HP used its Computex 2026 keynote in Taipei to pull back the curtain on a pair of Windows laptops that signal a new chapter in AI-accelerated computing. The HP OmniBook Ultra 16 and OmniBook X 14 are the first devices confirmed to ship with NVIDIA’s RTX Spark platform, a long-rumored silicon collaboration that blends an ARM-based CPU with NVIDIA’s GPU and AI prowess. Rather than an immediate retail drop, HP positioned the machines as the vanguard of a “late-2026 AI upgrade,” hinting that both hardware and a future Windows update will unlock experiences far beyond today’s Copilot+ features.

The Debut: A Computex 2026 Exclusive

On a stage already crowded with AI PC announcements, HP’s unveiling stood out for its tangible details. The company shared early renders of the OmniBook Ultra 16 in a dark aluminum chassis and the OmniBook X 14 in a lighter, ultraportable form factor. Both laptops are firmly planted in the Copilot+ ecosystem, featuring a dedicated Neural Processing Unit (NPU) that easily exceeds Microsoft’s 40 TOPS baseline. But the star of the show was the “RTX Spark” branding etched beside the keyboard — a clear signal that NVIDIA has moved beyond discrete GPUs for laptop AI into the system-on-chip (SoC) arena.

NVIDIA’s RTX Spark platform represents a joint effort with MediaTek, according to industry sources. The SoC is expected to combine a high-performance ARM Cortex-X series CPU cluster with a custom NVIDIA graphics architecture derived from the Ada Lovelace family, alongside a fourth-generation Tensor Core NPU. This integrated design allows for shared memory between CPU, GPU, and AI engines, a critical factor for advancing on-device inference and real-time content creation. While NVIDIA officials declined to share exact core counts or clock speeds during the event, the platform’s “Spark” moniker alludes to the role it’s meant to play: igniting a new class of AI experiences on thin-and-light laptops.

HP’s decision to announce the OmniBook Ultra 16 and X 14 now, months ahead of actual availability, is a strategic bet. It gives the company first-mover advantage in the RTX Spark race and positions these devices as the definitive answer to Qualcomm’s Snapdragon X Elite, which has dominated the early Copilot+ market. The “late-2026 AI upgrade” framing suggests that HP, Microsoft, and NVIDIA are coordinating a major software moment later this year — possibly tied to the long-rumored Windows 11 “Hudson Valley” update or an early preview of Windows 12 features. This update is expected to unlock the full potential of RTX Spark’s hardware, enabling advanced capabilities like persistent on-device assistants, local LLM fine-tuning, and AI-powered creative tools that today’s PCs can only dream of.

What Exactly Is RTX Spark?

For the past two years, Windows on Arm has been synonymous with Qualcomm’s Snapdragon X. The RTX Spark platform shatters that monopoly by bringing NVIDIA’s graphics heritage and AI leadership directly into the Copilot+ equation. Here’s what we know so far:

  • Architecture: RTX Spark is a heterogeneous SoC that integrates CPU, GPU, and NPU on a single die using a unified memory architecture. The CPU side is ARM-based and reportedly designed by MediaTek, leveraging the latest Armv9.2 ISA with a mix of performance and efficiency cores. The GPU block is custom NVIDIA, featuring ray tracing cores, Tensor cores, and the media engine found in GeForce RTX 40 series GPUs, though scaled for laptop thermals.
  • AI muscle: While exact TOPS figures remain under wraps, the platform is said to deliver over 100 TOPS of combined AI performance, significantly outpacing the 45 TOPS of the Snapdragon X Elite’s Hexagon NPU. This headroom is crucial for running larger language models natively and for PC-specific AI tasks such as real-time video style transfer, 3D object generation, and advanced voice synthesis.
  • Windows on Arm native: Unlike previous NVIDIA Grace or Tegra attempts, RTX Spark is built from the ground up for Windows on Arm. It will run native ARM64 Windows and come with NVIDIA’s full DirectX, Vulkan, and CUDA support. Emulation for x86-64 apps is handled by Microsoft’s improved Prism technology, but the platform’s real strength lies in running ARM-native games and productivity apps that can tap into the GPU’s parallel processing power.
  • Efficiency: Early whispers from the HP event suggest the RTX Spark package has a configurable TDP ranging from 15W to 45W, making it suitable for both fanless ultraportables and performance-oriented laptops. This flexibility likely explains why HP chose two distinct form factors for the launch.

NVIDIA’s entry into the PC SoC space is not entirely out of character. The company’s Tegra line powered the Nintendo Switch, and its DRIVE platform dominates automotive. But RTX Spark is the first chip designed for the mass Windows market, and NVIDIA is betting that its AI ecosystem — CUDA, TensorRT, cuDNN — will attract developers who have been lukewarm on the Qualcomm-only Copilot+ vision.

The HP OmniBook Ultra 16 and X 14

HP’s OmniBook brand, resurrected in 2024 for its next-generation AI PCs, now expands with two distinct models targeting different audiences. Both will feature the RTX Spark platform, but they differ in screen size, thermal design, and intended use cases.

HP OmniBook Ultra 16

The Ultra 16 is the performance flagship, aimed at creators, developers, and gamers who need more screen real estate and sustained power. Its 16-inch display is expected to offer a 120Hz OLED panel with 100% DCI-P3 color coverage, making it a compelling tool for video editing and design work. The chassis, rendered in meteor gray aluminum, appears slightly thicker than the X 14, likely to accommodate a dual-fan cooling solution for the higher TDP SKU of RTX Spark.

Key anticipated features:
- 16-inch 2.8K OLED, 120Hz, HDR600
- Up to 64GB LPDDR5x unified memory
- Two Thunderbolt 5 ports, USB-A, HDMI 2.1
- Wi-Fi 7 and Bluetooth 6.0
- 1080p IR camera with physical shutter
- Copilot+ key and dedicated AI assistant button

HP executives on stage emphasized that the Ultra 16 will be able to run locally-hosted AI models with up to 13 billion parameters without straining the system, thanks to the unified memory architecture that allows the GPU and NPU to access the same data without copying. This could make it the first consumer laptop to offer seamless, real-time AI denoising in photo apps or AI-assisted game upscaling directly within the operating system.

HP OmniBook X 14

The X 14 takes a thinner and lighter approach, clearly targeting professionals and students who want a Copilot+ PC they can carry everywhere. It shares design DNA with the existing OmniBook X 14 (which uses a Snapdragon X Elite), but the RTX Spark variant introduces a new cooling solution and a slightly larger battery to power the more capable NPU.

Key anticipated features:
- 14-inch 2.5K IPS or OLED, 90Hz, touch optional
- Up to 32GB LPDDR5x unified memory
- Two USB4 ports, USB-A, microSD reader
- Sub-2.5 lbs weight
- Studio-quality quad speakers
- Wi-Fi 7 and optional 5G

HP is billing the X 14 as the ultimate mobile AI companion, capable of handling Microsoft Teams’ real-time translation, advanced background effects, and even local Stable Diffusion image generation without a cloud connection. The integrated RTX Spark GPU is powerful enough to play AAA titles at 1080p with medium settings using DLSS, a feat that remains elusive for Arm-based ultrabooks today.

Both laptops will ship with Windows 11 version 24H2 and will receive the “late-2026 AI upgrade” as an over-the-air update when it becomes available. HP stopped short of committing to a specific launch date, but “later this year” points to a Q3 or Q4 2026 release, with the AI upgrade possibly landing in November or December.

The Late-2026 AI Upgrade: A Glimpse Into Windows’ Future

The “Prep Late-2026 AI Upgrade” language in HP’s press materials is more than marketing fluff. Microsoft has been working on a next-generation AI stack that goes far beyond the current Copilot sidebar. Multiple sources indicate that the upgrade will introduce:

  • Persistent Copilot: An assistant that lives in the system tray, can see and interact with any app, and remembers your habits across sessions. It will be able to complete multi-step tasks like booking travel or editing documents based on natural language commands.
  • AI-powered search: Local semantic search that indexes your files, emails, photos, and even videos, letting you find content by describing it rather than remembering file names.
  • Creative Studio: A system-level app that uses the RTX Spark NPU and GPU to generate images, music, and short videos locally, with advanced editing tools like AutoReframe, content-aware fill, and style transfer.
  • Windows Recall 2.0: A more privacy-focused and encrypted version of the controversial Recall feature, with on-device processing and automatic exclusion of sensitive content.
  • Game AI: NVIDIA’s DLSS 4 and Reflex integration into DirectX, giving RTX Spark laptops access to frame generation and latency reduction in any compatible game.

These features require a hardware foundation that current Copilot+ PCs based on Snapdragon X or Intel Meteor Lake cannot fully support due to NPU bandwidth limitations or lack of a powerful integrated GPU. RTX Spark’s unified memory architecture and high TOPS count make it the primary target for this upgrade, though it’s possible some features will trickle down to older devices with discrete NVIDIA GPUs. HP is betting that early adopters of the OmniBook Ultra 16 and X 14 will be poised to experience the most complete version of Windows’ AI vision.

Competition Heats Up: RTX Spark vs. Snapdragon X Elite vs. Intel Lunar Lake

HP’s announcement sets the stage for a three-way battle in the AI PC space. Qualcomm’s Snapdragon X Elite has a head start, with dozens of Copilot+ models already shipping. Intel’s Lunar Lake processors, now official, also pack a potent NPU and improved Arc graphics. But RTX Spark introduces a different value proposition: NVIDIA’s ecosystem.

Developers who have spent years optimizing for CUDA and TensorRT will find a familiar toolchain on RTX Spark. Porting AI models from desktop or server environments to the laptop becomes a matter of recompilation rather than rewriting for a different NPU SDK. Gaming studios, in particular, have deep experience with NVIDIA’s graphics drivers and DLSS technology. This could mean that RTX Spark laptops will offer a far superior gaming experience on Arm compared to Qualcomm’s Adreno-based platform, where x86 emulation overhead still clips performance.

Intel is countering with its own AI Playground app and broad x86 compatibility, but its integrated Arc GPUs lack the raw TOPS and developer mindshare that NVIDIA commands. HP’s strategy appears to be ecosystem diversification: it already sells Intel-based and Qualcomm-based OmniBooks. Adding an NVIDIA option gives customers a choice of AI platform, and HP likely expects that creators and gamers will gravitate toward RTX Spark.

Design Philosophy and Early Impressions

From the limited hands-on time offered to press after the Computex keynote, the OmniBook Ultra 16 and X 14 share HP’s now-familiar minimalist design language: clean lines, aluminum construction, and subtle keyboard decks. The X 14’s magnesium alloy chassis feels impossibly light, while the Ultra 16’s larger footprint allows for a centered keyboard without a numpad, which many creatives will appreciate.

One standout detail is the new “AI Boost” hotkey located next to the right Control key. HP says it will summon a context-aware AI toolbar that can offer quick actions like summarizing a webpage, generating a caption for a selected image, or toggling real-time translation in a video call. This hardware shortcut mirrors the Copilot key but focuses on instant, utility-driven AI tasks rather than a full assistant pane.

The displays, too, are clearly aimed at visual work. The Ultra 16’s OLED panel is Pantone-validated and factory-calibrated, while the X 14 offers a privacy screen option for business users. Both laptops feature 5-megapixel cameras that support Windows Studio Effects and automatic framing, leveraging the NPU to offload these tasks from the CPU and GPU.

Thermal management is always a question with powerful SoCs in thin chassis. HP claims that the RTX Spark models use a new vapor chamber cooling system with dual fans and liquid crystal polymer blades, a carryover from its Spectre line. The company promises quiet operation even during sustained AI workloads, though real-world testing is needed.

Pricing and Availability

HP did not reveal pricing at Computex, which is typical for products still months away from shelves. Given the positioning of the OmniBook Ultra as a premium creator laptop, it’s expected to start around $1,799, with the OmniBook X 14 likely priced from $1,299. These figures would place them slightly above equivalent Qualcomm-based models but within the range of Intel Evo premium laptops.

Availability is targeted for late Q3 or early Q4 2026, with pre-orders possibly opening in September. The “late-2026 AI upgrade” will be delivered via Windows Update, and HP confirmed that existing OmniBook RTX Spark owners will receive it at no additional cost. Whether the update will be visible as a new Windows version or a feature pack remains to be seen; Microsoft typically holds such news for its own events.

A Pivotal Moment for Windows on Arm

The HP-NVIDIA partnership, unveiled at Computex 2026, marks a turning point for Windows on Arm. For the first time, an industry heavyweight with deep AI and graphics expertise is entering the client SoC market, directly challenging Qualcomm’s dominance. The OmniBook Ultra 16 and X 14 are not just incremental updates; they are forward-looking devices designed to harness a wave of AI innovation that will crest later this year.

Users who purchase these laptops when they launch will effectively be buying a ticket to that future, trusting that HP, NVIDIA, and Microsoft will deliver on their ambitious roadmap. The “Prep Late-2026 AI Upgrade” promise sets high expectations, and any delay or feature downgrade could sour early adopters. But if the pieces fall into place, the RTX Spark platform could redefine what we expect from a thin-and-light Windows PC, making it as capable at AI tasks as it is at spreadsheets and streaming.

In the coming months, expect more OEMs to join the RTX Spark camp, and NVIDIA to host its own deep-dive sessions for developers. For now, the OmniBook Ultra 16 and OmniBook X 14 stand as the first concrete evidence that the next phase of AI computing is just a few months away — and it will arrive on the shoulders of a spark.