Nvidia is no longer content being the chip inside someone else’s laptop. On October 16, 2026, the company confirmed that NVIDIA RTX Spark laptops are available today from six major PC makers, a launch that pairs Nvidia’s own Grace CPU and Blackwell RTX GPU inside a single Windows 11 machine for the first time. Starting at $2,599, the lineup marks the clearest step yet toward Nvidia selling not just the silicon, but the whole computing platform underneath it.
The announcement, detailed in Nvidia’s investor relations release, lists ASUS, Dell, HP, Lenovo, Microsoft and MSI as launch partners, each shipping its own RTX Spark-based model. That is a notable list: five of the six are Nvidia’s longtime GPU customers, now building machines around a chip platform Nvidia designed almost entirely in-house. For a company that built a $5 trillion-plus valuation on selling components to other people’s finished products, RTX Spark looks like the start of something different.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Nvidia Announced on October 16
The core news is straightforward: nine RTX Spark laptop models from six manufacturers are now on sale, running Windows 11 and built around what Nvidia calls the RTX Spark superchip. The confirmed lineup includes the ASUS ProArt P14 and ProArt P16, the Dell XPS 16, the HP OmniBook Ultra and OmniBook X 14, the Lenovo Yoga Pro 9n, the Microsoft Surface Laptop Ultra, and the MSI Prestige N16 Flip AI+ and Prestige N16. Pricing and configurations vary by region, but the entry point across the lineup is $2,599.
Physically, Nvidia says the RTX Spark laptop class runs as thin as 14 millimeters and as light as three pounds, spanning 14- to 16-inch screen sizes. That puts the category in direct visual competition with premium ultrabooks from the same six brands, except every one of these machines now carries Nvidia-designed silicon at its core rather than an Intel, AMD or Qualcomm chip alone.
| Manufacturer | Model(s) | Starting Price |
|---|---|---|
| ASUS | ProArt P14, ProArt P16 | $2,599+ |
| Dell | XPS 16 | $2,599+ |
| HP | OmniBook Ultra, OmniBook X 14 | $2,599+ |
| Lenovo | Yoga Pro 9n | $2,599+ |
| Microsoft | Surface Laptop Ultra | $2,599+ |
| MSI | Prestige N16 Flip AI+, Prestige N16 | $2,599+ |
Every model on that list runs on the same underlying superchip, which is the part of this story that reads less like a product refresh and more like a platform launch. Nvidia is not supplying one GPU SKU for OEMs to slot into their own designs. It is supplying an entire CPU-plus-GPU reference architecture and branding the resulting class of PCs after itself.
Inside the RTX Spark Superchip
The RTX Spark superchip combines an Nvidia Blackwell RTX GPU with up to 6,144 CUDA cores and an Nvidia Grace CPU with up to 20 cores. That is a meaningfully different architecture from the x86 chips that have powered nearly every Windows laptop for two decades, and it is also distinct from the Arm-based Qualcomm Snapdragon chips that power today’s Copilot+ PCs. Nvidia designed both halves of this chip itself, then wired them together with the kind of unified-memory approach it previously reserved for data-center hardware.
That unified-memory pool is the headline spec: up to 128GB of LPDDR5X shared between CPU and GPU, which Nvidia says delivers up to 1 petaflop of FP4 AI performance. The GPU side includes fourth-generation Ray Tracing Cores and fifth-generation Tensor Cores, drawing between 45 and 80 watts depending on configuration. For context, most premium Windows laptops top out well below 128GB of total system memory, let alone memory that both the CPU and GPU can address directly without copying data back and forth.
Nvidia founder and CEO Jensen Huang framed the pitch in stark terms: “RTX Spark brings the power of NVIDIA AI, RTX graphics and CUDA together with Microsoft Windows in one PC for creating, gaming and running personal AI agents,” he said, according to Nvidia’s official announcement. In a separate statement tied to the launch, Huang put it even more bluntly: “The PC is being reinvented.”
That is an aggressive claim for a 14-millimeter-thin laptop, and it is worth treating as marketing language rather than settled fact. What is confirmed is the hardware: a CPU and GPU Nvidia designed in-house, a memory pool far larger than typical Windows laptops carry, and a Windows 11 operating system running on top of it with full driver and software support from day one.
Why Microsoft Built a Flagship Around It
Of the six launch partners, Microsoft’s involvement carries the most symbolic weight. The company that owns Windows itself chose to build its own Surface Laptop Ultra around a third party’s complete CPU-GPU platform rather than defaulting to Intel, AMD or even its own Qualcomm-based Copilot+ design. Pavan Davuluri, Corporate Vice President of Windows + Devices at Microsoft, confirmed as much: “We built Surface Laptop Ultra around NVIDIA RTX Spark,” he said, per Nvidia’s blog covering the joint announcement. Microsoft details the resulting machine on its own Surface Laptop Ultra and RTX Spark product page.
Davuluri also previewed what Microsoft expects the memory capacity to unlock: “With up to 128 gigs of unified memory and up to a petaflop of AI compute, you can run models on this laptop that simply don’t fit on a traditional machine,” he said, according to the same source. That statement is really the thesis of the entire RTX Spark category: large local AI models, running without a cloud connection, on hardware thin enough to carry in a backpack.
Microsoft’s decision to co-brand a flagship Surface device around Nvidia silicon is a reversal of the usual OEM dynamic, where Microsoft has typically pushed its own Arm-based Surface Pro and Surface Laptop lines as showcases for Qualcomm’s Snapdragon X chips and the Copilot+ PC program it launched in 2024. RTX Spark does not replace that program, but it does give Microsoft a second, GPU-heavy flagship tier that Qualcomm’s chips were never designed to serve.
A Platform Launch, Not Just a New Chip
Nvidia’s framing matters here because of how it differs from past Nvidia laptop news. For roughly two decades, “Nvidia inside a laptop” has meant a discrete GeForce GPU slotted next to an Intel or AMD CPU that the OEM selected independently. RTX Spark changes that relationship. Nvidia now designs the CPU, the GPU, the memory architecture and the software stack as one package, then hands finished reference designs to OEMs to brand and sell.
That is a meaningfully bigger role in the Windows PC supply chain than Nvidia has ever held before. Intel and AMD have long controlled the “brain” of a Windows laptop while Nvidia supplied the graphics add-on. Qualcomm, since 2024, has supplied a complete Arm-based CPU platform of its own for Copilot+ PCs, but without a competitive discrete GPU to pair with it. RTX Spark is Nvidia’s answer to both: a complete platform that also happens to include the strongest consumer GPU architecture on the market.
Whether that amounts to Nvidia “entering the laptop market” in the way headlines are already framing it is a fair question. Nvidia is not selling its own laptop under its own brand the way Apple sells MacBooks. It is still routing every RTX Spark machine through OEM partners who handle manufacturing, retail, support and branding. But the chip-maker-to-platform-owner shift is real, and it is the most direct challenge Nvidia has mounted yet to the roles Intel, AMD and Qualcomm have played inside Windows PCs for a generation.
Nvidia’s History of Trying This Before
This is not Nvidia’s first attempt to own more of the computing stack than a single chip. The company’s nForce motherboard chipsets, introduced in the early 2000s, bundled graphics, networking and memory control for AMD and later Intel processors. The chipset business faded as CPU makers absorbed those functions directly into their processors, and Nvidia eventually exited that market entirely.
Tegra, Nvidia’s Arm-based system-on-chip line, found real homes in Microsoft’s Surface RT and the Nintendo Switch, but it never became a mainstream Windows PC platform. Project Denver, Nvidia’s early custom Arm CPU effort, mattered more as a stepping stone toward today’s Grace CPU than as a shipping consumer product. And the Shield line of Android gaming and streaming devices proved Nvidia could sell branded consumer hardware directly to buyers, but Shield stayed a niche line rather than a mass-market business.
Seen against that history, RTX Spark looks like the most complete version of an idea Nvidia has circled for roughly 20 years: pairing its own CPU with its own GPU and selling the combination as a platform rather than a part. The difference now is that Nvidia has a mature CUDA software ecosystem, a dominant position in AI compute, and a CPU design (Grace) that has already proven itself in data centers. Those pieces did not exist together during the nForce or early Tegra era.
Where Intel, AMD and Qualcomm Still Hold Ground
RTX Spark does not erase the three companies that have defined Windows PC chips for years. Intel’s Core Ultra processors remain the default choice across mainstream business and consumer laptops, backed by full x86 compatibility and the broadest OEM relationships in the industry. AMD’s Ryzen AI chips compete on similar turf, typically undercutting Intel on multi-core performance and integrated graphics in thinner, lighter, less expensive systems than anything in the RTX Spark lineup. Qualcomm’s Snapdragon X platform, meanwhile, owns the battery-life and always-connected end of the Copilot+ PC category, an area where Nvidia’s 45-to-80-watt GPU power draw is not designed to compete.
What RTX Spark changes is the top of the market. At $2,599 and up, Nvidia is not trying to win the mainstream laptop buyer away from Intel or AMD. It is building a new premium tier aimed at developers, creators and AI power users who need local model capacity that ordinary 16GB or 32GB laptops simply cannot provide. That is a narrower fight than “Nvidia versus the Windows PC industry,” but it is a fight on turf none of the three incumbents currently occupy with a comparable combined CPU-and-GPU product of their own.
| Platform | Chip Designer | CPU Architecture | Dedicated High-End GPU | Primary Strength |
|---|---|---|---|---|
| NVIDIA RTX Spark | Nvidia (Grace + Blackwell) | Arm (Grace) | Yes, up to 6,144 CUDA cores | Local AI model capacity, CUDA ecosystem, graphics |
| Intel Core Ultra | Intel | x86 | Integrated only | Broad OEM support, mainstream pricing, x86 compatibility |
| AMD Ryzen AI | AMD | x86 | Integrated Radeon graphics | Multi-core performance, value thin-and-light systems |
| Qualcomm Snapdragon X | Qualcomm | Arm | Integrated only | Battery life, always-connected Copilot+ features |
Readers tracking the broader chip rivalry can find more detail in our earlier coverage of how Intel and AMD stock reacted to Nvidia’s PC chip push, and how the three companies are positioned in the broader handheld and laptop chip war that has been building through 2026.
Market Impact: What RTX Spark Means for Nvidia’s Business Mix
Nvidia’s revenue has long been dominated by data-center GPU sales, with gaming and consumer products making up a smaller, if still substantial, share of the business. RTX Spark does not change that mix overnight. At $2,599 and up, these are premium devices aimed at developers, creative professionals and AI enthusiasts rather than mainstream buyers, which means unit volumes are unlikely to rival Intel’s or AMD’s mainstream laptop chip shipments any time soon.
The more important impact may be strategic rather than financial in the short term. By co-designing the complete platform rather than supplying a single component, Nvidia gets to showcase CUDA, TensorRT and its full AI software stack running on Windows hardware it had a direct hand in building end to end. That is valuable marketing for Nvidia’s broader AI ecosystem, independent of how many units of the Surface Laptop Ultra or ASUS ProArt P16 actually sell. It also puts Nvidia’s brand directly in front of Windows buyers for the first time in a market where Intel’s “Intel Inside” and Qualcomm’s Snapdragon branding have historically done that work.
There is also a channel dynamic worth watching. Every RTX Spark laptop currently ships through an existing OEM partner, which limits the risk of Nvidia competing directly against its own customers’ retail businesses. But if RTX Spark proves popular, the incentive for Nvidia to expand its share of the laptop’s bill of materials, and its share of the margin, only grows. That is the kind of shift that tends to make long-term OEM partners nervous, even when the near-term numbers look modest.
The Case for Why This Could Work
The strongest argument in RTX Spark’s favor is one Nvidia has already proven in the data center: pairing a custom Arm CPU with a leading-edge GPU and a large unified memory pool produces capabilities that chip combinations assembled from separate vendors struggle to match. Grace-Blackwell systems already power some of the fastest AI infrastructure in the world. Shrinking a version of that architecture into a 14-millimeter laptop is an ambitious bet, but it is grounded in hardware Nvidia has already shipped successfully at scale elsewhere.
CUDA is the other half of that case. Nvidia’s software ecosystem remains the default target for AI developers, and a laptop that runs the full CUDA and TensorRT stack locally, with up to 128GB of addressable memory, offers a development and testing environment that Snapdragon X’s Arm-based Windows machines and most Core Ultra or Ryzen AI laptops cannot replicate without a cloud GPU instance. For professional AI developers who need to prototype against large local models, that could be a genuinely differentiated pitch, independent of whether RTX Spark ever becomes a mainstream consumer product.
The Case for Caution
The skeptical case starts with price. At $2,599 for the entry configuration, RTX Spark sits well above where most Core Ultra, Ryzen AI and Snapdragon X laptops are priced, and that gap limits the realistic buyer pool to professionals and well-funded enthusiasts rather than everyday consumers. Power draw is a second consideration: the 45-to-80-watt GPU range Nvidia lists is substantial for a thin laptop chassis, raising the usual thermal and battery-life trade-offs that come with discrete-class graphics in a portable form factor.
Software compatibility is the least-discussed risk. Grace is an Arm-based CPU, which means RTX Spark laptops inherit the same category of Windows-on-Arm application compatibility questions that have shadowed Qualcomm’s Snapdragon X launch since 2024. Some Windows software runs natively, some runs through emulation, and some simply does not run at all. Nvidia’s announcement does not resolve how thoroughly that gap has closed, and buyers evaluating RTX Spark for general-purpose use, rather than AI development specifically, will want to confirm their critical software runs well on Arm before committing $2,599 or more.
How This Fits the Broader AI PC Push
RTX Spark arrives alongside a broader industry shift toward running AI models locally rather than exclusively in the cloud, a trend every major chipmaker has chased since Microsoft launched its Copilot+ PC certification in 2024. Intel, AMD and Qualcomm have each added neural processing units to their chips to qualify for that program. Nvidia’s answer skips the modest NPU approach entirely and instead brings a full discrete-class GPU and a massive unified memory pool to the same fight, which is a fundamentally different scale of local AI hardware than the Copilot+ baseline requires.
That scale gap is the clearest way to understand what Nvidia is actually selling. A typical Copilot+ PC can run small, efficient local AI features, like live captioning or on-device image generation, within its NPU’s power budget. RTX Spark is built for a different job: running substantially larger AI models locally, the kind that would otherwise require a cloud GPU subscription or a dedicated desktop workstation. Microsoft’s Davuluri made that gap explicit when describing the Surface Laptop Ultra’s 128GB memory ceiling, and it is the single clearest technical argument for why RTX Spark exists as a separate category rather than just another Copilot+ PC tier.
What Comes Next: Five Predictions
- Expect more OEM models, not fewer. Nvidia already has six manufacturing partners on board at launch; if early sales and reviews are reasonably strong, additional OEMs or additional models from the same six are likely within the next product cycle.
- Intel and AMD will respond with their own large-memory, AI-focused flagship tiers. Neither company currently ships a mainstream laptop platform with anything close to 128GB of unified CPU-GPU memory, and RTX Spark’s early reception will shape how urgently they move to close that specific gap.
- Price will become the central argument against adoption outside AI and creative professional circles. At $2,599 and up, RTX Spark is positioned well above the price sensitivity threshold for mainstream Windows laptop buyers, and that gap is unlikely to close quickly.
- Windows-on-Arm software compatibility will remain the most-asked question in early reviews. Because Grace is Arm-based, RTX Spark inherits the same compatibility scrutiny that has followed Qualcomm’s Snapdragon X since 2024, and how well Nvidia and Microsoft have addressed it will shape consumer sentiment more than raw specs will.
- Nvidia’s channel strategy will stay OEM-first for now. Nothing in the current announcement suggests Nvidia plans to sell a laptop under its own brand the way it sells Shield hardware; the more likely path is deepening its role inside OEM partners’ lineups rather than competing against them directly.
How RTX Spark Compares to Nvidia’s Own Recent AI PC Pushes
RTX Spark is not Nvidia’s only recent attempt to put its chips closer to end users. The company has also pushed DGX-branded desktop AI systems and software that links idle Windows PCs together for local AI acceleration. Readers interested in how Nvidia’s consumer GPU lineup, including the RTX 50-series cards that remain popular with gamers and creators, fits alongside this new laptop push can check our coverage of the latest Steam hardware survey showing RTX 5070 adoption trends. The RTX Spark launch also lands at a moment when Nvidia’s market capitalization has been trading places with Apple near the top of the S&P 500, a rivalry covered in our report on Apple crossing $5 trillion as it chases Nvidia’s valuation crown.
Nvidia is also not the only company building custom AI silicon to compete with its own ecosystem. Microsoft’s internal chip ambitions are part of the same broader story, detailed in our coverage of Microsoft’s Maia 300 chip and its claimed performance edge over Nvidia in data-center workloads. That Microsoft is simultaneously building competing silicon and co-designing a flagship RTX Spark laptop says something about how layered the Nvidia-Microsoft relationship has become: partners in Windows AI PCs, and increasingly rivals in data-center chips.
For readers who want the original specification breakdown of the RTX Spark platform before this market-entry angle, our earlier launch coverage of the RTX Spark laptop specs, pricing and 128GB memory ceiling walks through the hardware in more technical detail.
The Bigger Question: Is This Really a New Market Entry?
It is worth being precise about what Nvidia did and did not announce. Nvidia’s own materials describe RTX Spark as a new class of Windows PCs, not explicitly as Nvidia entering the laptop market as a standalone brand. Every RTX Spark laptop ships under an OEM’s name, through an OEM’s retail channels, with OEM-provided support. That is meaningfully different from, say, Apple selling MacBooks directly under its own name with its own retail stores and support infrastructure.
What has genuinely changed is how much of the laptop Nvidia now controls before it reaches an OEM. In the GeForce-GPU-as-a-component era, Nvidia supplied one part of a system someone else designed. In the RTX Spark era, Nvidia supplies the CPU, the GPU, the memory architecture and a chunk of the software stack as a near-complete reference platform, leaving OEMs to handle industrial design, manufacturing, branding and sales. That is a substantially larger slice of the laptop’s value than Nvidia has ever claimed before, even if the word “entering” overstates how directly Nvidia is competing with its own OEM partners today.
Frequently Asked Questions
When did NVIDIA RTX Spark laptops become available?
Nvidia confirmed availability starting October 16, 2026, with models from ASUS, Dell, HP, Lenovo, Microsoft and MSI shipping or available to order the same day.
How much do RTX Spark laptops cost?
Pricing starts at $2,599, with final pricing and configurations varying by region and manufacturer.
What chip powers RTX Spark laptops?
The RTX Spark superchip pairs an Nvidia Blackwell RTX GPU with up to 6,144 CUDA cores and an Nvidia Grace CPU with up to 20 cores, connected through a shared memory architecture.
How much memory do RTX Spark laptops support?
Up to 128GB of unified LPDDR5X memory, shared between the CPU and GPU, which Nvidia says enables up to 1 petaflop of FP4 AI performance.
Which companies are making RTX Spark laptops?
ASUS, Dell, HP, Lenovo, Microsoft and MSI are the six confirmed launch manufacturers, spanning nine total models including the ASUS ProArt P14 and P16, Dell XPS 16, HP OmniBook Ultra and OmniBook X 14, Lenovo Yoga Pro 9n, Microsoft Surface Laptop Ultra, and MSI Prestige N16 and N16 Flip AI+.
Does this mean Nvidia is becoming a laptop brand like Apple?
Not exactly. Every RTX Spark laptop ships under its OEM partner’s brand, through that OEM’s own retail and support channels. Nvidia designed the core chip platform, but it has not announced plans to sell a laptop under its own name.
How does RTX Spark compare to Qualcomm’s Snapdragon X Copilot+ PCs?
Both use Arm-based CPUs, but RTX Spark adds a discrete-class Nvidia GPU that Snapdragon X’s integrated graphics cannot match, at the cost of higher power draw and a significantly higher starting price.
Has Nvidia tried to sell complete PC platforms before?
Yes, though with mixed results. Nvidia’s nForce motherboard chipsets, Tegra mobile chips, Project Denver custom CPU research and Shield consumer devices all represented earlier attempts to move beyond supplying a single GPU component, none of which became a dominant mainstream Windows PC platform on their own.
