Tuesday, July 15th 2025
NVIDIA's N1x CPU Hits New Roadblock: Launch Pushed to Late 2026
NVIDIA's long-awaited entry into Arm-based laptop CPUs has hit yet another obstacle, according to insiders close to SemiAccurate. Despite publicly declaring the N1 and its sibling N1x is in full production, the company now faces fresh engineering challenges that threaten to push shipping dates as far back as late 2026. Sources speaking to SemiAccurate describe that the newest issue may require a modification to the actual silicon. This setback follows an earlier hiccup, reported in early 2025, when some subtle flaws emerged during initial validation. NVIDIA engineers managed to correct those without a respin, restoring confidence and nudging the timetable back to early next year.
Performance teasers we spotted a month ago showed a prototype "NVIDIA N1x" scoring 3,096 in single‑thread and 18,837 in multi‑thread tests on Geekbench 6.2.2. The sample chip, believed to power an HP "8EA3" development notebook with 20 logical cores clocked at 2.81 GHz and backed by 128 GB of RAM running Ubuntu 24.04.1 LTS, suggested a significant performance in big.LITTLE arrangement built from standard 10 Cortex‑X925 performance cores and 10 Cortex‑A725 efficiency cores. The N1x's integrated graphics and neural‑processing unit would close the gap with Qualcomm's Snapdragon Elite series and even Apple's M3‑class silicon. Now, with the latest issues looming, OEM partners may have to recalibrate their Windows laptop plans. NVIDIA will need to balance the risk of further delays against the need to demonstrate a polished, reliable experience before its CPU ambitions can truly challenge established players in the laptop arena. Late 2026 now appears set to become the new milestone for when the first N1‑powered notebooks might finally reach store shelves.
Source:
SemiAccurate
Performance teasers we spotted a month ago showed a prototype "NVIDIA N1x" scoring 3,096 in single‑thread and 18,837 in multi‑thread tests on Geekbench 6.2.2. The sample chip, believed to power an HP "8EA3" development notebook with 20 logical cores clocked at 2.81 GHz and backed by 128 GB of RAM running Ubuntu 24.04.1 LTS, suggested a significant performance in big.LITTLE arrangement built from standard 10 Cortex‑X925 performance cores and 10 Cortex‑A725 efficiency cores. The N1x's integrated graphics and neural‑processing unit would close the gap with Qualcomm's Snapdragon Elite series and even Apple's M3‑class silicon. Now, with the latest issues looming, OEM partners may have to recalibrate their Windows laptop plans. NVIDIA will need to balance the risk of further delays against the need to demonstrate a polished, reliable experience before its CPU ambitions can truly challenge established players in the laptop arena. Late 2026 now appears set to become the new milestone for when the first N1‑powered notebooks might finally reach store shelves.
63 Comments on NVIDIA's N1x CPU Hits New Roadblock: Launch Pushed to Late 2026
And you simply cannot tell me that nGreedia will not sell this thing for an ARM and a leg!!! It also will not be upgradeable. Windows ARM is a joke, and compatibility issues will be a pox on the platform.
I see no reason why ARM should be in high-end consumer PC's and laptops.
On another note nvidia took notes of snapdragon when they launched and has quite a number of issues… not a bad move they need to build up more portfolio instead of being a gpu and Ai hardware provider and are branching into different hardware… soon we will see all nvidia laptops and desktops… team green fans will most likely throw their wallets at nvidia
On Twitter people talk that after Panther Lake Intel will bring this major evolution design to market.
So at the end of the day RISC-V will come at the end of everything, because
- there can be only one!
Nvidia has already it's plan B.
Didn't you know that actually Nvidia has it's own RISC-V cores?
Also!
The thing I'm trying to say in the previous comment is:
Everybody from Intel to Qualcomm will switch to RISC-V, because
- why paying ARM if you can get a thing which not only is free, but also rationally better?
There are clear benefits to APX, but it requires compiler modifications and recompilation of software to benefit from it.
Internally modern x86 cores are RISC-like anyway, so the whole RISC vs. CISC debate has long been obsolete due to that, and traditionally RISC architectures like ARM adding CISC-like features to their design.
At the moment RISC-V is even more fragmented than ARM, so software support is a problem.
Recently RISC-V RVA23 was deemed the minimum for the next Ubuntu Linux LTS version, basically making most current RISC-V hardware unusable with it.
It has compatibility issues? Yes, it does. Growing pains, and that's on Microsoft and their partners (to this date, Qualcomm only) to take care of.
My X1E device is way more agile than any Intel laptop I've ever had, principally when unplugged. Running on battery, even Lunar Lake loses more performance than the Snapdragons, so I won't get started on any other Intel or AMD arch. So I bet that when NV gets its N1(x) to market it can only be better than what QC has already brought, especially because Adreno drivers still suck big time.
How it's even possible?
How are you even supposed to take open source RISC-V & make it proprietary for internal use?
I agree that Arm designs have a major flaw for DIY desktop space and will never be able to chalkenge Intel and AMD until Apple, Nvidia and Qualcomm offer a truly modular and socketable CPUs, motherboards and other components. There is just as many mini-PCs and soldered systems they can possibly sell.
It's astonishing that Nvidia and Apple, some of the richest companies in the world, have no offer for DIY desktop segment. They have no courage to enter this segment and will always be on fringes with soldered systems, unless they change this.
Nothing is stopping vendors from making their core designs proprietary, almost all non-toy RISC-V SoCs to date are proprietary. Chiefly because you need a high performance internal bus, PCIe controllers, RAM controllers, USB controllers and all the other misc periphery that makes a SoC work. All those things require years and sometimes hundreds of millions to develop, especially cutting edge like PCIe 5.0 or DDR5. Nobody is going to give that up for free.
You keep banging on about stuff that your saying is going to become the standard. But that requires it to offer something over what we already have today, and for the masses to adopt it. It's not going to happen away from the datacentre dude. Not for at least another 15 years because it offers NOTHING TANGIABLE to consumers.
Any replacement to x86 has to be compatible, more performant, cheaper and just as expandable and upgradable. ARM has none of those things. RISC-V does have potential, but it has ZERO compatibility and is going to be slow after emulation and expensive.
Then the handheld.