NVIDIA’s RTX Spark pre-orders went live on October 7, 2026, and the first units reached buyers on October 16. Six PC makers, ASUS, Dell, HP, Lenovo, Microsoft and MSI, now sell laptops built around the RTX Spark superchip, which pairs an NVIDIA Grace CPU with a Blackwell RTX GPU. It is an Arm-based design the company is pitching as a bridge between gaming hardware and on-device AI compute. The rtx spark price starts at $2,599 for complete laptops, according to NVIDIA’s launch announcement, with configurations and prices varying by region.
What’s missing from nearly every listing so far is the one thing PC gamers usually look for first: a discrete GeForce GPU branding and a frame-rate chart. Instead, NVIDIA and its partners are selling RTX Spark almost entirely on TOPS and TFLOPS, the units used to measure AI inference and training throughput. That framing, paired with the fact that RTX Spark runs Windows 11 on Arm rather than the x86 architecture nearly every PC game is built for, has opened a real question going into retail launch day: can these systems actually game, or is “gaming and AI” more marketing shorthand than an apples-to-apples upgrade over a traditional Windows laptop.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
RTX Spark Pre-Orders Went Live at Six OEMs
NVIDIA’s October 16 announcement named six shipping partners: ASUS, Dell, HP, Lenovo, Microsoft and MSI. The company said additional systems would follow. Not every manufacturer has published the same configurations or pricing, and prices vary by region. Dell’s XPS 16 became available on October 16, and Microsoft has separately said its developer-focused Surface RTX Spark Dev Box is scheduled to ship to U.S. customers in November 2026.
Pre-orders opened October 7, and NVIDIA confirmed on October 16 that RTX Spark laptops were globally available through OEM websites and major retailers. That was a tight window, and delivery for some orders could still vary with retailer inventory, a caveat several outlets flagged given how tight memory supply has been across the PC industry this year.
The Full RTX Spark Price Breakdown
The rtx spark price spread across confirmed listings is wide, stretching from a $2,599 entry laptop to a $5,999 developer desktop, with some configurations reportedly approaching $7,000 once memory and storage are maxed out. The table below collects the specific models and prices that have surfaced in retailer listings and reporting so far.
| OEM / Model | Price | Memory / Storage | Form Factor | Availability |
|---|---|---|---|---|
| Microsoft Surface Laptop Ultra EP2 | $2,599 | 24GB RAM / 512GB SSD | Laptop | October 16 |
| ASUS ProArt P14 | $2,599.99 (U.S.) | 24GB / 512GB | 14-inch 3K OLED laptop | October 16 |
| ASUS ProArt P16 | $2,799 (U.S.) | 32GB configuration | 16-inch creator laptop | October 16 |
| HP OmniBook X 14 | $2,999.99 | 32GB LPDDR5X-9400 / 1TB PCIe 5.0 SSD | 14-inch laptop | October 16 |
| Best Buy RTX Spark listing | $3,799.99 | 32GB / 512GB SSD | Laptop | October 16 |
| HP OmniBook Ultra 16 | $3,199.99 | 32GB LPDDR5X-9400 / 1TB PCIe 5.0 SSD | 16-inch laptop | October 16 |
| RTX Spark Dev Box | $5,999 | Not fully specified | Small desktop / dev workstation | Scheduled for November 2026 |
PCWorld has reported that some configurations could climb toward $7,000 depending on the OEM, memory capacity and overall build. That ceiling puts RTX Spark machines well above typical premium ultrabooks and closer to professional mobile workstation territory, a pricing decision that lines up with how NVIDIA and its partners are actually marketing these machines: as creator and developer tools first, gaming laptops a distant second if at all.
Inside the N1X Platform: Grace CPU Meets Blackwell GPU
RTX Spark is built around what NVIDIA calls the N1X platform, pairing an Arm-based Grace CPU with a Blackwell-generation GPU on a single superchip package. That’s a different architecture than the x86-plus-discrete-GeForce formula that has defined gaming laptops for two decades, and it comes in two configurations that NVIDIA and its OEM partners have detailed publicly.
| Spec | Entry Configuration | Higher Configuration |
|---|---|---|
| CPU | 18-core Grace | 20-core Grace |
| GPU | 5,120-core Blackwell | 6,144-core Blackwell |
| Unified memory | 24GB-32GB LPDDR5X | 24GB-128GB LPDDR5X |
| Peak FP32 performance | Up to 31 TFLOPS | Up to 31 TFLOPS |
| Peak AI performance | Up to 1,000 TOPS NVFP4 | Up to 1 petaflop (1,000 TOPS) NVFP4 |
The headline number NVIDIA keeps repeating is the roughly 1,000 TOPS of NVFP4 AI throughput on the higher-end configuration, framed by the company as roughly 1 petaflop of AI compute in a system that fits on a desk or inside a 16-inch laptop chassis. NVIDIA’s own product messaging, found on its official site, leans almost entirely on that figure rather than on traditional gaming metrics like ray-tracing performance or native frame rates in popular titles.
Memory is the other defining trait of the platform, and it cuts both ways. Because RAM and VRAM are combined into a single unified, soldered pool shared by the CPU and GPU, a machine advertised with 32GB or 128GB doesn’t split that capacity the way a conventional gaming laptop would between system memory and graphics memory. That design favors large AI models that need big, fast memory pools, which is exactly the workload NVIDIA is targeting. It also means buyers can’t upgrade memory after purchase, a trade-off that matters more on a $5,999 or $7,000 machine than it would on a budget laptop.
Six OEMs, Mostly Creator Branding
Look at the product names attached to RTX Spark so far and a pattern emerges. ASUS is selling it inside the ProArt line, its creator-focused laptop family, not the ROG gaming series. HP has placed it in OmniBook, a productivity and content-creation line rather than its Omen gaming brand. Microsoft’s entry is a Surface Laptop variant and a separate developer-oriented desktop box, both aimed squarely at professionals rather than gamers. Dell went the furthest toward a dual-purpose pitch with its XPS 16 RTX Spark Creator Edition, which still sits inside Dell’s creator-tier XPS branding rather than its Alienware gaming line.
That branding pattern is a meaningful data point on its own. If NVIDIA and its partners genuinely believed RTX Spark could compete as a gaming platform out of the gate, the obvious move would have been to launch it inside ROG, Alienware, Omen or a gaming-specific Surface line. Instead, every confirmed SKU so far sits inside a creator or productivity product family, which suggests the “gaming and AI” positioning in NVIDIA’s launch messaging is aspirational marketing layered on top of what is, in practice, a workstation-class AI chip.
The Windows on Arm Gaming Question
RTX Spark machines run Windows 11 on Arm, the same operating system variant that has powered Snapdragon-based laptops for the past two years. Windows on Arm has improved substantially since its early stumbles, but the vast majority of PC games are still compiled for x86 processors, not Arm. Those titles run through an emulation layer, and while Microsoft’s Prism emulator has closed much of the performance gap for productivity software, graphically demanding games with kernel-level anti-cheat systems, like many competitive shooters, have historically been the hardest category to get working reliably on Arm-based Windows machines.
That’s a separate problem from raw GPU horsepower. A Blackwell GPU with 6,144 cores is a capable piece of silicon, but if a game’s anti-cheat driver refuses to load under emulation, or if a title simply hasn’t been tested against Arm Windows, the hardware’s theoretical performance becomes irrelevant. NVIDIA has named launch titles, but none of the RTX Spark reporting so far includes independent native gaming benchmarks or frame-rate comparisons, which is notable given how central performance charts are to every other GPU launch NVIDIA has done in recent years.
None of this means RTX Spark machines can’t run games at all. Many will, especially titles without aggressive anti-cheat or ones with native Arm builds. But the absence of gaming benchmarks at launch, combined with the creator-focused branding across the OEM lineup, points toward a product category built primarily for local AI development, model fine-tuning and creative workloads, with gaming compatibility as a secondary, unverified bonus rather than a core selling point.
Which Games and Anti-Cheat Systems NVIDIA Has Named
NVIDIA’s gaming announcement of October 13, 2026, written by Andrew Burnes, laid out the launch lineup. Launch support spans single-player adventure and online multiplayer games including Cyberpunk 2077, Fortnite, Gears of War: E-Day, Marvel Rivals, World of Warcraft and NBA 2K27. AION 2 with DLSS 4.5 was slated for RTX Spark laptops on October 16th, and NVIDIA lists DLSS, Reflex, ray tracing, Remix and G-SYNC among the supported RTX technologies.
Anti-cheat is where the timeline stretches out. NVIDIA says each online game is protected by robust anti-cheat solutions, and EA Javelin Anticheat is coming to RTX Spark devices in Early 2027. In 2027, Call of Duty will be coming to Windows on RTX Spark, although the exact date and complete compatibility details have not been confirmed. KRAFTON intends to make PUBG: BATTLEGROUNDS available on the platform by the first half of 2027. Independent reporting dated October 15, 2026 said NVIDIA was working to support broad game compatibility and anti-cheat systems, and noted that those claims are not equivalent to confirmation that every Windows game works.
For buyers, the practical step is to check each title you care about against NVIDIA’s launch list before ordering, especially competitive shooters whose anti-cheat support is still listed for 2027. Price is worth confirming too: NVIDIA’s investor release gave a launch starting price of $2,599, while a corporate blog post dated October 17, 2026 says laptops started at $2,399, so verify the figure on the retailer page at checkout.
How RTX Spark Stacks Up Against Traditional Gaming Laptops
A $2,599 to $3,199 price band, where most of the confirmed RTX Spark laptops sit, overlaps directly with high-end x86 gaming laptops carrying discrete RTX 5070 Ti- or 5080-class GPUs, machines with years of driver optimization, verified anti-cheat compatibility and published benchmark suites across hundreds of titles. Those competing laptops also carry traditional, upgradeable memory and storage configurations in many cases, rather than RTX Spark’s soldered unified pool.
Where RTX Spark pulls ahead is in a workload traditional gaming laptops were never built for: running large local AI models without a cloud connection. A unified memory pool of up to 128GB lets a single machine hold language models that would otherwise require a multi-GPU workstation or a cloud GPU instance. For a developer prototyping agentic AI tools or a data scientist fine-tuning models locally, as covered in NVIDIA’s Surface Dev Box launch, that capability matters more than frame rates ever would. The honest comparison, then, isn’t RTX Spark versus a gaming laptop. It’s RTX Spark versus a small AI workstation, a category where NVIDIA’s own prior compact AI developer hardware and competing Arm AI boxes are the real rivals, not GeForce-branded gaming rigs.
Historical Context: From Developer Boxes to Retail Shelves
RTX Spark didn’t appear out of nowhere. NVIDIA has spent the better part of two years building out desktop-class AI hardware that sits below its data-center GPUs but above consumer graphics cards, a category that has included compact AI developer boxes sold directly and through partners. What’s new with RTX Spark is the move from a niche, developer-only distribution model into mainstream retail channels: Best Buy, direct OEM stores and the six major PC manufacturers NVIDIA named at launch simultaneously, a distribution footprint NVIDIA has never used for this class of hardware before. That’s a meaningful shift from selling a few thousand units to developers toward trying to move RTX Spark at the volume of a flagship laptop launch, a strategy shift also visible in how NVIDIA framed its RTX Spark Builder PC push toward mainstream desktop buyers.
It also arrives at a moment when the broader PC industry is leaning hard into AI-branded hardware. Market reporting tied to Gartner-style forecasts cited in recent coverage estimates that AI PCs, broadly defined, could account for roughly 143 million units and about 55% of total PC shipments globally in 2026, though that figure describes the AI PC category as a whole rather than RTX Spark specifically, and should be read as a forecast rather than a confirmed result. RTX Spark is NVIDIA’s attempt to plant a flag inside that broader AI PC wave with a product that’s differentiated enough to stand out.
What the OEM Lineup Says About NVIDIA’s Strategy
Running RTX Spark through six named shipping manufacturers at once, rather than launching an exclusive with one or two partners, mirrors how NVIDIA has handled reference-design GPU launches in the past: define a platform spec, let OEMs build around it, then let retail pricing sort out the differentiation. ASUS and HP shipped the earliest confirmed consumer pricing, Dell positioned its entry as a creator-tier XPS variant, and Microsoft split its lineup between a consumer Surface laptop and a separate developer desktop box, a dual-track approach that hedges between selling to professionals who want a laptop and developers who want a stationary workstation.
Acer and Gigabyte appeared in launch coverage but were not among the six shipping partners NVIDIA named on October 16, and the company said additional systems would follow, a gap also visible in earlier coverage of the initial RTX Spark pre-order rollout. That gap suggests those two manufacturers may be waiting to see how the first wave of pricing and early reviews land before committing to specific configurations and launch dates, a cautious approach given how unproven the retail demand for Arm-based AI laptops still is outside of the Snapdragon ecosystem.
Market Impact: Memory Prices and the AI PC Premium
RTX Spark’s pricing lands during a period of tight memory supply across the PC industry, and the platform’s dependence on large pools of soldered LPDDR5X makes it more exposed to component cost swings than laptops with conventional, separately sourced RAM and VRAM. A 128GB unified memory configuration consumes far more premium memory silicon than a typical 16GB or 32GB gaming laptop, which helps explain why top-tier RTX Spark pricing pushes toward the $6,000-$7,000 range PCWorld has reported, well above what buyers have historically paid for a single consumer laptop.
That dynamic puts RTX Spark at an interesting crossroads for retailers like Best Buy, which is listing a $3,799.99 configuration alongside conventional gaming laptops in the same price bracket. Shoppers browsing that shelf without context could reasonably assume they’re comparing like-for-like gaming hardware, when in practice they’re choosing between a discrete-GPU x86 gaming laptop with verified game compatibility and an Arm-based AI workstation with unverified gaming performance. How retailers label and explain that distinction at checkout over the next few weeks will shape early returns and reviews as much as the hardware itself does. The same tension already shows up in coverage of RTX Spark’s push into enterprise and agentic-PC deployments, where buyers are weighing AI capability against unproven security and compatibility footing.
What Developers and Creators Should Actually Expect
For the audience NVIDIA appears to actually be targeting, developers, data scientists and content creators, RTX Spark’s value proposition is more straightforward than the gaming question suggests. A unified memory pool scaling up to 128GB means local inference on larger open-weight language models without renting cloud GPU time, a capability that lines up with growing interest in self-hosting, including the kind of workflow described in this site’s guide to running Ollama locally. The 20-core Grace CPU and 6,144-core Blackwell GPU combination on the higher-end configuration gives creators enough throughput for video encoding, 3D rendering and AI-assisted content workflows that benefit from unified memory bandwidth rather than a split CPU/GPU memory architecture.
The trade-off is cost and flexibility. At $2,999.99 to $3,199.99, HP’s OmniBook X 14 and OmniBook Ultra 16 sit well above typical 32GB x86 creator laptops, and buyers can’t add memory later if a project outgrows the initial configuration. For professionals whose workloads map directly onto what RTX Spark is built for, that premium may be easy to justify. For anyone expecting a gaming laptop with AI features bolted on, the current lineup, including HP’s own earlier RTX Spark pricing leak, is likely to disappoint.
Five Predictions for RTX Spark’s First Six Months
- Independent reviewers will publish the first native gaming compatibility tests within weeks of the October 16 launch, and results will likely be mixed, with productivity and non-anti-cheat titles working reasonably well while competitive shooters with kernel-level anti-cheat remain unreliable.
- Acer and Gigabyte, which NVIDIA did not name among its six shipping partners on October 16, will reveal RTX Spark pricing and specific models within the following one to two months, likely undercutting the $2,599 entry price set by Microsoft and ASUS to compete on volume.
- NVIDIA will continue marketing RTX Spark around AI throughput metrics like TOPS and petaflops rather than gaming frame rates, reinforcing that the platform’s primary audience is developers and creators rather than gamers.
- Pricing at the high end of the lineup, approaching the roughly $7,000 ceiling PCWorld has reported, will remain sensitive to ongoing memory supply constraints, with the possibility of price increases on 128GB configurations if LPDDR5X costs continue climbing industry-wide.
- Retailers such as Best Buy will adjust product descriptions and marketing copy to more clearly separate RTX Spark’s AI-workstation positioning from conventional gaming laptops, following likely consumer confusion and return requests from buyers who expected GeForce-class gaming performance.
Open Questions After the October 16 Launch
Several details remain unresolved now that pre-orders have converted into shipped hardware. NVIDIA has named launch titles, but independent gaming benchmarks and driver-level details about how specific anti-cheat systems behave under Windows on Arm emulation have not been published. Acer and Gigabyte, which appeared in launch coverage but not in NVIDIA’s list of six shipping partners, are also essentially absent from the current retail picture. And while NVIDIA has published the two N1X configurations’ core specs, full technical documentation, including sustained (rather than peak) performance figures and thermal behavior in thin laptop chassis, has not yet been independently verified by third-party reviewers.
Those gaps matter because they’re exactly the kind of details that typically separate a successful hardware launch from a confusing one. NVIDIA’s GPU launches have historically leaned on detailed benchmark charts to justify pricing; RTX Spark’s launch, by contrast, has leaned almost entirely on raw AI compute figures and OEM diversity, a pattern more common to enterprise and workstation hardware rollouts than consumer gaming launches.
Frequently Asked Questions
What is NVIDIA RTX Spark?
RTX Spark is NVIDIA’s N1X platform, an Arm-based design combining a Grace CPU and a Blackwell-generation GPU on one superchip, built primarily for local AI workloads and creative applications rather than traditional PC gaming.
What is the RTX Spark price range?
Confirmed pricing runs from $2,599 for Microsoft’s Surface Laptop Ultra EP2 and ASUS’s ProArt P14 up to $5,999 for the RTX Spark Dev Box, with some higher-memory configurations reportedly approaching $7,000, according to PCWorld.
When do RTX Spark pre-orders ship?
Pre-orders opened October 7, 2026, and RTX Spark laptops became available worldwide on October 16, 2026 through manufacturer websites and major retailers, though delivery for some orders could vary with retailer inventory.
Which companies are selling RTX Spark PCs?
NVIDIA’s October 16 availability announcement named six shipping partners: ASUS, Dell, HP, Lenovo, Microsoft and MSI. Launch coverage had listed eight, adding Acer and Gigabyte, and NVIDIA said additional systems would follow. Microsoft, ASUS and HP have published specific model pricing so far.
Can RTX Spark laptops play PC games?
RTX Spark runs Windows 11 on Arm, which relies on emulation to run most x86-built PC games. NVIDIA said hundreds of games were available at launch, including Cyberpunk 2077, Fortnite and Marvel Rivals, but no manufacturer has published independent native gaming benchmarks, and the product branding across the OEM lineup targets creators and developers rather than gamers.
How much memory do RTX Spark PCs have?
Configurations range from 24GB to 128GB of unified LPDDR5X memory, shared between the CPU and GPU. This memory is soldered and cannot be upgraded after purchase.
What is the peak performance of RTX Spark?
The higher-end N1X configuration is rated for up to 31 TFLOPS of FP32 performance and up to roughly 1,000 TOPS, or about 1 petaflop, of NVFP4 AI performance, according to NVIDIA’s published specifications.
Is RTX Spark meant for gaming or AI work?
Based on current OEM branding, pricing, and NVIDIA’s own marketing focus on TOPS and petaflop figures rather than gaming benchmarks, RTX Spark is positioned primarily as an AI development and creator workstation platform, with gaming use treated as a secondary, largely unverified capability.
