ARM made an architecture and CPU/GPU designs for others. They really had no stake in the chip making game, instead having others - Samsung, Apple, Qualcomm, Huawei, Mediatek, Rockchip, etc make the chip (that being a list of competitors for Nvidia). Their business model was in making the ARM ecosystem available for all manufacturers (including Nvidia) in order to sell licenses. This, I think, is the primary reason ARM chips dominate at this moment.

Nvidia as a company is notorious for keeping a tyrannical grip on their customers, systems, & hardware. CUDA is a wonderful software stack, with massive hardware adoption... tied very strongly to Nvidia hardware. Nvidia GPUs are wonderful pieces of hardware... with frustrating Linux drivers, dogged resistance to open-source driver efforts, & a history of resisting mainstream collaboration & doing things themselves (their EGLStreams vs DMA-BUF/GBM, which, after half a decade+ of making Nvidia on Linux semi-useless, is finally showing signs of retreat. i'll allow that their devs may have had some points, perhaps?). Nvidia is one of the only companies with a proprietary interlink system, NVLink, whereas historically IBM (OpenCAPI), HP (GenZ), AMD (not anymore alas, with HyperTransport giving way to Infinity Fabric, so sad), great collaborations (CCIX/CXL), countless others promoted & used open interlinks, neutral technology that enabled & drove innovation & jump started the personal computer revolution ("gang of nine"[1]). Nvidia releases no documentation for their chips. If you buy a Nvidia computer- very powerful, energy efficient dev boards- Linux4Tegra is expected, woe be unto those who try to venture out & use the common Linux distros they already know & love.

In Stratchery's wonderful write up of how quixotic & weird & strange the Nvidia acquisition of ARM seems[2], Ben highlights CEO Haung's core interesting idea of this acquisition:

"Arm’s business model is brilliant. We will maintain its open-licensing model and customer neutrality, serving customers in any industry, across the world, and further expand Arm’s IP licensing portfolio with NVIDIA’s world-leading GPU and AI technology."

And I'd love to expound on how weird, how strange this sounds. But Ben's done it far better than I could:

"Notice that last bit: Huang is not only arguing that Nvidia will serve Arm customers neutrally, but that Nvidia itself will adopt Arm’s business model, licensing its IP to competitive chip-makers. It’s as if this is an acquisition in reverse: the $318 billion acquirer is fitting itself into a world defined by its $40 billion acquisition.

Color me skeptical; not only is Nvidia’s entire business predicated on selling high margin chips differentiated by highly integrated software, but Nvidia’s entire approach to the market is about doing what is best for Nvidia, without much concern for partners or, frankly customers. It is a luxury afforded those that are clearly best in class, which by extension means that sharing is anathema; why trade high margins at the top of the market for low margins and the headache of serving everyone?"

I would LOVE to see Nvidia acquire arm & gain a soul, figure out what it means to innovate, openly, to drive tech forward while working hand in hand with others to see it get adopted & grow & thrive, in ways far beyond their own limited control & vision. Nvidia is up to great things, has such an amazing mastery of high technology that so few share, building amazing chips, releasing amazing products. They are so integrative, building great chips (gpus & cpus both), building ways to make chips work together (nvlink), building consumer electronics (shield tablet, shield tv), building ultra-dense servers (dgx).

But they do not share. "Take what you can, give nothing back!" is how a lot of people feel about Nvidia, and on many days I share that view. It's so hard to imagine. It would be such a change.

Ben tails the end of his write-up with a wonderful question posed to Huang, why does Nvidia need to own ARM? Huang comes up with great answers, seems very thoughtful, and I am excited to see his future, as he talks about it happen. But it would not be Nvidia as we know it any more doing this. The company would need to be reborn, be something else. Huang seems to get that, he sees and says that, but it seems just so much more probable that the real end would be a miserly, sad fate for ARM, performance going up yes, but control & closeness shutting down the relevance & interestingness & availability & affordability.

[1] https://en.wikipedia.org/wiki/Extended_Industry_Standard_Arc...

[2] https://stratechery.com/2020/nvidias-integration-dreams/

ARM makes their money by licensing the design to other huge companies, who then integrate the processor with other peripherals and manufacture it. They have no interest in extracting money from the end user, instead the documentation is readily available. In comparison, Nvidia, for example, sells you a video card, then you have to pay them again for a license to do deep learning. Oracle requires the end user of Java to pay a license to use the JRE. The fear is that if Qualcomm or Nvidia buy ARM, they're going to turn ARM into a similar business model where they try to nickel and dime the end user for everything.