NVIDIA’s new RTX Spark laptops are basically the moment the “AI PC” hype stops being a slide deck and turns into actual hardware you can buy. They are NVIDIA’s first serious push into consumer laptop chips, built around a new Arm-based “superchip” that tries to fuse CPU, GPU, and AI acceleration into a single platform – and the first wave of machines looks like a cross-section of the Windows laptop market, from creative workhorses to 2-in-1s and XPS-style flagships.

What makes this launch interesting is not just that NVIDIA is finally stepping directly into your next laptop’s silicon. It is that Spark arrives right in the middle of the AI PC land grab, where Qualcomm’s Snapdragon X, Intel’s Lunar Lake, and AMD’s latest Ryzen AI parts are all trying to convince you their version of “on-device AI” is the one that matters. RTX Spark is NVIDIA’s answer: an Arm-based chip with 20 CPU cores, 6,144 GPU cores, and support for up to 128GB of LPDDR5X memory in its flagship configuration, essentially mirroring the GB10 chip used inside NVIDIA’s DGX Spark “personal AI supercomputer.” In other words, NVIDIA is trying to bring a slice of its data center DNA into something that looks like a regular premium laptop.

The company is not launching this alone. Microsoft, ASUS, Dell, HP, MSI, Lenovo, Acer, and Gigabyte are all on deck with RTX Spark systems slated for this fall, with NVIDIA saying more than 30 models are in the pipeline. For a first generation, that is a big statement: this is not a niche dev kit or a single halo product. It is a full ecosystem bet.

Under the hood, the Spark story is about consolidation and efficiency. Instead of bolting a discrete GPU onto a conventional x86 CPU, Spark puts Arm CPU cores and a serious GPU on the same package, tuned for AI workloads and wrapped in LPDDR5X memory. NVIDIA’s flagship configuration combines those 20 CPU cores and 6,144 GPU cores with up to 128GB of unified memory, which is a wild number if you have been living in the 16GB or 32GB laptop world. That kind of memory ceiling matters for running large models locally, but NVIDIA is also planning cheaper, more modest SKUs with as little as 16GB for more mainstream machines.

NVIDIA positions this as a “personal AI supercomputer” class experience for consumers, echoing what it is doing with DGX Spark on the pro side. The idea is straightforward: instead of treating AI as something that only happens in the cloud, you pull more of that inference work down onto the device, whether that is running multimodal assistants, creative tools with generative features, or games that lean on AI-enhanced graphics. At Computex, NVIDIA leaned into claims of “all-day battery life” and designs as slim as 14mm for RTX Spark laptops, signaling that this is meant to compete directly with thin-and-light Windows machines, not only chunky workstations.

Of course, the chip is only part of the story. The more interesting part is what the OEMs are doing with Spark. So far, we have a pretty clear picture of Microsoft, ASUS, and MSI’s plans, plus a more mysterious tease from Dell, HP, and Lenovo.

Microsoft’s RTX Spark-powered Surface Laptop Ultra is the headline act on the Windows side and arguably the most “mainstream” expression of the platform. Microsoft is putting Spark into a 15-inch laptop with a mini-LED PixelSense Ultra touchscreen that the company calls “the brightest display we’ve ever shipped,” promising up to 2,000 nits of peak HDR brightness and 262 pixels per inch with high-precision color accuracy. In practical terms, that is squarely aimed at people who care about HDR content, HDR gaming, or just having a laptop that holds up outdoors.

The Surface Laptop Ultra also looks like the first Spark system to aggressively tackle the everyday creator and power user. You get HDMI, USB-C, USB-A, an SD card slot, and a headphone jack, plus what Microsoft says is the largest haptic touchpad it has ever shipped on a Surface. Microsoft is already calling it “the most powerful Surface Laptop ever built,” which is marketing, sure, but it also hints at how tightly this thing will be integrated with Windows features like Copilot, local AI agents, and potentially even NVIDIA’s own RTX-accelerated experiences. What we still do not know is the full spec stack – exact CPU/GPU bins, storage, and pricing are all still under wraps – but this is clearly intended as the reference Spark laptop most people will recognize.

ASUS, on the other hand, is going straight for creatives with its ProArt line. The new ProArt P16 and P14 drop later this year, and ASUS has already shared more about them than most of NVIDIA’s launch partners. Both machines are Spark-powered, both have touch-enabled displays, and both ditch the old ProArt dialpad wheel in favor of a haptic touchpad implementation that still aims to give you fine-grained control in creative apps.

The ProArt P16 is the bigger, more “desktop replacement” option. It has a 16-inch 4K OLED panel with a 16:10 aspect ratio and 120Hz refresh rate, rated for up to 1,600 nits of brightness with badges like TUV, Pantone, Trueblack 1000, and Dolby Vision hanging off the spec sheet. ASUS says you can configure it with up to 128GB of LPDDR5X-9400 memory and up to 2TB of Gen 5 SSD storage across two M.2 slots, plus a 99.9Wh battery and a healthy port selection – three USB-C (Type 4), one USB-A 3.2, HDMI 2.1, a headphone jack, and a full-size SD card slot. On paper, that is the kind of machine that can replace a traditional mobile workstation for video editors, 3D artists, and photographers who want to run generative workflows locally.

The ProArt P14, meanwhile, is the compact sibling. It shrinks the chassis down to a 14-inch 3K OLED at 120Hz with the same 16:10 ratio, and still hits up to 1,600 nits with the same color certifications. ASUS caps storage at up to 1TB on a single Gen 5 SSD here, and pairs that with a 90Wh battery and the same triple USB-C, single USB-A, HDMI 2.1, headphone jack, and SD card setup. The headline spec is that you can still go up to 128GB of LPDDR5X-9600 memory, which is overkill in the best possible way if your workflow leans on heavy AI tools or massive timelines. For creators, these two ProArt Spark machines look like the most coherent “MacBook Pro alternative” story in the lineup right now.

Then there is MSI with the Prestige N16 Flip AI+, a name that sounds like someone fed a marketing generator too many focus group notes but represents something genuinely different in the Spark lineup. It is a 2-in-1 with a 16-inch 3840 x 2160 OLED display, promising “over 1,000 nits” of brightness, 100 percent DCI-P3 coverage, Calman verification, and Delta E under 1 color accuracy – the kind of numbers that matter if you are grading footage or doing color-critical work. It also packs a 99.9Wh battery, which is about as large as you can legally ship in a laptop for air travel, suggesting MSI really wants this to be a do-everything machine.

What makes the Prestige N16 Flip AI+ interesting is its accessory story. It supports the MSI Nano Pen, a stylus co-developed with Microsoft that doubles as an AI microphone for Copilot, and can fully charge in about 30 seconds. That suggests MSI is thinking about Spark not just as compute, but as part of a new UI layer where you talk, draw, and gesture your way through AI-assisted workflows. Port-wise, the images show at least two USB-C ports, an HDMI port, and a headphone jack, though MSI has not confirmed the full IO layout yet. We also don’t have firm details on storage, memory tiers, or pricing – which is becoming a theme across the whole first-gen Spark wave.

The other big OEMs are playing it more coy in public, but they are clearly in. Dell is preparing a new XPS 16 refresh with RTX Spark, and the clearest look we have so far shows a design with three USB-C ports, an HDMI port, a headphone jack, and an SD card slot. That screams “creator and prosumer” again, and it fits the pattern of NVIDIA seeding Spark into machines that already have credibility in photo, video, and coding circles. Lenovo’s upcoming Yoga Pro 9n and HP’s OmniBook X 14 and Ultra 16 are also confirmed as Spark devices, but both companies are holding back detailed specs and pricing until closer to launch.

HP has at least shown off the sides of the OmniBook X 14, revealing two USB-A ports, two USB-C ports, an HDMI port, and a headphone jack – a refreshing spread in a world where many thin-and-lights lean too hard into USB-C-only minimalism. NVIDIA says all RTX Spark laptops will offer “all-day battery life” and bodies as slim as 14mm, which suggests that even these more traditional clamshells are being tuned for efficiency rather than raw, desktop-class performance at any cost. Meanwhile, Acer and Gigabyte are also on NVIDIA’s list of upcoming Spark partners, though there are no public model names or timelines yet beyond a general “later this year” window.

For buyers in the US, the obvious question is: where does Spark sit in a landscape already crowded with “AI PCs”? Right now, Qualcomm’s Snapdragon X laptops are pushing Arm-based efficiency and tight Microsoft integration, while Intel and AMD push x86 compatibility and NPUs as the backbone of Windows’ AI features. RTX Spark complicates that narrative. NVIDIA brings a GPU-first mindset and a massive existing ecosystem of CUDA-accelerated apps and RTX features, which could give it an edge in creative and gaming scenarios if developers lean in. The promise is that you can run heavier models locally, get better AI-enhanced graphics, and still keep battery life in check.

But there are unanswered questions that will matter when you are actually about to spend money. We don’t yet know how RTX Spark Windows laptops will handle the messy reality of legacy x86 apps, especially compared with Qualcomm’s emulation story or the native x86 options from Intel and AMD. We don’t have firm US pricing, which will decide whether these systems end up as premium halo products or actually compete in the midrange where most buyers live. And we don’t yet have independent benchmarks to show how Spark’s CPU cores stack up against its rivals when you are doing boring but important stuff like compiling, exporting, or running multiple VMs.

What is clear is that NVIDIA is treating Spark less like a side project and more like a genuine client platform, backed by big OEM commitments and a launch tied to marquee events like Computex. If you are already living in Adobe, Resolve, Blender, or a maze of AI tools, these machines look like they are designed for you first, not for the enterprise IT buyer. And because this is NVIDIA, you can expect a lot of marketing around RTX-accelerated apps, “personal AI supercomputers,” and integrations with its broader ecosystem of software and services over the next year.