with money and don't have said "liquidity" to cash out those stable coins when you want to... happened before (with relatively tightly regulated banks, especially in comparison to crypto and exchanges)

so yes, you are absolutely correct, as long as there isn't a run on the stable coins, or others and there is blind trust, they'll remain stable.

they are trying some new stuff/techniques now where they'll drop the price low enough and just long enough to get on the best seller list for the duration of the prime days sale. (noticed this with certain power tool manufacturers)

Most sellers also didn't get the message that the prime days sale lasts longer than a few hours... and tons of sellers were just changing the prices up and down every few hours... with a bulk of the items going to prices even higher than when the sale started just 24 hours later.

There are items you can get on sale during the sale/event, but to find steep savings, you really need to be looking, tracking and be lucky. So like with most "sales"

so, casually eating up .5% of the entire US electrical capacity in one deal... that's just processing.. not including cooling and operations... 6 Gigawatts is equivalent to two states of UTAH. or just Oregon. Depending on the facility design, there might be an additional .6 to 3 gigawatts needed for overhead. So, this deal will need about 6-9 nuclear reactors to power. Where is this generating capacity going to come from? I hope these will replace older chips, and not add...

These numbers are insane... and weird that we went from compute power, to just electrical power. Which when applied to computers is just a stupid metric to use... but sure...i guess for AI chips, might be more difficult to find a standard metric for the processors... But electricity used is just weird.

On a side note, weird that Microsoft decided to shift away from AMD for their Ai processors (and go for internally designed ones), only for OpenAI (a partner of Msft) to pick up the AMD chips.

I thought it stemmed from the fact that these organizations are purchasing power generation facilities, or part stake holders in them.

Msft with their plan to revive three mile island and be sole recipient of all it's power generation, others with plans to purchase SMRs and other power generation facilities once construction is complete. A litany of renewable power generation projects owned by them.

Personally, I see this as a failure of public and private enterprise in actually providing for market needs. Power companies failed to add capacity (sure as hell raked in the profits and charged for "adding capacity". Governments failed to ensure enough capacity was being added.

Then these corps come in, buy up existing infrastructure pennies on the dollar, originally funded heavily by public money, built in an environment when it was cheaper and quicker to build things (less red tape, cheaper materials, cheaper labor)... and are taking away capacity needed BY the public. Now, for the rest of the population to get electricity, they need to invest heavily into new projects, fight for decades in courts about regulations, and it'll be 25 years before a new watt of capacity is added. All this will result in the corporations getting cheap electricity for themselves, higher rates for everyone else, and the public is left to pay through the nose .

used to dislike Torvalds... but the older I get, the more his ways/views aligns with mine. He simply has no tolerance for people wasting time with stupidity and has been true to that with his work. If you're going to paste a link, it NEEDS to add something of value. If it simply circle jerks itself with not added context/new information, then they can fuck off. It's akin to msft and their updates and looking them up... often, the links circle back into each other not clarifying what it actually is... You'll end up running in circles trying to figure out what they're actually for.

"Anyone, we repeat anyone, can build a data center" really? with what stable and cheap power, plus water for cooling in a traditionally hot/dry environment struggling to provide water as it is.

The US with it's developed and relatively modern power grid is struggling to provide the power for the data center boom driven by AI... The only nation with relative comfort in the power generation sector that can support this growth is China. And they are shoveling money and resources into adding capacity. India is screwed on this front for at least the next 20 years and that is IF they had a nation level, strategic, massive investment in modernizing and adding capacity to their grid.

India, even if it has viable AI solutions (and not a bunch of techs sitting on the other side of the prompt) to provide willing buyers... is retarded by a lack of infrastructure, domestic $$$ (users/companies to purchase access to the AI service), corruption (bribes are needed for nearly anything), and domestic regulatory red tape that already hinders domestic firms, but is especially hostile to foreign firms. At least China has a way around this- partner with domestic firms and you can operate there, you'll eventually be supplanted by a domestic clone.. but at least you have some foothold there... India - you can't even partner... unless your definition of partnering is handing over all IP/tech rights, and still paying for everything... and hoping to still be involved if the venture becomes successful.... but left holding the bill if/when it goes bust.

On an aside, doubtful the government is keen on replicating the effects AI has had on the job market in tech and other jobs that India relies on heavily. Massive job losses everywhere AI is adopted... What will happen to India when suddenly millions of people are out of relatively well paying jobs? and it won't be 1 or 2 million people... it will be a massive wave of unemployment. They have no short term incentive to see this sector work. Long term, all nations need it - few have the mechanisms/resources to absorb the short term negative effects on society. Ai has the potential to do for the digital age what industrialization did for manufacturing and agriculture.