throwaway31131

So much for the $1.4B spent on NUVIA.

https://www.allaboutcircuits.com/news/startup-key-apple-goog...

Maybe, or maybe the size of the chair market grows because with $2 chairs more buyers enter. The high end is roughly unaffected because they were never going to buy a low end chair.

Just out of curiosity, what software product were you making in two weeks before using AI? Or maybe I’m misunderstanding your use of shipping.

And if you believe the numbers from the press on Google’s AI spending, that’s an amazing deal.

https://www.indiatoday.in/technology/news/story/google-ai-bo...

I guess we’re being a bit vague on timeframe but chrome books launched in 2011 so they’re one of those products that took ~10 years to be an overnight success, with 2020 being an accelerant. So my vote is no.

What’s the “Ender’s Game Approach “? I’ve read the book but I’m not sure which part you’re referring to.

Interesting. Thanks for the suggestions.

I posted this example before but academic papers on algorithms often have pseudo code but no actual code.

I thought it would be handy to use AI to make the code from the paper so a few months ago I tried to use Claude (not GPT, because I only have access to Claude) to recreate C++ code to implement the algorithms in this paper as practice for me in LLM use and it didn’t go well.

https://users.cs.duke.edu/~reif/paper/chen/graph/graph.pdf

100GW per year is not going to happen.

The largest plant in the world is the Three Gorges Dam in China at 22GW and it’s off the scales huge. We’re not building the equivalent of four of those every year.

Unless the plan is to power it off Sam Altman’s hot air. That could work. :)

https://en.wikipedia.org/wiki/List_of_largest_power_stations

Cost per transistor is increasing. or flat, if you stay on a legacy node. They pretty much squeezed all the cost out of 28nm that can be had, and it’s the cheapest per transistor.

“based on the graph presented by Milind Shah from Google at the industry tradeshow IEDM, the cost of 100 million transistors normalized to 28nm is actually flat or even increasing.”

https://www.tomshardware.com/tech-industry/manufacturing/chi...

I’m with you, I’m not sure the volume or cost would be less once you factor in capacitors that are high enough quality for the application.

The datasheet mentions low profile a lot. That does make sense as one can make a flat, high quality, capacitor. Making a flat high quality inductor is harder and probably more expensive and likely consumes more volume overall. I can imagine some applications where being flat is important, like the back of a panel.

These types of switching circuits are very common inside ASIC where the high speed isn’t an issue, you don’t need to move all that much charge, and one can’t easily (if at all) support inductors.

Intel opened a packaging facility in New Mexico in January that’s supposed to their largest.

https://newsroom.intel.com/intel-foundry/updates-intel-10-la...

Maybe valued at >$1B, but who do you sell it to?

Packaging is something Intel does very well and they have facilities is the USA. The rumor is many companies might do final assembly with Intel.

https://www.eetimes.com/intels-embarrassment-of-riches-advan...

For some definitions of serious, sure. The main critical piece that’s missing is all the testing infrastructure. Buying 100 or so ASICs for university use is one thing. Buying 100K, or more, is another.

Not the gdb support via jtag that software engineers need, they have that. But the various manufacturing test suites, which do modify gate netlists, and automated circuit characterization techniques that electrical engineers and the manufacturing engineers use.

Details on the ETH Zurich open source ASICs can be found here:

https://github.com/open-source-eda-birds-of-a-feather/open-s...

Presented at DAC 2025

That’s an interesting idea.

But the magic is in the “find patterns” stuff as memorization is just data storage. If you think of the machine learning algorithms as assigning items a point in a space, then it does uncover neighbors, sometimes ones we might not expect, and that’s interesting for sure.

But I’m not sure it’s analogous to what people do when they uncover patterns.

Definitely interesting to ponder though.

That’s interesting because I can’t imagine learning a subject without a textbook. I have a hard time believing another medium would have the depth and density to get all the points across. Although it does depend on the subject matter and one’s learning goals.

But I also do read textbooks for fun… Now that I have a few decades of experience in a lot of these subjects I get way more out of the books. And I can start to understand more of the meta information. Like, of all the things the author could’ve used as an example, why did they pick that. Also, it’s hugely interesting for me to look at the homework problems and theorize why this particular problem was picked. Especially fun for electrical engineering books. But ya, I’m weird like that.

Isn’t the primary issue that newer designers don’t know they show run ERC (or that ERC even exists)? Isn’t your tool going to have the same issue? i.e. how do user even know they should run it in the first place? How do you plan to overcome that barrier?

I’m not against more automated checkers, I’m very much for automated checkers, but I’m curious how you plan to not repeat the mistakes of the past.

Ingesting data sheets is an interesting angle compared to normal ERC, which KiCAD supports out of the box, but how good is it at the ingesting?

Datasheets themselves are inconsistent and incomplete so I’m wondering how you evaluated the accuracy of the import and what your acceptance criteria is.

I had the same difficulty. i.e. what was the issue that caused a headcount reduction from a 1000 to 10 or less? It’s possible the organization had larger issues than generated code but again, difficult to draw conclusions without details.

I think that’s what everybody wants. The problem is the statement 'to each according to his contribution' is subjective. Weighing “contribution” in a way everyone agrees on just hasn’t materialized. It’s probably going to get harder too as technology improves and changes more and more rapidly.

The fan would move the heat off that given sled but if the channel between the sleds can’t remove that extra heat you probably didn’t help yourself.

My point is the picture doesn’t show any details on the room or what’s outside the rack so it’s hard to know what’s optimal.

For many of the “very good” engineering schools that I know of they got “very good” status because of their graduate programs. In graduate school a 1:few relation is almost certain. In undergraduate, not so much.

You hinted to it but at what point are you basically giving individual oral exams to the entire class for every assignment? There are surveys where 80% of high school students self report using AI on assignments.

I guess we could go back to giving exams soviet Russia style where you get a couple of questions that you have to answer orally in front of the whole class and that’s your grade. Not fun…

I’m not sure even $1T has been spent. Pledged != spent.

Some estimates have it at ~$375B by the end of 2025. It makes sense, there are only so many datacenters and engineers out there and a trillion is a lot of money. It’s not like we’re in health care. :)

https://hai.stanford.edu/ai-index/2025-ai-index-report/econo...

I'm not familiar with Ben Shapiro or Steven Crowder but what makes you so sure that their point of view is unreasonable? It seems you agree that their viewpoint is opposing, which likely means their premise are different, so isn't it entirely possible that their conclusion is reasonable and logical when one starts from their position?

Also, there isn't one source that can represent the "conservative" viewpoint because there isn't one conservative viewpoint. There are many factions within the Republican party with sometimes shockingly different points of view. Just like the Democratic party representing the "liberal" agenda.

I could just as easily ask, where is the one source I go to get an understanding of the liberal agenda? (Just a rhetorical question, I actually don't follow the news and don't plan to.)

I don't think this was the point of the post at all.

Their bottom line summed it up perfectly.

"We’re not saying “never use CUDA-on-AMD compilers or CUDA-to-HIP translators”. We’re saying don’t judge AMD based on them."

Having your own ASIC comes with a huge sunk cost. One gets advantages from that, but it's still a lock, just a lock of a different color. But with Google money and manpower, management can probably pursue both paths in parallel and not care.