Massive "no shit"/"told you so" moment. ARM is an incredibly important foundational company but they barely make any money, hell they lose money about half the

Massive "no shit"/"told you so" moment. ARM is an incredibly important foundational company but they barely make any money, hell they lose money about half the time, including during the pandemic, despite absolutely crazy electronics sales! Softbank isn't a charity, they want $40b+ valuation on a company that makes $0 +/- 1.5b ish. They want almost double what they paid for it, in fact. The only companies willing to pay that much money are the ones who see some kind of strategic synergy with owning ARM. NVIDIA could have gotten CUDA into every smartphone's default graphics IP, and become kingmaker licensing one of the most important ISAs in the world. Not only would they not kill that "just to sell some more tegras", but actually it likely would have been a watershed moment of NVIDIA moving towards licensing their IP to external companies. This would have been the start of the pivot towards semicustom - just like AMD already does. People just don't want to accept that because of "green man bad syndrome". But NVIDIA isn't exactly what people always present them as either - they are moving towards open linux drivers, they opened up to adaptive sync, both of which people would have (and did) defend to the death would absolutely never happen, hell would freeze over. And now that they are, still, nothing is ever good enough - well they have a closed userland, they have blobs (which so does AMD, in linux-firmware), they have a security processor! Even if those things are also true of other brands... the green man bad field means it'll never be good enough. There were always two alternatives to NVIDIA buying it - another sugar-daddy with business synergy, someone like TSMC ("bundle deal if you get it fabbed here") or Samsung or someone like that that could pair it with some other aspect of their business. Or finally, Softbank just makes it actually turn a profit. Who pays for that? Well, big tech, and probably ultimately consumers too. Companies like Google and Amazon and Tesla have grown fat on the "pay almost nothing for this ISA, bolt on our accelerator, sell it for billions" while ARM gets nothing and that's never been sustainable in the long term. Companies like HDMI already do make private tech that is licensed per-device, and again, you can't even have HDMI 2.1 in linux on AMD because it's closed-source and they won't license it in a compatible way, while NVIDIA can offer support because of those blobs. Qualcomm requires similar per-device licensing on their modems too, and wireless/modem blobs are some of the most notoriously opaque/blobby peripherals too. People deal with it, pay the licenses, and move on. https://gitlab.freedesktop.org/drm/amd/-/issues/1417 https://gitlab.freedesktop.org/drm/amd/-/issues/1417 And the "big tech forms a consortium and buys it and runs it as a nonprofit" was always a pipedream. Never going to happen, and none of the companies involved have any incentive to push tech forward in the same way as, say, Apple does. They just need it to be Good Enough And Cheap, and it already is. And scraping up $40b to keep it cheap is a bit of a mixed bag. Well, you didn't get NVIDIA. You did get big ARM layoffs (just as Jensen warned, and just as people scoffed at, ARM UK dropped 20% of their staff) and big price increases. Just as predicted. /shrug https://www.tomshardware.com/news/arm-uk-jobs-cut https://www.tomshardware.com/news/arm-uk-jobs-cut And yeah RISC-V exists and it's good to have an alternative to x86 and ARM too. There's a bit of a lifecycle here where they tend to get stagnant over time, maybe RISC-V being open-source will help that. But it also is going to be hell for ISA compatibility across vendors (a lot of the codes are deliberately left "vendor-defined") and there's a long way to go, RISC-V is where ARM was 10 years ago, and after companies pay to R&D to bring it up to speed, they're going to want a return on their investment too.