There is a big trend away from CISC to RISC even in conventional computers and datacentres. How far this will go is up to debate, but for now it’s got a lot of momentum (Macbooks going to ARM, Microsoft ARM emulation, more ARM instances in datacentres). From this perspective ARM seems to be in a great place.
Ultimately it seems that logic design is going to be dominated by Nvidia, AMD, Intel and ARM’s customers. While logic foundries will be split between TSMC and Samsung (perhaps also Intel).
While Nvidia is leading in GPU it lacks on the CPU side (compared to AMD, Intel which offer both) so taking on ARM would be a natural way to get into the CPU space. Nvidia will have more control over the CPU space if they successfully acquire ARM, they will have control over the instruction set and could use this to their advantage. It also fits with their general plan of having all-things logic (particularly AI).
If Nvidia does mishandle ARM as well as risking regulatory backlash it could push increased adoption of the open RISC-V standard. It’s one thing to pay licensing fees and have over to control to the relatively neutral ARM but rival companies may not want to be beholden to a competitor (Nvidia).
I don’t love the acquisition, I’d probably invest in an Arm IPO but I’m not sure it has the same value here.
Tangent
I don’t love Nvidia at this price, CUDA is basically the moat for Nvidia but it’s having to do a lot of heavy lifting to support the market cap right now and I’m not sure it can last forever.
(I’m long TSMC-AMD so I’m certainly biased against Samsung-Nvidia at the moment)