HP unveils world's thinnest RTX Spark laptops: 12K video editing, 128GB RAM, all-day battery

[Redaktion]

HP just announced the OmniBook Ultra 16 and OmniBook X 14, the world's thinnest laptops powered by the new Nvidia RTX Spark superchip. Featuring up to 128GB of unified memory and Blackwell GPU architecture, these next-gen AI PCs are designed to run complex local agents and advanced creative workflows for developers and power users, launching later in 2026.

https://www.notebookcheck.net/HP-unveils-world-s-thinnest-RTX-Spark-laptops-12K-video-editing-128GB-RAM-all-day-battery.1312200.0.html

[Julian M]

The funny part is that Nvidia isn't only a gaming hardware, but the hype is falsely very much on the gaming potential of these new chips - which they are not designed for.

So these are for AI/DL workflows and what not, fine. More hardware for the AI bubble, fine. It's priced accordingly too, fine.

Not sure this makes sense for a video editing workstation vs a Mac Studio, especially with Windows on ARM sucking so much and linux having little pro-grade editing software support, but hey to each their own.

So why are we talking about it as if it's a mainstream/prosumer product that will revolutionalize the market? It's Nvidia, that's it.

[Utkarsh Tanwar]

Funny enough as they have marketed themselves as Macbook Pro rivals as we are well aware of Windows RAM hungry Features and bloatware plus at Apple side of things Everything is designed by Apple from chip to software to hardware which help them make it more sleek and fast.

Waiting for real life benchmarks compared to macs.

[Moat]

...but the hype is falsely very much on the gaming potential of these new chips - which they are not designed for.
...
So why are we talking about it as if it's a mainstream/prosumer product that will revolutionalize the market?

It's a Blackwell GPU, what do you mean by not designed for? The same driver stack that is used other tasks can also be used for gaming just like other geforce graphics card.

And it's because Nvidia is a massive enough company that are able convince an entire industry of developers, publishers, ISVs to optimize & port their software/games to arm. Anti-cheat systems ported for online multiplayer working on it. Entire creative suite of audio video editing apps optimized for it. They're building pretty much a moat / entire ecosystem.

No other company can really push for arm this hard for arm adoption besides Apple and them.

Anyone else who does arm hardware and nobody really cares. You could argue Qcomm was doing it but let's face it, progress was painfully slow. For several reasons. Even now 2 years later there are still major issues. You can say Nvidia is piggybacking a lot of the initial and previous hard work done between Qcomm and MS. Maybe true but it is what it is.

The people who have tested DGX spark, both on Reddit and YouTube have all said the experience for running windows games on arm is light years ahead anything else they've seen in arm land. Despite the Nvidia issues, despite the running through translation layers, despite the CPU overhead. Still a million times better.

And that was on Linux unofficially, with no help from Nvidia or Microsoft like these companies are doing now collaborating and working together on.

It's revolutionary because halo has 4 hr of actual real use battery life. Intel despite it's best efforts now have a horrendous track record when it comes to GPU driver support.

Nvidia has a solid track record of support. They're still updating Nvidia shields from like the Meiji period. Lmao, I jest but ffs, they need to let it die.

Hope that answers your question.

[Gino188]

Lol. How expensive are these laptops going to be?

[Guesstimation]

Lol. How expensive are these laptops going to be?

The last known retailer leak showed these prices for n1x:

€3000 for 32 GB
€4000 for 64 GB

[RobertJasiek]

€3000 for 32 GB
€4000 for 64 GB

So €1000 for the second 32 GB or €2000 for 64 GB RAM. Now, if that is HBM, it may explain high prices partially. However, we all know that RAM prices are inflated by more than the factor 4. So in normal times, 64 GB RAM as HBM should cost ~€400 - 500 or ~2.5x the price of DDR5. This would be at least somewhat reasonable with total computer prices of then

€2200 ~ 2250 for 32 GB,
€2400 ~ 2500 for 64 GB

or less because currently some other components are also inflated. Nobody should pay more but wait for the hardware price boom to collapse. Mind you, if it has HBM indeed.

[Guesstimation]

So €1000 for the second 32 GB

Yes.

if that is HBM

It uses LPDDR5X. I believe 9400-9523 MT/s or around this speed.

[RobertJasiek]

Ok, so why do Nvidia's Spark PR sheets list HBM as a feature? Lie?

[Guesstimation]

Where? Link? :S

[jase]

It's a Blackwell GPU, what do you mean by not designed for? The same driver stack that is used other tasks can also be used for gaming just like other geforce graphics card.

Lol, just because it can game doesn't mean it's designed for gaming. It very much is NOT.

It only has 300GB/s of bandwidth, much lower than RTX 5070 mobile, and shared with 20 CPU cores at that. While iGPUs/APUs have been optimised for more bandwidth efficiency, I simply can't see it beating RTX 5070 mobile by that much, yet it has 6144 CUDA cores, more than even 5070 Ti Mobile. These extra CUDA cores can't really provide any gaming performance. So why would they not just reduce it to, say 5888 CUDA cores for better yield?

Also the silicon is on 3nm yet measures over 173㎟, GB205 only measures about 190㎟ without the GDDR7 PHY/MC, 24MB cache and one TPC. Let's call NVLINK and PCIe die area even. See the problem? There is virtually no die shrink between 4nm and 3nm. Nvidia must have done something to make it bigger than just a stripped down GB205 on 3nm.