NVIDIA has unveiled RTX Spark, a new superchip that combines a Blackwell GPU with a Grace CPU to deliver up to 1 petaflop of AI computing power in slim laptops and compact desktops, marking a fundamental shift in how personal AI agents will run directly on consumer devices. Announced by CEO Jensen Huang at COMPUTEX 2026, the RTX Spark represents the culmination of 30 years of NVIDIA innovation, integrating CUDA cores, Tensor Cores, and specialized AI acceleration technologies into a form factor thin enough to fit in a 14-millimeter laptop.

What Makes RTX Spark Different From Previous NVIDIA Chips?

The RTX Spark superchip is engineered specifically for the era of on-device AI agents, combining hardware and software in ways previous consumer GPUs were not designed to handle. At its core sits an NVIDIA Blackwell RTX GPU with 6,144 CUDA cores (the parallel processing units that power AI workloads) paired with a 20-core NVIDIA Grace CPU. The two processors connect via NVIDIA NVLink-C2C, a high-speed chip-to-chip interconnect that allows them to work together seamlessly.

The real innovation lies in the unified memory architecture. RTX Spark offers 128 gigabytes of unified memory, meaning the GPU and CPU can access the same data pool without constantly copying information back and forth. This is critical for running large language models and AI agents locally, where latency and power efficiency matter. The chip delivers up to 1 petaflop of AI compute capacity, enough to run sophisticated AI models entirely on your device without sending data to cloud servers.

Security is baked in from the ground up. NVIDIA and Microsoft are partnering to deliver a secure Windows platform for on-device agents built on new operating system security primitives and NVIDIA OpenShell, a new runtime for Windows. This means AI agents running on RTX Spark will have robust protections against unauthorized access or data leakage.

How to Maximize RTX Spark's AI Performance?

• Enable Local AI Inference: RTX Spark achieves 2X faster inference performance on top agentic models through optimized runtimes like llama.cpp and vLLM, allowing you to run large language models directly on your device without cloud dependency.
• Leverage Adobe and Blender Integration: Adobe and Blender have been rebuilt specifically for RTX Spark, meaning creative professionals can run AI-accelerated workflows in real-time without waiting for cloud processing.
• Utilize DLSS 4.5 Ray Reconstruction: Coming this August, DLSS 4.5 Ray Reconstruction uses AI to improve image quality in ray-traced games and creative applications, delivering 35 percent more compute capability while maintaining similar performance to previous models.

Which Manufacturers Are Building RTX Spark Devices?

RTX Spark laptops and compact desktops will be available this fall from major hardware makers including ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with additional models from Acer and GIGABYTE to follow. The laptops will come in 14 to 16-inch sizes, engineered to be as slim as 14 millimeters and as light as 3 pounds. They feature precision-machined aluminum chassis, color-accurate tandem OLED displays with NVIDIA G-SYNC technology for smooth visuals during gaming and creative work.

Desktop versions are designed to be small and ultra-efficient, built for agents, creative workloads, gaming, and everyday productivity. Major hardware makers are already rallying around the RTX Spark platform with many designs in development, signaling strong industry confidence in the technology.

What Other AI Improvements Is NVIDIA Rolling Out?

Beyond RTX Spark, NVIDIA announced significant upgrades to its DLSS technology, which uses AI to enhance gaming and creative graphics. DLSS 4.5 Ray Reconstruction, arriving this August, introduces a new second-generation transformer model that improves image quality in ray-traced and path-traced games. The model delivers 35 percent more compute capability and processes 20 percent more parameters while maintaining similar performance to the previous version.

The new Ray Reconstruction model features several major improvements. An efficient denoiser removes visual noise from ray-traced scenes more intelligently. Enhanced super resolution gives the model deeper spatial awareness across every scene, improving lighting accuracy, temporal stability, and motion clarity. The model was trained on a larger dataset, making it better at reconstructing images closer to ground truth. Developers also gain finer control for temporal accumulation, allowing precise tuning of model response for even better image quality.

DLSS 4.5 Ray Reconstruction is coming to Blender 5.3 this fall, allowing 3D artists to maintain interactive viewport experiences while getting final-quality renders in real-time. This keeps creative workflows unlocked rather than forcing artists into a slow wait-and-see approach with traditional hand-tuned denoisers.

How Widespread Is DLSS Adoption Among Game Developers?

DLSS adoption continues to accelerate across the gaming industry. Over 1,000 RTX-enhanced and accelerated games and applications are now available, and developers are rapidly adopting DLSS 4.5. NVIDIA announced 11 additional games adding DLSS 4.5 support, including Backrooms: Escape Together, CINDER CITY, Duet Night Abyss, Gothic 1 Remake, Hell Let Loose: Vietnam, Honeycomb: The World Beyond, Marvel Rivals, NARAKA: BLADEPOINT, Phantom Blade Zero, Squad, and Where Winds Meet.

The breadth of adoption reflects how DLSS has become a standard feature for modern gaming. With over 1,000 games and apps already supporting RTX acceleration, developers view NVIDIA's technology stack as essential for delivering high-quality visuals and performance to players.

What Additional Features Did NVIDIA Announce for Gamers?

NVIDIA also rolled out several quality-of-life improvements for gamers and content creators. The NVIDIA App received an update adding 240 FPS ShadowPlay recording, allowing streamers and content creators to capture gameplay at ultra-high frame rates. New GeForce RTX partner cards, laptops, systems, and G-SYNC compatible displays were unveiled at COMPUTEX, expanding the ecosystem of hardware optimized for NVIDIA's technology.

The company also launched a GeForce Rewards program, allowing users to claim a World of Tanks: HEAT pack while supplies last. These incremental improvements reflect NVIDIA's strategy of deepening integration across gaming hardware, software, and services.

RTX Spark represents a watershed moment for consumer AI. By bringing petaflop-scale computing power to devices thin enough to slip into a backpack, NVIDIA is fundamentally changing where AI inference happens. Instead of sending your data to distant cloud servers, personal AI agents will run directly on your device, offering faster response times, better privacy, and lower latency. With major manufacturers already building RTX Spark systems and availability set for this fall, the era of truly personal AI is arriving faster than many expected.