NVIDIA and MediaTek have locked in the launch plan for RTX Spark, the AI-first chip platform set to reshape the Windows laptop market this fall. Six OEMs, ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, are readying systems for a fall 2026 debut, with Acer and GIGABYTE confirmed to follow. Before a single unit has reached a retail shelf, ASUS and MSI both say their first production batches are already spoken for by channel partners, and a Morgan Stanley analyst note now puts early pricing at roughly $2,899 for the high-end N1X configuration.
That combination, a new Arm-based chip class, a surprise MediaTek collaboration, and demand that appears to be outrunning supply weeks before launch, makes RTX Spark one of the more consequential hardware stories of the year. It also raises a harder question: is this a genuine shift in how Windows PCs get built, or a tightly managed launch designed to look sold out on paper. Below is what’s confirmed, what’s estimated, and what it means for the PC market heading into Q4 2026.
What NVIDIA and MediaTek Actually Announced
NVIDIA’s own announcement, posted on its corporate newsroom, states that RTX Spark-powered slim Windows laptops and compact desktop PCs will ship “this fall” from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and GIGABYTE to follow (NVIDIA Newsroom). NVIDIA frames RTX Spark systems as small, efficient machines built to run personal AI agents around the clock at a desk, while still handling gaming and content creation through full RTX graphics.
MediaTek’s involvement is the twist most coverage has focused on. The Taiwanese chipmaker, long known for mobile silicon rather than premium PC hardware, co-developed the Arm-based CPU side of RTX Spark alongside NVIDIA, a partnership Wikipedia’s entry on the platform corroborates using NVIDIA’s own announcement details (Wikipedia). It’s MediaTek’s formal entry into NVIDIA’s AI PC ecosystem, and it puts the company in direct contact with the same OEMs, ASUS, MSI, Dell, HP, and Lenovo, that build the bulk of the world’s Windows laptops. For MediaTek, a company that has spent years supplying chips for budget phones and set-top boxes, landing a premium AI-PC design win next to NVIDIA is a meaningfully different market position.
Inside the RTX Spark Superchip: Grace CPU Meets Blackwell GPU
RTX Spark, which NVIDIA CEO Jensen Huang has also referred to by its internal name N1X, pairs a 20-core Grace CPU with a Blackwell RTX GPU carrying up to 6,144 CUDA cores and fifth-generation Tensor Cores with FP4 precision. NVIDIA rates the top configuration at up to 1 petaflop of FP4 AI performance. The two dies connect over NVLink-C2C at 600 GB/s, and the platform supports up to 128GB of unified memory shared between CPU and GPU, a design NVIDIA and reviewers say lets the chip run large local AI models, reportedly up to 120-billion-parameter models, without touching the cloud.
A cut-down variant, RTX Spark N1, trims that down substantially: 10 or 12-core Grace CPU configurations, up to 2,048 CUDA cores, and up to 64GB of unified memory, aimed at thinner and cheaper laptops that still carry the RTX Spark name. Power draw scales with the tier too, from roughly 18W on the lightest N1 builds up to 80W on full N1X configurations, which NVIDIA and OEM partners say still supports all-day battery life in the laptop form factor.
RTX Spark N1 vs N1X: The Spec Gap
The two-tier split matters for buyers trying to figure out which “RTX Spark laptop” they’re actually looking at, since OEMs are not required to disclose which chip variant ships in a given SKU until closer to launch. Here’s how the two configurations compare based on specs NVIDIA and partner OEMs have disclosed so far.
| Spec | RTX Spark N1 | RTX Spark N1X |
|---|---|---|
| CPU cores (Grace, Arm) | 10 or 12-core | 20-core |
| CUDA cores (Blackwell) | Up to 2,048 | Up to 6,144 |
| Unified memory | Up to 64GB | Up to 128GB |
| Peak AI performance | Not officially rated | Up to 1 petaflop FP4 |
| Power envelope | 18W – 45W | 45W – 80W |
| Estimated street price | ~$1,799 and up | ~$2,899 |
Those price estimates come from a Morgan Stanley analyst note cited by TechTimes and corroborated by PC Guide’s own reporting on channel checks with PC brands, not from official NVIDIA or OEM pricing, which partners including Microsoft, HP, ASUS, and MSI have so far deferred to “closer to launch.”
The Six Confirmed Launch Partners, and What They’re Building
NVIDIA’s OEM list for fall 2026 covers six companies at launch, with two more joining afterward. Named models have started to surface across OEM marketing pages and outlet reporting, though not every brand has finalized a public SKU name yet.
| OEM | Reported model | Launch status |
|---|---|---|
| ASUS | ProArt P16 / ProArt P14 | Fall 2026, first batch pre-ordered by channel partners |
| MSI | Prestige N16 Flip AI+ | Fall 2026, initial N1X batch reported near sell-out |
| Dell | XPS 16 (RTX Spark variant) | Fall 2026 |
| HP | OmniBook Ultra 16, OmniBook X 14 | Fall 2026, HP claims the thinnest RTX Spark laptop |
| Lenovo | Yoga Pro 9n | Fall 2026 |
| Microsoft | Surface Laptop Ultra | “Later 2026,” no confirmed date |
| Acer | Not yet named | To follow initial wave |
| GIGABYTE | Not yet named | To follow initial wave |
HP’s own newsroom post says RTX Spark will land in the OmniBook Ultra 16 and OmniBook X 14, and the company is positioning one of the two as the thinnest RTX Spark laptop on the market at launch. PC Guide’s reporting adds Dell’s XPS 16 and Lenovo’s Yoga Pro 9n to the confirmed list, filling out the picture across all six first-wave partners (PC Guide).
ASUS and MSI: Sold Out Before the Shelves Are Stocked
ASUS Says Channel Demand Outran Its Own Plan
On its Q2 2026 earnings call, ASUS told investors: “The market response has far exceeded our expectations. Essentially, the initial quantities we planned have already been fully pre-ordered by our channel partners,” a comment reported by WCCFTech and echoed by Yahoo Tech’s reporting on retail-partner demand for ASUS’s ProArt line (Yahoo Tech). That’s a distributor-level claim, not a consumer sellout. It means resellers and retail partners have committed to buying ASUS’s first production run, not that individual shoppers have exhausted retail stock, a distinction that matters for reading the story accurately, as we noted in our earlier coverage of ASUS and MSI’s initial RTX Spark order surge.
MSI Is Fielding Orders From Three Continents
MSI product manager Peng Renfang described similar pressure on the company’s own allocation: “Customers from mainland China, Taiwan, and the United States are actively placing orders, and the first batch of laptops equipped with the high-end N1X chip is also nearing a sell-out, with additional orders being placed.” NVIDIA’s own spec sheet for the platform, along with a broader OEM rundown, is tracked in detail by Kingy AI’s RTX Spark guide. Both companies say they’re going back to NVIDIA for additional chip supply ahead of the fall launch, which itself is a signal about how NVIDIA is metering initial RTX Spark output across partners.
Why a Pre-Launch Sellout Isn’t Necessarily What It Sounds Like
Reading “sold out” as consumer demand data is the most common mistake in coverage of this story so far. What ASUS and MSI have described is channel allocation, meaning the number of units NVIDIA agreed to supply for the first production run has already been claimed by distributors and retailers who plan to sell through them once RTX Spark laptops actually ship. It doesn’t tell us how many units NVIDIA planned to build, how large that first allocation actually was, or whether real-world consumer demand will match it once these laptops carry a confirmed $2,899-and-up price tag.
It’s also a familiar NVIDIA playbook. The company has run tight initial allocations on GPU launches for years, generating headlines about scarcity that build anticipation ahead of a wider retail ramp. That doesn’t make the channel demand fake, ASUS and MSI are public companies discussing real order data with investors and press, but it does mean the “sold out” framing should be read as a supply-planning story first and a demand story second.
Competitive Landscape: RTX Spark vs Apple Silicon and Snapdragon X
RTX Spark enters a Windows-on-Arm race that Qualcomm’s Snapdragon X series has been running largely alone for two years, while Apple’s own M-series chips have set the bar for battery life and on-device AI in the Mac lineup. NVIDIA’s pitch is different from both: rather than compete purely on CPU efficiency, RTX Spark leans on its GPU heritage, positioning the chip as a machine that can run a 120-billion-parameter model locally while also playing modern games above 100 fps at 1440p, a combination neither Snapdragon X nor Apple’s integrated GPUs currently target as their headline use case.
That framing puts RTX Spark closer to a hybrid workstation-laptop than a battery-life champion, which explains the $2,899 estimated price ceiling on the N1X tier, well above what Snapdragon X-based laptops typically command and in the same range as high-end configurations of Apple’s own recently launched M6 and M5 Ultra Mac Studio lineup, covered in our earlier report on Apple’s 80-core GPU launch. Whether creators and developers pay a premium for that combination, rather than splitting the job between a cheaper laptop and a cloud AI subscription, is the open question NVIDIA is betting on.
Historical Context: NVIDIA’s Long Push Into PC Silicon
NVIDIA has tried to get inside the Windows PC CPU market before. Its Tegra line powered early Windows RT tablets in the 2010s before Microsoft abandoned that platform, and NVIDIA has spent the years since focused almost entirely on discrete GPUs and, more recently, on data center AI accelerators. RTX Spark marks its most serious attempt yet at owning the full compute stack inside a mainstream Windows laptop, CPU and GPU both, rather than supplying just the graphics half of the system.
The timing lines up with NVIDIA’s broader push into AI infrastructure, detailed in our coverage of NVIDIA’s record $96.2 billion quarter, where data center AI chips now dominate revenue. RTX Spark extends that AI-first strategy down into consumer and prosumer hardware, effectively asking Windows laptop buyers to pay for the same class of on-device inference capability that’s driving NVIDIA’s data center business. It also isn’t the only recent example of demand outrunning supply around this launch window, a pattern we tracked separately in our report on OpenAI’s own hardware buying spree as RTX Spark pre-orders built up. For the full run of NVIDIA and AI-chip coverage, see our hardware section.
Market Impact: A Test Case for On-Device AI Pricing
If Morgan Stanley’s $2,899 estimate holds, RTX Spark’s N1X tier prices meaningfully above typical premium Windows laptops, a bet that buyers will pay extra for local AI horsepower rather than relying on cloud subscriptions from OpenAI, Anthropic, or Google. That’s a real gamble: cloud AI subscriptions cost a fraction of a laptop’s price premium per month, and most consumer AI workloads today still run comfortably in the cloud. RTX Spark’s case rests on privacy-sensitive or latency-sensitive workloads, plus a genuine appetite among developers and creators for local inference, an audience that’s real but historically smaller than the mainstream laptop market NVIDIA needs to hit its volume targets.
The launch also lands in the middle of an industry-wide memory crunch. DRAM and NAND prices have climbed sharply through 2026 as AI data center demand competes with consumer electronics for the same fabs, a dynamic we’ve tracked in our reporting on RAM prices rising amid the AI memory shortage. RTX Spark’s 128GB unified memory ceiling on the N1X tier makes it especially exposed to that cost pressure, since unified memory designs can’t be cheaply downgraded the way a discrete RAM stick swap can.
Risks: Software Support, Pricing Confirmation, and Real Availability
Three open risks stand out heading into the fall launch. First, Arm-based Windows machines have a mixed software compatibility track record, and RTX Spark will need broad application support, not just emulation, to justify its price versus a traditional x86 laptop. Second, none of the six OEMs have confirmed final consumer pricing, so the $2,899 and $1,799 figures remain analyst estimates rather than locked numbers, and actual street pricing could land higher given current memory costs. Third, “fall 2026” and “later 2026” are still the only public timing windows for several partners, including Microsoft’s Surface Laptop Ultra, which means some RTX Spark laptops announced today may not actually reach buyers before year-end.
What Happens Next: Five Predictions
- NVIDIA and its OEM partners confirm official pricing within four to six weeks of the fall launch window opening, likely landing close to Morgan Stanley’s $2,899 N1X estimate given current memory costs.
- ASUS and MSI both announce a second production allocation before the end of Q4 2026, extending the “sold out” narrative into a second wave rather than resolving it with broad retail stock.
- Acer and GIGABYTE confirm their first named RTX Spark models by early Q1 2027, rounding the OEM list out to eight.
- Qualcomm responds with an updated Snapdragon X pricing or performance push aimed squarely at the N1 tier, where RTX Spark is most exposed on price.
- Independent reviewers flag real-world AI performance well below the 1 petaflop FP4 marketing figure once third-party benchmarks arrive, a pattern common to first-generation AI PC chips.
How RTX Spark Fits Into NVIDIA’s Wider Product Line
RTX Spark sits alongside, not instead of, NVIDIA’s DGX Spark line, which targets developers and researchers running local AI workloads on dedicated desktop hardware rather than laptops. NVIDIA is effectively running two tracks from the same underlying Grace-Blackwell architecture: DGX Spark for a compact AI workstation audience, and RTX Spark for OEM laptops and compact desktops aimed at a broader creator and prosumer market. That shared architecture is also why RTX Spark isn’t sold as a standalone processor the way a CPU or GPU normally would be, it only ships inside finished OEM systems from the confirmed partner list.
What Buyers Should Actually Watch For
For anyone considering an RTX Spark laptop at launch, the practical advice is to wait for confirmed pricing and independent reviews rather than pre-ordering off spec sheets alone. Check which chip tier, N1 or N1X, a given model actually ships with, since OEMs haven’t been consistent about labeling that distinction in early marketing. And weigh whether the target workload genuinely needs on-device AI at this scale, versus a cheaper laptop paired with a cloud AI subscription, before paying the premium these systems are expected to carry.
Frequently Asked Questions
What is NVIDIA RTX Spark?
RTX Spark is a new chip platform from NVIDIA, co-developed with MediaTek, that pairs a Grace Arm CPU with a Blackwell RTX GPU and up to 128GB of unified memory. It’s designed for Windows laptops and compact desktops that run AI workloads locally alongside gaming and content creation.
When will RTX Spark laptops be available?
NVIDIA says RTX Spark laptops and compact desktops will be available this fall, 2026, from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with Acer and GIGABYTE following afterward. Exact per-model dates have not been confirmed.
How much will an RTX Spark laptop cost?
Official pricing hasn’t been announced. A Morgan Stanley analyst note estimates roughly $2,899 for the high-end N1X configuration and around $1,799 and up for the lower N1 tier, though OEMs have deferred confirmed pricing to closer to launch.
What is MediaTek’s role in RTX Spark?
MediaTek co-developed the Arm-based Grace CPU side of the RTX Spark platform with NVIDIA. It marks MediaTek’s entry into NVIDIA’s AI PC ecosystem after years of focusing primarily on mobile and embedded chips.
What’s the difference between RTX Spark N1 and N1X?
N1X is the higher-end tier, with a 20-core Grace CPU, up to 6,144 CUDA cores, and up to 128GB of unified memory. N1 is a cut-down variant with a 10 or 12-core CPU, up to 2,048 CUDA cores, and up to 64GB of memory, aimed at thinner, less expensive laptops.
Are RTX Spark laptops already sold out?
ASUS and MSI have both said their initial production allocations are fully pre-ordered by channel partners, meaning distributors and retailers, not individual consumers, have claimed the first batch. Both companies say they’re seeking additional supply from NVIDIA ahead of the fall launch.
Can RTX Spark run AI models without an internet connection?
Yes. NVIDIA and reviewers say the N1X tier’s 128GB unified memory and 1 petaflop FP4 performance are sized to run large local AI models, reportedly up to around 120 billion parameters, entirely on-device rather than through a cloud service.
Which laptop should I compare RTX Spark against?
The closest comparisons are Qualcomm Snapdragon X-based Windows laptops on the Arm side, and Apple’s M-series MacBooks and Mac Studio line on the on-device AI side. RTX Spark’s estimated pricing sits above typical Snapdragon X laptops and closer to premium Apple Silicon configurations.
