NVIDIA RTX Spark Price: $2,599 PCs Launch [2026]

NVIDIA opened sales of its RTX Spark Windows PC lineup on October 16, 2026, putting a new class of machine built around local AI agents on store shelves for the first time. The starting price is $2,599, and six manufacturers, ASUS, Dell, HP, Lenovo, Microsoft and MSI, shipped laptops on day one. Compact desktops from the same roster follow in November, NVIDIA said, with Acer and GIGABYTE expected to join later.

The launch caps months of buildup that started at GTC Taipei during COMPUTEX in June, when NVIDIA first unveiled the RTX Spark platform alongside its developer-focused DGX Spark system. What is new this week is retail availability: real laptops, real prices, and a direct pitch to buyers who want to run AI models on their own hardware instead of renting cloud GPU time. For readers tracking the AI model NVIDIA bundles with the launch, our earlier coverage of the RTX Spark Nemotron 74B routing feature goes deeper on the software side of this release.

Google · Preferred Sources

Don't miss new tech stories on Google

Add TrendinTech once in the Google app and our stories appear in your news suggestions.

Add Now

NVIDIA RTX Spark Price and Launch Date: What Shipped Today

NVIDIA RTX Spark is both a chip and a PC platform name. The company uses it to describe a new superchip, and it also brands the laptops and desktops built around that chip as NVIDIA RTX Spark PCs. Laptop preorders opened in early October and general sales began October 16, 2026, matching the date NVIDIA set when Jensen Huang and Microsoft CEO Satya Nadella discussed the initiative on stage at a Microsoft event in San Francisco.

The nvidia rtx spark price starts at $2,599, though NVIDIA has said configurations and final pricing vary by region and retailer. That entry price buys a Windows laptop with NVIDIA’s full AI stack built in, not a stripped-down developer kit. The pitch is straightforward: a portable computer that can run substantial AI models locally, without a subscription or a round trip to a data center.

Six brands launched laptops simultaneously, which is an unusually broad rollout for a first-generation NVIDIA PC platform. Historically, new NVIDIA silicon trickles out through one or two lead partners before the rest of the market catches up. RTX Spark skipped that staggered pattern entirely.

Inside the Superchip: Blackwell GPU Meets Grace CPU

RTX Spark pairs an NVIDIA Blackwell RTX graphics architecture with an NVIDIA Grace CPU on a single package, the same general design philosophy NVIDIA has used in its data center superchips, shrunk down for a laptop chassis. NVIDIA describes RTX Spark as combining the capabilities of a gaming PC, a creative studio and a personal AI supercomputer in one chip, rather than treating AI as an add-on feature bolted onto a conventional laptop processor.

That framing matters for how the product is positioned against rivals. Qualcomm’s Snapdragon X2 and Intel’s Core Ultra Series 3 both chase efficient on-device AI through NPUs paired with conventional CPU cores, an approach our Snapdragon X2 versus Core Ultra Series 3 comparison breaks down in detail. NVIDIA’s bet is different: it wants the GPU, not a separate NPU block, doing the heavy AI lifting, with CUDA as the software layer that ties everything together.

128GB of Unified Memory: Why It Changes What Fits On a Laptop

The headline spec, after price, is memory. RTX Spark PCs top out at up to 128GB of unified memory shared between the CPU and GPU. On a conventional gaming laptop, the GPU’s own memory is walled off from system RAM, which caps how large a model can load before it runs out of room. Unified memory removes that wall, so a larger share of the total pool is available to whatever AI model is running.

NVIDIA founder and CEO Jensen Huang framed the practical upshot plainly: “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 during NVIDIA’s local-AI event with Microsoft. That is a different value proposition than raw frame rates or render times. It is about whether a model loads at all, which is the kind of binary constraint that has pushed many AI developers toward cloud GPUs in the first place.

It also explains why NVIDIA keeps pointing to its own Nemotron family of models as a reference workload. Our breakdown of how RTX Spark runs a 74 billion parameter Nemotron model locally shows the kind of task this memory pool is built for: a model too large for most laptop GPUs, running entirely on-device instead of calling out to a cloud API for every request.

1 Petaflop of FP4 Compute: Putting the Number in Context

NVIDIA rates RTX Spark at up to 1 petaflop of FP4 AI performance. FP4 is a low-precision number format increasingly used to run large language models faster and with a smaller memory footprint, at some cost to numerical precision compared with the FP16 or FP32 formats older GPUs relied on for AI work. A petaflop figure quoted in FP4 is not directly comparable to a petaflop figure quoted in a higher-precision format, so buyers comparing spec sheets across brands should check which precision a number refers to before assuming two chips are equally fast.

What the figure does confirm is that NVIDIA built RTX Spark around inference workloads first. Petaflop-class AI compute in a laptop chassis was a data center benchmark a few years ago. Readers who have followed our coverage of RTX 5090 versus RTX 4090 AI benchmarks will recognize the pattern: NVIDIA keeps pushing AI throughput numbers that used to belong to server racks down into consumer and prosumer hardware, one generation at a time.

Six PC Makers Ship on Day One

ASUS, Dell, HP, Lenovo, Microsoft and MSI all launched RTX Spark laptops on October 16, 2026. NVIDIA has said Acer and GIGABYTE will add models later, without giving a specific date. That is a wide bench of partners for a brand-new chip family, and it mirrors the breadth NVIDIA achieved years ago when it first pushed GeForce RTX graphics into thin-and-light laptops rather than confining the silicon to bulky gaming rigs.

Microsoft’s involvement goes beyond being one of six OEMs. The company co-developed the Windows-side tooling that lets RTX Spark machines manage local AI agents, and Satya Nadella appeared alongside Jensen Huang to introduce the initiative at a dedicated event in San Francisco. Microsoft’s own blog post on the launch described the collaboration as producing what the company called the most powerful and efficient thin-and-light Windows PCs it has shipped.

Compact Desktops Follow in November

Laptops are first out the door, but NVIDIA has also committed to compact desktop versions of RTX Spark, which it says will go on sale in November 2026 from the same group of manufacturers. Desktop form factors typically allow for better sustained cooling than laptops, which could matter for workloads that keep the chip running at full tilt for extended periods, such as fine-tuning a model or batch-processing a large dataset, rather than short bursts of agent activity.

The staggered rollout, laptops now, desktops next month, suggests NVIDIA and its partners prioritized the mobile form factor for the initial marketing push, possibly because a laptop running petaflop-class AI locally is a more striking demonstration of the platform’s premise than a desktop box doing the same thing on a desk.

RTX Spark vs DGX Spark vs Traditional AI Laptops

NVIDIA now has two products with Spark in the name, and it is worth being precise about the difference. DGX Spark is the developer-focused system NVIDIA introduced earlier in 2026 as a dedicated local AI workstation, sold directly through NVIDIA’s own channels to developers and researchers. RTX Spark takes a related architectural idea, a GPU and CPU sharing one large unified memory pool, and packages it inside mainstream Windows laptops and desktops sold by familiar PC brands rather than through NVIDIA’s developer storefront.

Against conventional gaming laptops with a mobile RTX GPU and a separate CPU, RTX Spark’s pitch is the unified memory pool rather than raw gaming frame rates. A gaming laptop built around a discrete mobile GPU can still outperform RTX Spark in traditional rendering benchmarks, since RTX Spark’s design trades some of that specialization for a larger shared memory pool and an architecture tuned for AI inference. Apple’s Mac Studio line has offered a broadly similar unified-memory concept on its own Apple Silicon chips for several years, but it runs macOS, not Windows, and does not plug into NVIDIA’s CUDA ecosystem, which remains the dominant software layer for AI development.

Platform	Form factor	Primary audience	Operating system	Unified memory approach
NVIDIA RTX Spark PCs	Laptop (now), compact desktop (November)	Consumers, creators, developers	Windows	CPU and GPU share one memory pool, up to 128GB
NVIDIA DGX Spark	Developer desktop system	AI researchers and engineers	NVIDIA’s Linux-based stack	Dedicated unified-memory design, sold direct through NVIDIA
Traditional RTX gaming laptop	Laptop	Gamers, creators	Windows	Separate GPU memory and system RAM
Apple Mac Studio (Apple Silicon)	Desktop	Creative professionals	macOS	Unified memory on Apple’s own chip, no CUDA support

NVIDIA RTX Spark Confirmed Specifications

Here is what NVIDIA has confirmed about the platform as of launch day, stripped of marketing language, for readers comparing it against other machines on a spec sheet basis.

Spec	Confirmed detail
Chip design	Blackwell RTX GPU paired with Grace CPU in one superchip
Starting price	$2,599, varies by region and configuration
Maximum unified memory	Up to 128GB shared between CPU and GPU
AI performance	Up to 1 petaflop of FP4 AI compute
Operating system	Windows
Laptop launch partners	ASUS, Dell, HP, Lenovo, Microsoft, MSI
Future partners	Acer and GIGABYTE, timing not yet announced
Laptop availability	October 16, 2026
Compact desktop availability	November 2026

Where RTX Spark Fits in NVIDIA’s AI PC Roadmap

NVIDIA has spent the past two years pushing AI capability down through its product stack, from data center GPUs, to workstation cards, to gaming GPUs with AI-assisted features such as DLSS, and now to a chip purpose-built for personal AI agents in a laptop. RTX Spark is the clearest sign yet that the company sees local, on-device AI as a mainstream consumer category rather than a niche for developers with dedicated workstations.

That roadmap shift shows up elsewhere in NVIDIA’s recent announcements too. The company’s decision to make more of its AI software stack freely available, detailed in our report on NVIDIA’s open access push for European developers, lowers the barrier for developers to build the kind of local-agent software RTX Spark is meant to run. A hardware platform with unused software support is a weaker pitch than one backed by a growing library of tools, and NVIDIA appears to be building both sides of that equation at once.

Market Impact: Intel, AMD and Qualcomm Face a New Challenger

RTX Spark puts NVIDIA in direct competition with Intel and Qualcomm on a part of the PC market it has historically not played in directly: the core processor inside a Windows laptop. NVIDIA has long supplied the discrete GPU sitting alongside an Intel, AMD or Qualcomm processor. With RTX Spark, NVIDIA’s own Grace CPU is doing the general-purpose computing work too, at least in these specific laptop models.

That is a meaningful shift in how the PC supply chain is organized. For decades, the processor and the graphics chip in a Windows laptop typically came from different vendors. RTX Spark collapses that into a single NVIDIA part. If the platform sells well through six major OEMs, it gives NVIDIA use over a slice of laptop bill-of-materials that used to belong entirely to CPU makers.

It also lands at an awkward moment for Intel and AMD’s own AI PC pushes, both of which have spent the past two years marketing NPU-equipped chips for on-device AI without matching the memory headroom RTX Spark offers. Our coverage of NVIDIA’s decision to pull the RTX 5090 and shift its GB202 silicon toward RTX Pro products shows a company increasingly willing to reshuffle its product lineup around where AI demand is strongest, rather than protecting legacy categories.

Historical Context: From CUDA to the Personal Agent Era

NVIDIA’s CUDA platform, which turned its GPUs into general-purpose computing engines rather than graphics-only chips, is now two decades old and underpins almost everything the company has built since, including RTX Spark’s AI software stack. The jump from CUDA as a tool for researchers to RTX Spark as a mainstream retail product took most of that time, with the AI boom of the past several years compressing the final steps.

RTX Spark itself moved unusually fast by NVIDIA’s own standards. The platform was unveiled at GTC Taipei during COMPUTEX in June 2026, alongside the DGX Spark developer system, and reached retail shelves across six manufacturers within about four months. That is a tighter unveiling-to-shipping window than NVIDIA has used for most of its major GPU architecture launches, which typically carry a longer gap between announcement and broad retail availability.

The Software Stack Developers Get on Day One

NVIDIA says the global PC app ecosystem is ready for RTX Spark at launch, pointing to updated tools spanning local AI agents, NVIDIA Broadcast, DLSS updates and developer frameworks for building and running agents locally rather than through a cloud API. For developers specifically, NVIDIA and Microsoft jointly published guidance on building personal AI agents on Windows PCs using the new hardware, treating RTX Spark as a platform to build on rather than a sealed consumer appliance.

That developer framing is consistent with how the device is being marketed to the software-engineer audience specifically, as a machine that keeps agents running continuously at no extra per-query cost, since the compute happens on hardware the user already owns rather than through a metered API. For a sense of how a competing on-device approach handles that same tradeoff, our look at Liquid AI’s on-device agent model covers a smaller, more efficiency-focused alternative built for phones and lighter hardware rather than a 128GB laptop.

What NVIDIA and Microsoft Are Saying

NVIDIA’s own description of the launch leans heavily on the idea of a shift in what a personal computer is for. In the company’s words, NVIDIA RTX Spark reinvents Windows PCs for the era of personal AI agents, offering a new class of computer that moves from tool to teammate, as detailed in NVIDIA’s original platform unveiling at COMPUTEX.

Jensen Huang’s comments at the Microsoft event focused on the practical ceiling the extra memory removes, telling attendees that the unified memory and compute headroom let a laptop run models that simply do not fit on a traditional machine. NVIDIA’s broader messaging around the launch repeats that RTX Spark puts the company’s full AI stack into Windows laptops and compact desktops, rather than treating AI as a bolt-on feature.

Microsoft’s framing, published on its own Windows Experience blog, described the collaboration in similarly sweeping terms, crediting the joint engineering work with producing the most powerful and efficient thin-and-light Windows PCs the two companies have shipped together. NVIDIA’s investor-facing announcement used nearly identical language, calling RTX Spark a new superchip that reinvents Windows PCs for the era of personal AI agents.

Launch Partner Lineup at a Glance

Manufacturer	Product type confirmed	Status as of October 16, 2026
ASUS	Laptop	Shipping
Dell	Laptop	Shipping
HP	Laptop	Shipping
Lenovo	Laptop	Shipping
Microsoft	Laptop (Surface)	Shipping
MSI	Laptop	Shipping
Acer	Not yet specified	Announced, no date given
GIGABYTE	Not yet specified	Announced, no date given

Who Actually Needs a $2,599 AI Laptop

At $2,599 and up, RTX Spark is priced well above mainstream laptops and squarely in prosumer territory, closer to a high-end creator or gaming laptop than an everyday productivity machine. That price point makes sense for software engineers building and testing local agents, researchers who want to iterate on models without cloud GPU bills, and creative professionals running generative tools that benefit from a large memory pool. It is a harder sell for someone who mainly wants a fast Windows laptop for browsing, office work or casual gaming, since those tasks do not touch the hardware NVIDIA built the price premium around.

That split matters for how OEMs market their individual RTX Spark configurations. Expect ASUS, Dell, HP, Lenovo and MSI to lean on developer and creator messaging rather than general consumer advertising, at least for this first wave, while Microsoft’s Surface-branded model likely leans on its own ecosystem of Windows AI features to justify the price to a broader buyer.

Predictions: Where RTX Spark Goes From Here

Expect Acer and GIGABYTE to announce specific RTX Spark models within the next one to two quarters, rounding out the eight-brand roster NVIDIA has signaled.
Watch for Intel and AMD to respond with updated AI PC messaging that emphasizes memory capacity rather than just NPU throughput, since RTX Spark’s unified-memory pitch targets the exact limitation their current chips share.
Compact desktop RTX Spark units arriving in November will likely draw comparisons to NVIDIA’s own DGX Spark on price and target audience, and NVIDIA will need to keep the two product lines clearly differentiated in its marketing.
Software partners building on NVIDIA’s local-agent tooling are likely to expand quickly, given NVIDIA and Microsoft’s joint developer documentation published alongside the launch.
Pricing above the $2,599 entry point will probably stay volatile through the end of 2026 as retailers and OEMs adjust configurations to regional demand, consistent with NVIDIA’s own statement that pricing varies by region.

Frequently Asked Questions

What is the NVIDIA RTX Spark price at launch?
RTX Spark laptops start at $2,599, with NVIDIA noting that exact configurations and pricing vary by region and retailer.

When did NVIDIA RTX Spark PCs go on sale?
General availability began October 16, 2026, following laptop preorders that opened earlier in October.

Which brands sell NVIDIA RTX Spark laptops?
ASUS, Dell, HP, Lenovo, Microsoft and MSI shipped RTX Spark laptops on launch day. NVIDIA has said Acer and GIGABYTE will join later.

How much memory does RTX Spark support?
Up to 128GB of unified memory shared between the Grace CPU and Blackwell RTX GPU, depending on configuration.

What is the difference between RTX Spark and DGX Spark?
DGX Spark is NVIDIA’s developer-focused desktop system sold directly by NVIDIA. RTX Spark applies a related unified-memory design to mainstream Windows laptops and desktops sold by established PC brands.

When do RTX Spark desktops go on sale?
NVIDIA has said compact desktop versions of RTX Spark will be available in November 2026 from the same manufacturer lineup as the laptops.

What operating system does RTX Spark run?
RTX Spark PCs run Windows, distinguishing them from NVIDIA’s DGX Spark developer system, which uses NVIDIA’s own Linux-based software stack.

Is RTX Spark meant for gaming or AI work?
NVIDIA markets RTX Spark as combining gaming, creative work and local AI in one machine, though its unified-memory design and FP4 compute rating are built primarily around AI inference rather than maximizing traditional gaming frame rates.

NVIDIA RTX Spark PCs Hit $2,599, 6 Makers Ship Day 1 [2026]