RTX Spark Laptop Guide: UK Prices, Specs and Is It Worth It for AI?

An RTX Spark laptop is a new type of Windows laptop built around NVIDIA’s N1X chip, which combines an Arm-based CPU, a Blackwell GPU and up to 128 GB of shared memory in one package. The first models went on sale on 16 October 2026, and NVIDIA is pitching them as laptops that can run large AI models locally. This guide explains what RTX Spark is, which laptops use it, UK and US prices, what you can realistically do with one for AI work, and the trade-offs you should understand before spending £2,599 or more.

Last updated: 16 October 2026. Specifications come from NVIDIA’s product page and announcement. UK prices were checked on the Microsoft UK Store and Currys, and US prices come from Windows Central’s launch list, all on 16 October 2026. We have not tested these laptops ourselves; where we mention early performance figures from other sources, we say so.

What is an RTX Spark laptop?

NVIDIA announced RTX Spark at GTC Taipei on 31 May 2026. Instead of pairing a separate Intel or AMD processor with a discrete graphics card, RTX Spark laptops use the N1X system-on-chip, which NVIDIA co-designed with MediaTek. The CPU uses the Arm architecture, and the CPU and GPU share one pool of fast LPDDR5X memory, similar in concept to Apple’s unified memory.

That shared memory is the key to the AI pitch. On a conventional gaming laptop, AI models must fit in the GPU’s own memory, often 8–16 GB. On an RTX Spark laptop with 128 GB, the GPU can use most of that memory, so much larger models fit.

RTX Spark laptop specs

NVIDIA lists two N1X configurations:

NVIDIA says RTX Spark can run AI models of up to 120 billion parameters with context windows of up to 1 million tokens locally. It supports CUDA, TensorRT and the NVFP4 format natively, and includes the OpenShell runtime for AI agents, which we explain in our NVIDIA OpenShell guide. For games, it supports DLSS 5 and Reflex 2.

Which RTX Spark laptops are available?

NVIDIA’s launch partners are:

• ASUS ProArt P14 and ProArt P16
• Dell XPS 16
• HP OmniBook X 14 and OmniBook Ultra 16
• Lenovo Yoga 9n and Yoga Pro 9n
• Microsoft Surface Laptop Ultra
• MSI Prestige N16 Flip AI+

Acer and Gigabyte are expected to follow with desktop systems. Availability differs by country and model, so not every laptop is on sale in the UK or EU on launch day.

RTX Spark laptop price in the UK

At the time of writing, the clearest UK pricing is for the Microsoft Surface Laptop Ultra, available to pre-order on the Microsoft UK Store (prices include VAT):

Currys also listed the entry Surface Laptop Ultra (24 GB, 512 GB) at £2,599. The Surface has a 15-inch 3K PixelSense Ultra display with up to 2,000 nits HDR brightness, weighs about 2.0 kg, and has Wi-Fi 7, HDMI 2.1b and three USB4 ports. Microsoft claims up to 15 hours of video playback. A charger is not included in the box, although Microsoft was including a 140 W adapter with pre-orders.

UK prices for the other brands were not consistently listed when we checked, so compare retailers before buying.

RTX Spark laptop prices in the US

For reference, Windows Central reported these US starting prices:

High-memory configurations cost far more. PC Gamer reported about $7,000 for an ASUS ProArt P16 with 128 GB and 2 TB, and about $4,300 for a 64 GB Surface Laptop Ultra. US prices exclude sales tax, so do not convert them directly to pounds or euros.

What can you do with an RTX Spark laptop for AI?

Memory decides what you can run. As a rough guide:

• 24–32 GB models suit small and mid-size local language models, AI image generation and AI features in creative apps.
• 64 GB models can handle larger quantised language models and heavier image or video workflows.
• 128 GB models are the ones NVIDIA’s 120-billion-parameter claim applies to.

Run local language models

Tools such as Ollama and LM Studio let you run open models privately on your own machine. Our guide on how to run an LLM locally walks through the setup. Check that your chosen tool has a native Windows on Arm build with CUDA support before you rely on it.

Note that even the 128 GB model cannot run the largest open-weight models. Meta’s new Muse Spark Open, for example, needs roughly 700 GB of GPU memory, so you would use it through an API instead. Smaller models such as Muse Glimmer 30B are a realistic local fit.

Use local models in coding agents

Developers can point open-source coding agents at a local model server so code never leaves the laptop. We compare that approach with a cloud agent in OpenCode vs Claude Code.

Generate images and video locally

Node-based tools like ComfyUI benefit from a large memory pool. See our ComfyUI tutorial for a starting point, and our roundup of the best AI video generators if you would rather use cloud tools.

Early performance figures

Formal independent reviews were still emerging at launch. One early roundup on dev.to summarised third-party reports of about 58–89 tokens per second on a 35B model, 63 tokens per second on a 120B model, and roughly one hour of battery life under sustained AI load. We could not verify the original sources, so treat these as indicative only and wait for full reviews.

Limitations and trade-offs

• Arm-based Windows. RTX Spark uses an Arm CPU. Most mainstream apps now run, but some older software, games with anti-cheat systems and specialist drivers may need emulation or may not work. PC Gamer notes the entry models rely heavily on emulation for some games.
• Memory is soldered. You cannot upgrade it later, so buy the capacity you need for AI work now. For serious local LLM use, 24 GB is limiting.
• Price. Starting at £2,599 in the UK, these are premium machines, and the 128 GB models cost far more.
• Battery under AI load. Sustained AI workloads use a lot of power, and early reports suggest battery life drops sharply.
• Gaming is not the focus. Early expectations put gaming performance roughly around an RTX 5070 laptop, which is good but not top tier.

RTX Spark laptop vs MacBook for AI

The most common alternative for local AI on a laptop is a MacBook Pro with Apple Silicon, which also uses unified memory. The main differences are software and ecosystem:

• Choose RTX Spark if you need Windows, CUDA, TensorRT or NVIDIA-specific tools, or if your team’s AI stack is built on NVIDIA.
• Choose a Mac if you are happy with macOS and tools built on Apple’s MLX framework, or you value long battery life and mature Arm software support.

Compare equal memory capacities, because memory determines which models you can run on either platform.

Who should buy an RTX Spark laptop?

• Good fit: developers and data scientists who want to run and test open models privately on Windows; creative professionals using AI-heavy tools; organisations with strict data policies that prefer local processing.
• Probably not: people who mainly use ChatGPT, Claude or Gemini in a browser; budget buyers; and gamers who want maximum frame rates per pound.

How to choose the right RTX Spark laptop

1. Decide your largest model. If you want 70B-plus models locally, aim for 64 GB or 128 GB.
2. Pick the chip tier. The 20-core / 6,144-core N1X gives more GPU performance and access to 128 GB.
3. Check your software. Confirm that the apps you depend on support Windows on Arm.
4. Choose screen size and weight. 14-inch models are more portable; 16-inch models usually have more cooling headroom.
5. Compare UK retailers. Check stock, return policies and whether a charger is included.
6. Wait for reviews if you can. Independent testing of thermals, battery life and app compatibility will be more reliable than launch claims.

FAQ

How much is an RTX Spark laptop in the UK?

The Microsoft Surface Laptop Ultra starts at £2,599 including VAT on the Microsoft UK Store, rising to £5,899 for the 128 GB model.

Can an RTX Spark laptop run ChatGPT offline?

No. ChatGPT is a cloud service. You can run open-weight models offline instead, using tools like Ollama or LM Studio.

Is RTX Spark good for gaming?

It supports DLSS 5 and Reflex 2, but it is designed mainly for AI and creative work. Some games may run through emulation or may not support Arm.

Can I upgrade the RAM later?

No. The memory is part of the chip package, so choose your capacity at purchase.

When did RTX Spark laptops go on sale?

Retail availability began on 16 October 2026, with some HP models following on 4 November in the US.

Conclusion

An RTX Spark laptop is the most serious attempt yet to put data-centre-style AI capability into a Windows laptop, and the large shared memory pool is a real advantage for running local models. It is also expensive, Arm-based and, at launch, largely untested by independent reviewers. If local AI on Windows is central to your work, a 64 GB or 128 GB model is worth shortlisting. If you mainly use cloud AI tools, a conventional laptop will serve you well for far less.

Sources: NVIDIA RTX Spark product page; NVIDIA announcement; Microsoft UK Store: Surface Laptop Ultra; Currys listing; Windows Central; PC Gamer; dev.to early review roundup.