The rumored specs for Nvidia's expected first PC CPU are starting to float around the leaky corners of the x, and the latest claim is that the green team is already working on a 2025 chip for the processing part, but the strange rumor is that Nvidia will use Intel's own fab to create yet another silicon nail in the x86 coffin.

Nvidia Creating an ARM-based processor after all the AI PC noise of the year makes absolutely sense, especially with historical rumors of the green team's partnership with MediaTek in creating a reference design for laptops. Allowing the company to create fully built laptops around its own processors and Gpus has traditionally made discrete Gpus thin and light La

So Nvidia is building a System on chip (SoC) that uses Arm's processing core and incorporates its own graphics silicon as an integrated GPU for low-power/high-performance chips, making it easier for other laptops. It may sell out to bookmakers. It feels like it's very simple.  The spec rumor comes from @XpeaGPU on X, which claims to be a Cortex X5CPU cluster with Blackwell iGPU and will have sport LPDDR6 memory in the package.

It also states that it will be built using TSMC's N3P nodes, but this is a claim quickly contested by another fellow x rumourer, @Kepler_L2. They claim that Nvidia will actually use the Intel3nm manufacturing process.

Intel doesn't have a 3nm node itself in particular, so it could simply mean Intel3. But it could also mean an Intel3-T update that includes the Foveros Direct3D (PDF warning) stacking feature to bring together chips and wafers from different foundries, which means Nvidia will continue to use the TSMC N3P to build the Blackwell GPU core, and Intel will use the ARM Cortex X5 core and the base on which they all sit. You can create tiles. 

There have been rumors for some time that Nvidia could see INTEL as an alternative foundry partner for TSMC, and that different GPU lines could be manufactured at Intel, which would have been strange given the different packaging methods used by the two foundries and would mean extra work porting designs from different production facilities. If you're just using Intel to build processing components, and perhaps an interposer, then it's suddenly more plausible.

All this, however, does not mean Nvidia's custom cores, which both Qualcomm and Apple have gone down its route for their own new silicon. But that's not necessarily bad for a green team. Plucking pre-made cores from Arm's catalog means you don't have to mess with designing your own processor cores - this is the same simple process that Nvidia Drive platform is used in the automotive sector using standard ARM Neoverse cores. 

Apple touted it as something its own custom chip brought to the table, at least in part because Rosetta was so effective at translating the x86 app for the M-series chips. The expectation is that Qualcomm may have done a similar thing with the Orion core for Snapdragon X Elite processors, particularly with Microsoft claiming Prism is just as good for Windows as Rosetta was for iOS.

Nvidia's ability to combine the Blackwell GPU architecture with the ARM Cortex X5CPU core, codenamed Blackhawk. Given, there is also a certain pleasing symmetry. 

So, when can we see Nvidia's Cpus hit the market?"To see its place in the AI PC market, according to Michael Dell (not a clever name for Dell Technologies)."Next year he joked in a Bloomberg interview that he was sitting with Jen-Hsun Huang.

But I don't know if this means Nvidia laptops, just like Apple laptops, which control the entire hardware ecosystem. The Green team may only be planning to create Socs for other laptop manufacturers to eventually incorporate into the thin and lightweight market without discrete Gpus, but I'm sure it's hard for Nvidia to resist creating at least a MacBook-y reference design.

There are also rumors that they are working with MediaTek on portable gaming SoCs outside of potential locations for the next generation Nintendo Switch. So, again, we want a new shield.