Nvidia’s ARM acquisition is stalled

[Lakados]

Nvidia started hiring quite a few high-profile positions for RISC-V development right around the time the ARM deal surfaces, I reckon it's a move to both kickstart their own architecture research and to be able to integrate ARM into their portfolio quickly. Now that the deal died, what's to prevent Nvidia from doing in-house chip design and licensing or selling full SoC solutions with RISC-V based CPUs?

Apple is also dabbling in RISC-V CPU design, Samsung already uses RISC-V for their ISP and 5G radio, presumably for their Exynos line.

It feels like every big player is preparing to jump ship from ARM if Nvidia closes the M&A and decides to be a dick. Now that the deal fell through, it'll be interesting to see whether they'd continue with the ARM licensing or just go fully in-house.

RISC-V does real time data manipulation very well. Storage controllers, memory management interfaces, networking controllers. These are all things that we are going too see or already see RISC-V chips doing and with increasing frequency. The open nature of its ISA allows for extremely specialized processors with a smaller physical, thermal, and electrical foot print. Samsung uses them on all their SSD’s and DDR5 modules. RISC-V is unlikely to become an architecture we see directly powering our phones, tablets, or PC’s any time soon, if at all. But it can do substantial work at taking the load off those chips.

Specialized chips for specialized jobs are again going to become a thing because performance gains on general computing is hitting a wall, gains come slower and what gains do come are more expensive and complex.

PCIE5 and the recently demo’d PCIE6 have bandwidth to the degree that some things can again be moved off the CPU and GPU cores to a net performance gain while decreasing overall core cost and complexity.

And yes RISC is now at a stage where it’s fast enough to take on many small loads, you could easily use it for smaller networking switches, sensors, and many consumer devices (mostly children's toys). It’s application in the automotive field alone could be worth billions especially with the ever increasing amount of data their various sensors are generating which needs real time analysis. The possibilities for agriculture as well given the ever increasing amount of automation on modern farms isn’t anything to sneeze at.

So yes the industrial applications for RISC-V are numerous and profitable should they get the right products to market at the right time.