Nvidia RTX Spark Superchip Named CES 2027 Innovation Award Honoree

The Consumer Technology Association named Nvidia RTX Spark, a Grace CPU plus Blackwell GPU superchip for thin laptops, a CES 2027 Innovation Awards honoree.

Nvidia RTX Spark superchip laptop platform promotional image
Nvidia RTX Spark superchip platform. Image: Nvidia.

Nvidia's RTX Spark Superchip and Laptops platform has been named a CES 2027 Innovation Awards honoree, the Consumer Technology Association (CTA) confirmed as part of its November 9, 2026 announcement of the 2027 honoree class. The recognition, listed on CTA's ces.tech site in the Gaming & eSports category, covers the Grace-CPU-plus-Blackwell-GPU "superchip" platform Nvidia is building into slim Windows laptops and compact desktops — the same architecture family that, in data-center form, underpins Nvidia's GB-series and DGX systems.

The award itself is largely a confirmation of a platform Nvidia already unveiled earlier in 2026: RTX Spark pairs a Grace CPU (up to 20 cores) with a Blackwell RTX GPU (up to 6,144 CUDA cores) over an NVLink-C2C interconnect, with up to 128GB of memory shared between the two processors, fifth-generation Tensor Cores with FP4 support for up to one petaflop of AI compute, and support for local AI models as large as 120 billion parameters with a 1-million-token context window, according to the CES Innovation Awards listing. Nvidia's own newsroom confirmed in October that the first RTX Spark laptops are already shipping from OEM partners, though with some figures that differ from the ones on the CES award page — a gap this article lays out below.

What actually happened

CTA, the organization behind CES, revealed its 2027 Innovation Awards honorees on November 9, 2026. The program drew nearly 4,000 submissions across 37 categories — a record and roughly 10% more than the prior year — with CTA president and CEO Kinsey Fabrizio citing "incredible momentum across our industry" in the announcement. Nvidia RTX Spark Superchip and Laptops appears on CTA's ces.tech site as a Gaming & eSports honoree, with its own dedicated award page describing the platform in detail.

It's worth being precise about what this award is and isn't. CTA's November 9 press release itself calls out two other Nvidia honors by name — the Nvidia Vera CPU and Nvidia DLSS 5, both tagged "Best of Innovation" in their categories — but does not mention RTX Spark in the release text. RTX Spark's honoree status comes from its own listing in CTA's online award directory rather than a Best of Innovation callout in the press release, a distinction that matters for anyone tracking which Nvidia products received the program's top tier versus a standard honoree mention. Additional honorees and full presentations are scheduled to be revealed at CES 2027, which runs January 6-9, 2027, in Las Vegas, with a showcase at the Venetian.

What the RTX Spark superchip actually is

RTX Spark is Nvidia's name for a superchip platform aimed at thin-and-light Windows laptops and small-footprint desktops, rather than at data centers or developer workstations. It combines a Grace CPU with a Blackwell-generation RTX GPU on a single module, connected by Nvidia's NVLink-C2C chip-to-chip interconnect — the same link technology Nvidia uses in its data-center Grace Hopper and Grace Blackwell superchips. Unlike a traditional laptop, where the CPU and discrete GPU each have their own dedicated memory pool, RTX Spark's CPU and GPU share a single unified memory pool of up to 128GB, letting large AI models, texture sets, and application data move between compute engines without being copied back and forth.

Per the CES Innovation Awards listing, the platform's ceiling specs are a 20-core Grace CPU, a Blackwell RTX GPU with up to 6,144 CUDA cores, and fifth-generation Tensor Cores supporting FP4 precision for up to one petaflop of AI compute. Nvidia pitches this as enough headroom to run local AI models with as many as 120 billion parameters and a context window up to one million tokens entirely on-device, without sending prompts or data to the cloud.

RTX Spark specs at a glance

SpecCES Innovation Awards listingNvidia's October 2026 launch details
CPUNvidia Grace, up to 20 coresNvidia Grace (core count not restated)
GPUBlackwell RTX, up to 6,144 CUDA coresBlackwell RTX (core count not restated)
InterconnectNVLink-C2CNVLink-C2C
Unified memoryUp to 128GB, shared CPU/GPUUp to 128GB, shared CPU/GPU
AI computeUp to 1 petaflop (FP4, 5th-gen Tensor Cores)Up to 1 petaflop (FP4)
Local model supportUp to 120B parameters, 1M-token contextNemotron 3.5 Bolt 74B-A9B (24GB) on supported configs
Creative workload12K video editing; 3D scenes over 90GB8K 4:2:2 video editing, incl. unplugged; scenes up to 90GB
GamingAAA titles at 1440p, above 100fpsAAA titles at 1440p, above 100fps; DLSS 5 adds up to 1.4x
Laptop thicknessAs thin as 14mm (planned)Not restated; described as "slim" designs
AvailabilityPlanned for fall 2026Available October 16, 2026, from launch OEMs
Starting priceNot listedFrom $2,599 (varies by region/configuration)

Why bringing a "superchip" to a laptop matters

The word "superchip" is doing real work here. Nvidia has used it previously for data-center and workstation parts — the Grace Hopper and Grace Blackwell superchips that power AI supercomputers, and the GB10-based Nvidia DGX Spark, a developer-focused mini desktop that shipped in late 2025 and can run models up to roughly 200 billion parameters on a single unit. RTX Spark is a different, consumer-and-creator-facing product built around the same basic idea — tightly coupling a CPU and GPU with a shared memory pool over NVLink-C2C — but packaged for thin Windows laptops rather than a desktop AI appliance. Readers should not conflate the two: DGX Spark is a developer mini PC built to run and fine-tune large models at a desk, while RTX Spark Superchip and Laptops is the platform aimed at the laptops and compact desktops people actually carry and buy at retail.

That distinction is the real story behind the CES recognition. Unified-memory superchip architecture has mostly lived in racks and developer boxes; RTX Spark is Nvidia's attempt to push that architecture into mainstream laptop form factors competing directly with Apple's unified-memory Apple Silicon Macs and Qualcomm- and Intel-based Copilot+ PCs. It's also part of a broader pattern: Nvidia's data-center business has scaled enormously (the company's most recent quarterly results showed data-center revenue alone topping $107 billion), and that data-center scale is now visibly funding and informing Nvidia's push into consumer silicon.

It's already shipping — with some numbers that don't match the CES listing

One nuance worth flagging clearly: by the time CTA named RTX Spark a CES 2027 honoree, Nvidia had already announced that RTX Spark laptops were available. Nvidia's own newsroom said on October 16, 2026, that RTX Spark PCs were "available today" globally through OEM sites and major retailers, with launch partners Asus, Dell, HP, Lenovo, Microsoft, and MSI shipping 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 Flip AI+ and Prestige N16, starting at $2,599.

That October announcement, however, cited somewhat different figures than the ones on CES's own award page. Nvidia's launch materials specified local support for a Nemotron 3.5 Bolt 74B-A9B model on supported configurations (not the 120-billion-parameter ceiling listed on the CES page), 8K 4:2:2 video editing rather than 12K, and did not restate a specific 14mm chassis thickness. It's unclear from public information whether that reflects a lower-tier launch configuration versus a higher-spec version still to come, rounding or aspirational framing in the award submission, or simply two different snapshots of the same evolving platform. We're flagging the discrepancy rather than resolving it, since neither Nvidia nor CTA has published a reconciliation.

The AI and creative pitch

Nvidia's case for RTX Spark centers on running substantial AI workloads locally rather than in the cloud. The shared memory pool means a chatbot, coding assistant, or image-generation model doesn't have to be split awkwardly between a small GPU memory allocation and a separate CPU memory pool — everything fits in one address space, up to 128GB. On the creative side, Nvidia is pitching RTX Spark as capable of editing 8K video (with CES materials citing a theoretical 12K ceiling) and handling 3D scenes larger than 90GB, workloads that have traditionally needed a tower workstation rather than a laptop that can supposedly be as thin as 14mm.

On gaming, Nvidia and CES both cite AAA titles running at 1440p above 100 frames per second, with Nvidia's October materials adding that DLSS 5's newer, faster AI model can boost frame rates by up to 1.4x on top of that baseline. None of these performance figures have been independently verified by outside reviewers at the time of this article; they are Nvidia's own claims.

What a CES Innovation Awards honoree designation means

The CES Innovation Awards are run by CTA and judged by a panel on innovation, functionality, and design, spanning 37 categories this year, including a new Art Tech category run in partnership with Christie's. CTA reported that AI, digital health, and robotics made up a combined 25% of this year's nearly 4,000 entries, with robotics and enterprise tech submissions each up 74% year over year. An "honoree" designation is CTA's standard recognition tier; a smaller number of products within each category additionally receive the "Best of Innovation" distinction, which RTX Spark did not receive (Nvidia's Vera CPU and DLSS 5 did, in separate categories). Selected honorees, including RTX Spark, are expected to get fuller public presentation when CES 2027 opens in January, though CTA's materials don't specify an exact embargo lift date beyond the show itself.

How RTX Spark fits against rival laptop silicon

RTX Spark enters a laptop-chip market already crowded with unified-memory and AI-focused designs. Apple's Mac lineup has used a unified CPU/GPU memory architecture for several chip generations now, and buyers weighing Apple's M6 against the previous M5 are asking many of the same questions Windows buyers will now ask about RTX Spark: how much shared memory is enough, and how that affects running large local AI models. On the x86 and Arm-for-Windows side, Intel has been pushing its own efficiency-and-AI pitch with chips compared in pieces like Panther Lake versus Lunar Lake, while Nvidia's RTX Spark instead leans on a discrete-class Blackwell GPU paired with an Arm-based Grace CPU rather than an integrated graphics block. That makes RTX Spark less a direct architectural match to Intel's or AMD's integrated designs and more a laptop-scaled version of Nvidia's own data-center superchip philosophy.

Quick facts

  • What: Nvidia RTX Spark Superchip and Laptops — a Grace CPU + Blackwell RTX GPU platform for thin laptops and compact desktops.
  • Award: Named a CES 2027 Innovation Awards honoree (Gaming & eSports), announced by CTA on November 9, 2026.
  • Peak specs cited by CES: 20-core Grace CPU, 6,144-CUDA-core Blackwell GPU, 128GB shared memory, 1 petaflop FP4 AI compute, 120B-parameter local models, 1M-token context.
  • Real-world status: First RTX Spark laptops already shipped October 16, 2026, from Asus, Dell, HP, Lenovo, Microsoft, and MSI, starting at $2,599.
  • Still unconfirmed: The specific 14mm-thin chassis, the full 120B-parameter/1M-token configuration, and exact pricing for every RTX Spark SKU.

What's still unknown, and what to watch at CES 2027

Several details remain open. Nvidia's CES award submission describes laptops "as thin as 14mm" as planned for fall 2026, but no specific model at that thickness has been confirmed in Nvidia's own October launch materials or by an OEM. It's also unclear whether the 120-billion-parameter, 1-million-token-context configuration cited by CES represents a higher-end RTX Spark tier still arriving, or simply a more optimistic framing of the same hardware already shipping with a smaller 74-billion-parameter reference model. Exact pricing beyond the $2,599 starting point, the full roster of OEM designs beyond the models already announced, and region-by-region availability also remain to be confirmed.

CTA says additional honorees and fuller presentations will be revealed at CES 2027, running January 6-9, 2027, in Las Vegas, with select winners shown at the CES Innovation Awards Showcase at the Venetian. Readers interested in RTX Spark's full story should watch for Nvidia's own CES keynote and booth presence in January, where the gap between the award page's maximum specifications and what's actually on store shelves is most likely to get clarified.

Frequently asked questions

What is Nvidia RTX Spark?

RTX Spark is Nvidia's superchip platform for thin Windows laptops and compact desktops. It pairs a Grace CPU with a Blackwell RTX GPU over NVLink-C2C, sharing up to 128GB of memory between the two processors for AI, content creation, and gaming.

Is RTX Spark the same as Nvidia DGX Spark?

No. DGX Spark is a developer-focused mini desktop built on Nvidia's GB10 superchip that shipped in late 2025. RTX Spark Superchip and Laptops is a separate, consumer-and-creator-facing platform built for thin laptops and small desktops sold through PC makers.

Why was RTX Spark named a CES 2027 Innovation Awards honoree?

The Consumer Technology Association listed Nvidia RTX Spark Superchip and Laptops as a 2027 CES Innovation Awards honoree in the Gaming and eSports category, as part of its November 9, 2026 announcement of nearly 4,000 honorees across 37 categories.

Are RTX Spark laptops available yet?

Yes. Nvidia's newsroom said RTX Spark laptops became available on October 16, 2026, from Asus, Dell, HP, Lenovo, Microsoft, and MSI, starting at $2,599, though some CES-listed ceiling specs, like 12K video editing and a 120-billion-parameter model limit, were not fully restated in that launch announcement.

How thin will RTX Spark laptops get?

CES's award listing says laptops as thin as 14mm are planned for fall 2026, but no specific 14mm model has been confirmed yet in Nvidia's own launch materials or by an OEM.

When will full RTX Spark details be revealed at CES?

CTA says additional honoree presentations will be shown when CES 2027 opens January 6-9, 2027, in Las Vegas, including a CES Innovation Awards Showcase at the Venetian.

Sources

More on Nvidia RTX Spark →NvidiaRTX SparkCES 2027AI laptopsGrace CPUBlackwell GPU
Mara Lindqvist
Written byMara Lindqvist

Mara Lindqvist edits the hardware desk. She covers graphics cards, processors, memory and storage, the foundries and chip designers behind them, and what the numbers on a spec sheet mean for people choosing a PC. Specifications in her stories come from manufacturer spec pages and datasheets.

More from Hardware

See all