ASUS and MSI have run out of RTX Spark computers before most shoppers even had a chance to place an order. Both companies confirmed that their first production allocations from Nvidia are already spoken for through channel pre-orders, and both are now going back to Nvidia for a second batch of the GB10-powered chips that make the machines possible. It is an unusual position for a brand-new consumer product category to be in: sold out before it has properly launched.
The news comes from two directions. Wccftech reported on August 13, 2026, that ASUS told investors its initial RTX Spark quantities had been fully claimed by channel partners, and separate Taiwan-based financial reporting today, August 31, 2026, indicates MSI is in the same position, with each company’s first shipment covering roughly 100,000 units. For a product line that Nvidia only unveiled at Computex in June, that is a fast sellout, and it says as much about how tight AI-related chip supply still is in late 2026 as it does about consumer appetite for on-device AI hardware.
What Actually Sold Out: RTX Spark, Explained
RTX Spark is Nvidia’s push to bring its GB10 Grace Blackwell superchip, the same silicon that powers the DGX Spark desktop “personal AI supercomputer”, into laptops and compact desktops built by PC brands rather than sold directly by Nvidia. The GB10 pairs a 20-core Arm-based Grace CPU (ten Cortex-X925 performance cores and ten Cortex-A725 efficiency cores) built on a 3-nanometer TSMC process with a Blackwell GB20B GPU carrying 48 streaming multiprocessors and 6,144 CUDA cores. Nvidia and its partners rate the chip at up to 1 petaflop of AI performance, or roughly 1,000 AI TOPS in FP4 sparse mode, backed by 128GB of LPDDR5X-8533 unified memory shared between CPU and GPU.
What makes that memory pool significant is not raw gaming performance, since RTX Spark is not a gaming chip in the RTX 50-series sense. It is the ability to load and run large language models locally, without a cloud subscription or an API bill. According to ASUS’s own pre-launch messaging, the RTX Spark platform is pitched at developers who want to run models with up to 120 billion parameters directly on a laptop, while the larger DGX-class desktop version, the ASUS Ascent GX10, is positioned for models up to roughly 200 billion parameters. That is a meaningfully different value proposition than a typical Copilot+ PC, and it is why the first wave of buyers looks less like consumers and more like AI labs, research groups, and independent developers who have been priced out of cloud GPU rental.
ASUS Confirms the Sellout in Its Own Earnings Call
The clearest on-record confirmation comes from ASUS itself. During its most recent investor call, reported by Wccftech and echoed across regional financial outlets, ASUS management described RTX Spark demand in blunt terms: “The market response has far exceeded our expectations. Essentially, the initial quantities we planned have already been fully pre-ordered by our channel partners, so we’re continuing to place additional orders and are very optimistic about next year’s sales.”
Two things are worth unpacking in that statement. First, “channel partners” means distributors and retailers, not end consumers, so this is a wholesale-level sellout rather than proof that individual buyers have already snapped up every unit. Second, ASUS did not disclose the exact number of units in that first allocation on the call itself, which is why the 100,000-unit figure now circulating comes from separate supply-chain reporting rather than from ASUS’s own investor materials. Even accounting for that caveat, a company going back to a supplier for a second production run before its first laptops have technically shipped is a strong demand signal, not routine pre-launch marketing.
ASUS’s first RTX Spark systems are expected to arrive under its ProArt line, including the ProArt P16 and ProArt P14 laptops plus a ProArt Mini PC, positioning RTX Spark within ASUS’s existing creator and professional hardware family rather than as a standalone gaming or consumer sub-brand.
MSI’s Matching Allocation and the 100,000-Unit Figure
MSI has not issued a public statement as detailed as ASUS’s earnings-call remarks, but Taiwan-based financial reporting citing UDN Money, published August 31, 2026, states that MSI’s first RTX Spark batch has also sold out through preorders at the distribution level, and that ASUS and MSI each received enough GB10 chips from Nvidia to build approximately 100,000 computers in their respective first shipments. That parity is notable: it suggests Nvidia allocated its first meaningful wave of RTX Spark silicon evenly between the two launch partners rather than favoring one over the other, and that both companies exhausted that allocation on comparable timelines.
MSI’s RTX Spark hardware is expected to include the Prestige N16 Flip AI+ laptop and an EdgeMesa N AI+ mini PC, both shown around Computex 2026 alongside early ASUS and HP entries. MSI already has experience in this exact hardware category through the EdgeXpert MS-C931, a DGX Spark-class desktop unit that launched in October 2025 and has been sold primarily through direct pre-order and enterprise channels rather than open retail, a distribution pattern that appears to be repeating with RTX Spark.
Why Nvidia Capped the First Batch This Tight
The 100,000-unit figure per partner is small by PC industry standards, where a single popular laptop SKU can ship millions of units a year. That scarcity traces back to the same memory constraints that have been squeezing Nvidia’s entire AI hardware lineup through 2026, a supply crunch Digitimes has tracked across Taiwan’s semiconductor and memory supply chain for much of the year. In February, Nvidia raised the price of the DGX Spark Founders Edition from $3,999 to $4,699, a $700 increase that Wccftech attributed directly to ongoing memory shortages affecting the LPDDR5X supply that GB10-based systems depend on. That same shortage has rippled into gaming GPUs too: Crypto Briefing reported on August 3, 2026, that GeForce RTX 50-series cards were becoming up to 30% more expensive in South Korea because of memory costs and constrained supply.
Nvidia CEO Jensen Huang addressed supply more broadly at Computex in June, where he unveiled RTX Spark while emphasizing that Nvidia had secured enough capacity to support continued GPU and CPU growth and pledged roughly $150 billion a year in Taiwan-based investment. Even with that commitment, the memory bottleneck specifically, rather than logic-chip fabrication, appears to be the binding constraint on how many RTX Spark units ASUS and MSI could build for this first run. Crypto Briefing’s August 11 coverage of Nvidia’s efforts to calm AI market jitters likewise pointed to global memory shortages as the factor forcing specification and pricing recalibrations across Nvidia’s AI product stack, not just RTX Spark.
RTX Spark vs. DGX Spark: Two Products, One Chip
It helps to separate two things that share the same GB10 silicon but serve different markets. DGX Spark is Nvidia’s own branded desktop reference platform and the partner-built equivalents certified under that program, aimed squarely at AI labs and enterprise developers. RTX Spark is the mobile and compact-desktop expansion of that same chip, sold as a feature inside mainstream laptop and mini PC lines from ASUS, MSI, and other PC brands, aimed at a slightly broader developer and creator audience who want portability alongside local AI inference.
As of August 10, 2026, the DGX Spark partner ecosystem included eight officially certified systems: Nvidia’s own DGX Spark Founders Edition, the Acer Veriton GN100, ASUS Ascent GX10, Dell Pro Max with GB10, Gigabyte AI TOP ATOM, HP ZGX Nano AI Station, Lenovo ThinkStation PGX, and MSI’s EdgeXpert MS-C931. RTX Spark laptops sit adjacent to that lineup rather than inside it, using the same GB10 foundation but packaged for mobility and, in ASUS and MSI’s case, priced to reach a wider audience than the roughly $4,000 DGX-class desktops.
| System | Maker | Category | Memory | Status (Aug 31, 2026) |
|---|---|---|---|---|
| DGX Spark Founders Edition | Nvidia | Desktop | 128GB LPDDR5X | Shipping; price raised to $4,699 |
| Ascent GX10 | ASUS | Mini desktop | 128GB LPDDR5X | Shipping via DGX Spark program |
| EdgeXpert MS-C931 | MSI | Desktop | 128GB LPDDR5X | Pre-order/direct-contact sales since Oct. 2025 |
| ProArt P16 / P14 / Mini PC | ASUS | Laptop/mini PC | Unified GB10 memory | First batch sold out via channel pre-order |
| Prestige N16 Flip AI+ / EdgeMesa N AI+ | MSI | Laptop/mini PC | Unified GB10 memory | First batch reportedly sold out via pre-order |
| ZGX Nano AI Station | HP | Mini desktop | 128GB LPDDR5X | DGX Spark-certified partner system |
| ThinkStation PGX | Lenovo | Desktop | 128GB LPDDR5X | DGX Spark-certified partner system |
| AI TOP ATOM | Gigabyte | Desktop | 128GB LPDDR5X | DGX Spark-certified partner system |
The GB10 Grace Blackwell Superchip, in Numbers
Every RTX Spark and DGX Spark system leans on the same core component, so understanding the chip explains why demand has outrun supply. The specs are consistent across every certified partner build:
| Spec | Detail |
|---|---|
| CPU | 20-core Grace Arm CPU: 10x Cortex-X925 + 10x Cortex-A725 |
| Process node | TSMC 3nm |
| TDP | 140W |
| Cache | 49MiB (25MiB L2 + 24MiB system-level cache) |
| GPU | Blackwell GB20B: 48 SMs, 6,144 CUDA cores, 192 5th-gen Tensor cores, 48 4th-gen RT cores |
| GPU clock | Up to 2.42GHz |
| AI performance | Up to 1 petaflop / ~1,000 AI TOPS (FP4, sparse) |
| Memory | 128GB LPDDR5X-8533 unified (CPU + GPU shared) |
| Local model capacity | Up to ~120B parameters (RTX Spark) / ~200B parameters (DGX-class desktop) |
Pricing: What RTX Spark Laptops Are Expected to Cost
Neither ASUS nor MSI has published a locked retail price for their first RTX Spark laptops, and any figure attached to the launch should be read as an estimate rather than a confirmed MSRP. Early analyst estimates that have circulated ahead of the fall launch put entry-level RTX Spark laptop configurations in the neighborhood of $1,799, with higher-memory or higher-performance configurations closer to $2,899. That would place RTX Spark laptops well below the roughly $4,000 price point of DGX-class desktops like the Ascent GX10 and EdgeXpert, which makes sense given the different target buyer, but it still sits far above a typical mainstream laptop, reflecting the cost of packing a 128GB unified-memory AI chip into a portable chassis.
For comparison, Nvidia’s own DGX Spark Founders Edition, sold with 4TB of storage, now costs $4,699 after February’s $700 increase, while the ASUS Ascent GX10, sold with 1TB of storage, was originally guided at $2,999. Both of those price points, first reported by TechRadar Pro, predate the current RTX Spark laptop sellout and reflect the desktop-class end of the GB10 lineup rather than the mobile systems that just ran out of stock.
Timeline: From Computex Reveal to a Sold-Out First Batch
The speed of this sellout only makes sense against the timeline. Nvidia unveiled RTX Spark at Computex on June 2, 2026, alongside Jensen Huang’s remarks about secured chip supply and continued Taiwan investment. Wccftech reported on July 22 that ASUS and MSI would be the first vendors to ship RTX Spark PCs, with Acer and Gigabyte following later, and confirmed a fall 2026 launch window. By August 13, ASUS was telling investors its channel allocation was already spoken for. And by August 31, the same pattern had been reported for MSI, with both companies’ combined first-wave allocation estimated at roughly 200,000 units before either laptop has technically reached a retail shelf.
That compresses what is normally a multi-month ramp, from unveiling to sellout, into under three months, and it happened entirely at the pre-order and channel-allocation stage rather than through public retail listings. Consumer availability for RTX Spark laptops is still expected in the fourth quarter of 2026, meaning the systems that just sold out to distributors have not yet reached the shoppers who will ultimately buy them.
Historical Context: Nvidia’s Pattern of Undersupplying New Categories
This is not the first time Nvidia has launched a new hardware category into an allocation-constrained market. The original DGX Spark desktop line, which MSI’s EdgeXpert joined in October 2025, was sold largely through direct pre-order and required buyers to “contact MSI directly” rather than purchase off a retail shelf, according to Tom’s Hardware‘s coverage of the launch. The February price increase to $4,699, driven by memory costs rather than a change in the chip itself, was an early sign that Nvidia’s AI hardware supply chain was running tighter than its GPU business could comfortably absorb. RTX Spark’s sellout extends that same pattern from a niche desktop product into a mainstream laptop category for the first time, which raises the stakes considerably: a shortage that was previously confined to AI labs and enterprise buyers is now hitting consumer PC brands’ core laptop lineups.
Competitive Landscape: Acer, Gigabyte, and the Rest of the Field
ASUS and MSI’s head start is deliberate. Nvidia structured RTX Spark’s rollout so that ASUS and MSI would ship first, with Acer and Gigabyte following in a second wave, according to Wccftech’s July reporting. That staggering means Acer and Gigabyte are watching the exact scenario playing out now, two launch partners selling out a 100,000-unit allocation apiece before shipping, as they finalize their own order volumes with Nvidia. The DGX Spark side of the ecosystem, where Dell, HP, and Lenovo are already certified partners alongside Gigabyte, gives some indication of how the field will fill out on the mobile RTX Spark side too, though none of those three has been named as an early RTX Spark laptop partner.
The bigger competitive question, one The Verge and other outlets have raised repeatedly through 2026, is whether Apple, AMD, or Qualcomm respond with a comparable unified-memory AI laptop platform. Apple’s M-series chips already offer large unified memory pools that developers use for local LLM inference, and Nvidia’s own GB10 pricing sits well above what an equivalent-memory MacBook Pro configuration costs, meaning RTX Spark’s real competition for developer dollars may come from outside the traditional Windows-laptop field entirely.
Market Impact: What a 200,000-Unit Sellout Means
In isolation, roughly 200,000 combined units between ASUS and MSI is a rounding error against global laptop shipments, which run into the hundreds of millions annually. The significance is not volume, it is what the sellout signals about Nvidia’s ability to allocate scarce GB10 and memory supply toward its most reliable OEM partners. ASUS explicitly told investors it is “very optimistic about next year’s sales” and is placing additional orders now, which suggests the company expects 2027 volumes to scale well past this initial allocation once memory supply loosens. For Nvidia, a fast sellout of a brand-new, higher-margin AI PC category, arriving right as its DGX Spark desktop line was already selling at raised prices, reinforces the case that AI-capable PC hardware, not just data-center GPUs, is now a meaningful and constrained revenue line.
It also puts pressure on the broader PC supply chain. If GB10 chips and the LPDDR5X memory that feeds them stay scarce into 2027, expect other AI PC categories, including Copilot+ laptops from Microsoft’s partner ecosystem, to compete for the same constrained memory fabs, a dynamic that has already pushed RTX 50-series GPU prices higher in markets like South Korea this year.
What This Means for Developers Waiting to Buy One
For the AI developers, researchers, and small labs this hardware is actually built for, the practical takeaway is straightforward: the first units are already gone at the wholesale level, and the fourth-quarter 2026 retail launch is likely to be tight from day one. Anyone planning to buy an RTX Spark laptop to run 100-billion-plus-parameter models locally instead of renting cloud GPU time should expect waitlists similar to what DGX Spark buyers faced in late 2025, when MSI directed interested buyers to contact the company directly rather than shop retail. Given that ASUS is already placing follow-on orders, a second wave of supply is coming, but Nvidia and its partners have not given a firm date for when that second batch will reach the same channel partners who just cleared out the first one.
Predictions: Where RTX Spark Goes From Here
- Retail listings for ASUS ProArt and MSI Prestige/EdgeMesa RTX Spark systems will likely carry waitlists or limited regional availability when they formally launch in Q4 2026, mirroring the pre-order-only pattern DGX Spark set in 2025.
- A second RTX Spark price adjustment is plausible if LPDDR5X memory costs keep climbing, following the same $700 increase Nvidia applied to the DGX Spark Founders Edition in February.
- Acer and Gigabyte’s later-2026 RTX Spark entries will likely launch with larger allocations than ASUS and MSI’s initial 100,000-unit batches, benefiting from a few extra months of GB10 and memory production ramp.
- Demand will keep concentrating among AI developers and small labs running local models rather than mainstream consumers, keeping RTX Spark a premium, dev-focused category through 2026 rather than a mass-market laptop line.
- Expect Nvidia to lean harder on RTX Spark messaging at CES 2027 as evidence that AI PCs, not just data-center GPUs, are a durable growth line, especially if ASUS’s “optimistic about next year” framing translates into confirmed 2027 order volumes.
Frequently Asked Questions
What is Nvidia RTX Spark?
RTX Spark is Nvidia’s platform for bringing the GB10 Grace Blackwell superchip, the same silicon used in Nvidia’s DGX Spark desktop AI systems, into laptops and compact desktops built by partners like ASUS and MSI. It combines a 20-core Grace Arm CPU with a Blackwell GPU and 128GB of unified memory, aimed at running large AI models locally.
Did ASUS and MSI RTX Spark laptops actually sell out?
Yes, at the channel level. ASUS confirmed on an investor call, reported by Wccftech on August 13, 2026, that its initial RTX Spark allocation had been fully pre-ordered by distributors and retailers. Taiwan-based financial reporting citing UDN Money reported on August 31, 2026, that MSI’s first batch sold out the same way, with each company’s first shipment covering an estimated 100,000 units.
When will RTX Spark laptops be available to buy?
Nvidia and its partners have targeted a fall 2026 launch window, with ASUS pointing to a fourth-quarter 2026 release for its ProArt P16, ProArt P14, and ProArt Mini PC RTX Spark systems. MSI’s Prestige N16 Flip AI+ and EdgeMesa N AI+ are expected on a similar timeline.
How much will RTX Spark laptops cost?
Neither ASUS nor MSI has confirmed final retail pricing. Early estimates circulating ahead of launch put entry configurations around $1,799 and higher-tier configurations near $2,899, though these figures are not official MSRPs and should be treated as estimates until each company confirms pricing closer to launch.
What is the difference between RTX Spark and DGX Spark?
Both use the same GB10 Grace Blackwell superchip, but DGX Spark refers to Nvidia’s own desktop reference platform and certified partner desktops like the ASUS Ascent GX10 and MSI EdgeXpert, aimed at AI labs and enterprises and priced near $4,000. RTX Spark is the laptop and compact-desktop version built into mainstream ASUS and MSI product lines, expected to launch at lower price points.
Why is Nvidia GB10 chip supply so limited?
Reporting from Wccftech and Crypto Briefing throughout 2026 has pointed to ongoing memory shortages, specifically LPDDR5X supply, as the main constraint, rather than a shortage of Blackwell GPU logic dies. That same memory pressure pushed Nvidia to raise the DGX Spark Founders Edition price by $700 in February 2026 and contributed to RTX 50-series GPU price increases of up to 30% in markets like South Korea by August 2026.
How many RTX Spark units did ASUS and MSI receive in their first batch?
According to Taiwan-based financial reporting citing UDN Money, ASUS and MSI each received enough GB10 chips to build approximately 100,000 computers in their respective first shipments, putting the combined first wave at roughly 200,000 units across both companies.
Can RTX Spark laptops run large AI models locally?
Yes. With 128GB of unified LPDDR5X memory, RTX Spark laptops are positioned to run AI models with up to roughly 120 billion parameters locally, without relying on cloud GPU rental, while the larger DGX-class desktop systems built on the same chip are positioned for models up to roughly 200 billion parameters.