Every laptop on the shelf today is stamped with "AI", "Copilot+" or "NPU", and the same question keeps coming up: is that little chip worth the extra money? The question hits harder this year, because every dinar in a build now counts — the global memory crunch has pushed RAM and storage prices up sharply, and IDC expects PC prices to climb by as much as 8% through 2026. Paying extra for something you will never use is no longer an affordable luxury.
The good news is that the answer is clearer than you think, especially if you are shopping for a powerful gaming laptop. Let's break it down without the marketing fog.
What is actually happening: the NPU is not what you imagine
An NPU (Neural Processing Unit) is a small block inside the processor with one job: running AI tasks at the lowest possible power draw — mic noise removal on calls, camera background blur, live translation, and the quiet Windows features running in the background. It was never meant to replace your CPU or GPU; it exists to take the repetitive work off their hands so your battery survives the day. That is its real benefit: power efficiency, not raw muscle.
Here is the part most buyers miss. Microsoft set a bar of at least 40 TOPS (plus 16GB of RAM and 256GB of storage) before a machine can be called a "Copilot+ PC". The chips that clear that bar belong to the thin-and-light, long-battery class — Snapdragon X, Intel Core Ultra 200V, and AMD Ryzen AI 300, which reaches 50 TOPS.
The heavy gaming chips from the HX family — like the Intel Core Ultra 9-275HX found in most machines in this category — carry an NPU of only around 13 TOPS, well under Microsoft's threshold. That is not a defect; it is a deliberate decision. Intel spent the silicon area and the power budget on cores and performance instead of the NPU, because these machines already ship with a discrete GPU that does the job far better. And when you see big "platform TOPS" numbers in an ad, most of that figure comes from the GPU, not the NPU.
What this means for you in practice
In short: on a gaming laptop, the real AI work runs on the graphics card, not the NPU. Image generation, running language models locally with no internet, video editing and render acceleration, upscaling, and in-game frame generation all lean on the GPU and on how much video memory (VRAM) it has. That number decides what you can run and what you cannot; a model that does not fit in memory simply does not run.
So the rule is simple: do not pay extra for an "AI" sticker on the lid. Pay for three numbers — VRAM, RAM, and SSD capacity. The NPU? Treat it as a pleasant extra that comes free with the processor, not as a reason to change your buying decision.
This is where the 2026 factor comes in. The current crunch exists because AI data centres have absorbed most of the world's memory output. DDR4 and DDR5 prices have risen steeply, SSD prices have doubled in some tiers, and memory chips now account for more than 80% of the cost of building a graphics card — to the point that the RTX 5090 has passed $4,600 in the US market, and some manufacturers have started selling prebuilt desktops with no RAM at all. Even graphics card production is being prioritised for data centres over ordinary users.
All of it boils down to one piece of advice that will still hold two years from now: buy the RAM and storage you need on day one. Upgrading later gets more expensive, not cheaper, and a laptop GPU cannot be upgraded at all. A machine that ships with 64GB of RAM and 2TB of storage is worth more than it looks on paper.
How to choose: the right pick for your case
If you do serious local AI work. You want the largest VRAM pool you can reach, backed by plenty of system RAM so nothing chokes. The MSI Raider A18 HX pairs an RTX 5090 with 24GB of video memory with 64GB of RAM, 2TB of storage, a Ryzen 9-9955HX3D, and an 18-inch UHD+ display. This is as far as a portable machine goes with all three numbers at once.
If you want the best memory-to-price balance. The Lenovo Legion Pro 7 in its 5GAX configuration combines 64GB of RAM and 2TB of storage with an RTX 5080 carrying 16GB of video memory, on a WQXGA OLED panel at 240Hz. By 2026 market logic, the RAM and storage here are the real bargain — those figures have become expensive to reach.
If you want the strongest GPU without the bulk. The 17US version of the Legion Pro 7 gives you an RTX 5090 with 24GB of video memory alongside 32GB of RAM, 1TB of storage, and a 240Hz OLED screen. A sensible pick when gaming and rendering matter more to you than hoarding a huge library.
If your goal is the cheapest route into the 5090 tier. The Acer Predator Helios 16 AI brings an RTX 5090 with 24GB of video memory and an Ultra 9-275HX on a WQXGA OLED 240Hz panel, but with 16GB of RAM and 1TB of storage. Excellent value, as long as you accept now that a RAM upgrade is in your future — and at today's prices, budget for it upfront rather than later.
If you want a big screen you will stare at for hours. The MSI Vector A18 HX offers an 18-inch QHD+ 240Hz display with 64GB of RAM, 2TB of storage, an RTX 5080 with 16GB of video memory, and a Ryzen 9-9955HX. That extra canvas pays off in editing, coding and design, not only in games.
If gaming is your first and last priority. The HP Omen Max 16 carries an RTX 5070 Ti with 12GB of video memory, 32GB of RAM, 2TB of storage, and a 240Hz WQXGA screen. That is the sweet spot for most players: high performance without paying the halo-tier tax, with RAM and storage that will not embarrass you in two years.
If the budget is tight but you still want real VRAM. The MSI Vector 16 HX AI comes with an RTX 5080 holding 16GB of video memory, an Ultra 7-255HX, 16GB of RAM, 1TB of storage, and a QHD+ 240Hz panel, in a no-OS version that trims the cost. Note the irony: "AI" is right there in the name, yet the part doing the AI work is the graphics card, not the NPU.
The bottom line
The NPU is not a scam, but it is not a reason to pick one gaming laptop over another. In this class specifically, the NPU is small and its role is marginal, while every heavy task lands on the GPU. If you were shopping for a thin machine with long battery life running light AI tasks all day, the NPU would genuinely matter — but on a machine with an RTX 50 series card, the right question is not "how many TOPS?" but "how much VRAM, RAM and storage?".
And one piece of advice that stays valid a year from now: in a memory market that tightens every month, a laptop bought today with generous RAM and storage was bought at a price that will not come back. Do not buy something you plan to upgrade later; buy something that is enough from day one. At Jowa Al-Souq, delivery covers every Iraqi province with cash on delivery, so you inspect the machine in your own hands before you pay.
