Computex 2026 opened in Taiwan with a flood of new laptops, gaming silicon and displays from Microsoft, Nvidia, AMD, Acer, HP and Samsung. Running through almost every marquee machine on the first day was one piece of hardware: Nvidia’s RTX Spark superchip, the platform now sitting at the heart of the most powerful Windows laptops at the show. The trade show runs June 2 to June 5, with keynotes that began on June 1.
The headlines read like a familiar hardware roundup. The development that is harder to spot in a spec sheet is that Nvidia has moved from supplying the graphics card to owning the whole compute platform, an Arm-based bid that drops it into direct competition with Intel, AMD and Qualcomm for the brain of the PC.
Inside the RTX Spark Superchip
Jensen Huang, Nvidia’s chief executive, used the show to launch RTX Spark, a part formerly known by the codenames N1 and N1X. It folds a processor, a graphics engine and memory onto a single platform aimed at running personal AI agents directly on the device rather than in the cloud. Nvidia has been clear that the chip is built as much for developers as for shoppers, a pitch laid out in its RTX Spark developer strategy for the PC market.
- 6,144 CUDA cores (Compute Unified Device Architecture, Nvidia’s parallel-computing engine) on a Blackwell RTX graphics block with fifth-generation Tensor Cores
- A 20-core Arm-based Grace CPU (central processor) co-developed with MediaTek, joined to the GPU by an NVLink chip-to-chip link
- 128GB of unified memory shared dynamically between the processor and the graphics engine
- Up to 1 petaflop of AI compute (one thousand trillion operations per second), enough to run models with 120 billion parameters locally
The unified memory is the part that matters most. By letting the CPU and GPU draw from the same large pool, the chip can hold a big language model in memory without shuttling data back and forth, which is what makes on-device agents practical on a thin laptop.
It also marks a strategy shift. Nvidia spent decades selling the add-in graphics card; the Grace pairing means it is now selling the processor too, with full CUDA support carried over so existing AI and creative tools run without a rewrite.
Microsoft and HP Hand Their Flagships to Nvidia
Microsoft put the chip front and center with the Surface Laptop Ultra, which the company calls the first Surface built on Nvidia silicon from the die up. It carries a 15-inch mini-LED PixelSense Ultra touchscreen rated at 2,000 nits of peak HDR (high dynamic range) brightness, weighs under 2kg, measures under 18mm thick, and packs the full 128GB memory ceiling. It ships later this year at a price Microsoft has not set.
HP leaned in just as hard with the OmniBook Ultra 16 and the OmniBook X 14, both running RTX Spark, which the company bills as the world’s thinnest laptops to carry the chip. HP also signaled a line of compact desktops on the same platform. The wider partner roster sits in Nvidia’s Computex 2026 GeForce RTX partner showcase, which lists Dell, ASUS, Lenovo and MSI alongside them.
Acer took the opposite road. Its Predator Helios 18 AI sticks with a discrete graphics card, pairing an Intel Core Ultra 9 290HX processor with an Nvidia GeForce RTX 5090, up to 256GB of memory and up to 6TB of storage behind an 18-inch 4K mini-LED panel running at 240Hz. It is the brute-force gaming approach next to the integrated AI-first design.
The split is easy to read across the three headline machines.
| Laptop | Chip | Display | Memory |
|---|---|---|---|
| Microsoft Surface Laptop Ultra | RTX Spark (Grace + Blackwell) | 15-inch mini-LED, 2,000 nits | Up to 128GB unified |
| HP OmniBook Ultra 16 | RTX Spark (Grace + Blackwell) | 16-inch class | Up to 128GB unified |
| Acer Predator Helios 18 AI | Intel Core Ultra 9 + RTX 5090 | 18-inch 4K mini-LED, 240Hz | Up to 256GB |
The Incumbents Nvidia Just Stepped On
By building its own Arm processor and selling it as the core of the machine, Nvidia is going after the brain of the PC, not just its graphics. That puts pressure on the three companies that have owned that slot, and the move was telegraphed before the show in coverage of Nvidia’s first Windows PC processors set to rattle Intel.
- Intel: the x86 incumbent loses the marquee Surface design to an Arm part, even as it keeps high-end gaming sockets like the one in Acer’s Predator.
- AMD: its Ryzen AI line now shares the premium-laptop conversation with a rival that brings its own GPU pedigree.
- Qualcomm: the Snapdragon X push made Arm laptops mainstream, and Nvidia is now contesting that ground, as detailed in the leak around Nvidia’s N1X and Qualcomm’s Arm hold on Windows.
None of this is settled. Windows on Arm still carries app-compatibility gaps for older x86 software, Nvidia has not published pricing, and the incumbents own enormous installed bases that do not move on one keynote. The opening has cracked, not the moat collapsed.
AMD’s Counter: X3D Revival and AM5 to 2029
AMD answered with a mix of nostalgia and a longevity promise. The standout is a Ryzen 7 5800X3D 10th Anniversary Edition for the older Socket AM4, marking ten years of that platform and shipping with Carbonaut thermal material in the box.
For current builders, the company added the Ryzen 7 7700X3D for Socket AM5, an eight-core chip with 104MB of total cache and boost clocks up to 4.5GHz, plus a global launch for the Radeon RX 9070 GRE graphics card with 12GB of video memory aimed at 1440p play. New EXPO memory profiles with ultra-low latency round out the gaming tweaks.
| Product | Platform | Price | Availability |
|---|---|---|---|
| Ryzen 7 5800X3D 10th Anniversary | Socket AM4 | $349 | June 25 |
| Ryzen 7 7700X3D | Socket AM5 | $329 | July 16 |
| Radeon RX 9070 GRE | Graphics card | Regional pricing | Global rollout |
AMD also confirmed Socket AM5 support through 2029, a pledge that lets buyers upgrade for years on one board. It is a different bet from Nvidia’s: instead of redefining the platform, AMD is selling continuity and value. The full plan sits in AMD’s Computex 2026 platform roadmap.
Samsung Display Builds Screens for the AI PC
The chips need panels, and Samsung Display brought close to 16 gaming OLED and QD-OLED (quantum-dot OLED) products to the floor in Taiwan, spanning handhelds to ultrawide monitors.
- An 8.8-inch OLED panel built for handheld gaming PCs
- A 31.5-inch (32-inch class) QD-OLED monitor panel using Penta Tandem technology
- A 49-inch ultrawide OLED gaming panel
- An Ultra Slim laptop OLED, more than 20% thinner at the module edge, spanning 165Hz to 240Hz refresh rates
The headline is the monitor panel, which Samsung says is the world’s first self-emissive screen to combine 4K resolution with a 360Hz refresh rate. The company credits a reworked internal circuit and drive system, plus a Penta Tandem stack that adds a fifth blue OLED layer for more brightness and longer life.
Lower down the size range, the 8.8-inch panel feeds the handheld category that has grown fast, while the Ultra Slim laptop screen is built for exactly the kind of thin, premium notebook the new chips target.
That alignment is the quiet point. The same show that put a new processor inside ultrathin laptops also produced the thin, high-refresh, color-accurate displays those laptops will use, as set out in Samsung Display’s full gaming OLED lineup for Computex 2026.
When the RTX Spark Laptops Reach Buyers
The calendar is the near-term story. Nvidia says RTX Spark laptops and small desktops arrive this fall from partners including Microsoft, HP, Dell, ASUS, Lenovo and MSI, with Acer and GIGABYTE models following; the Surface Laptop Ultra is slated for later in the year. AMD’s two new X3D chips land first, on June 25 and July 16, and DLSS 4.5 Ray Reconstruction is set for August. Full numbers sit in the RTX Spark specifications and partner list from Nvidia.
If the fall machines ship near MacBook Pro money and Windows on Arm runs the apps buyers expect, Nvidia turns a keynote into a platform. If the price comes in high or the software gaps bite, the incumbents get another year to answer.
Frequently Asked Questions
What is the Nvidia RTX Spark superchip?
RTX Spark is a single platform that combines a 20-core Arm-based Grace CPU, a Blackwell RTX GPU with 6,144 CUDA cores and up to 128GB of unified memory. Nvidia rates it at up to 1 petaflop of AI compute, enough to run models with 120 billion parameters on the device. It was previously known by the codenames N1 and N1X.
When do RTX Spark laptops go on sale?
Nvidia says RTX Spark laptops and compact desktops arrive in fall 2026 from partners including Microsoft, HP, Dell, ASUS, Lenovo and MSI, with Acer and GIGABYTE models following. Pricing had not been announced as of the show.
What is the Microsoft Surface Laptop Ultra?
It is Microsoft’s first Surface built on Nvidia’s RTX Spark silicon, with a 15-inch mini-LED display at 2,000 nits peak HDR brightness, up to 128GB of unified memory, a chassis under 18mm thick and under 2kg. Microsoft says it ships later in 2026 without confirming a price.
How much do AMD’s new Ryzen X3D chips cost?
The Ryzen 7 5800X3D 10th Anniversary Edition for Socket AM4 is $349 and arrives June 25. The Ryzen 7 7700X3D for Socket AM5 is $329 and arrives July 16. AMD also confirmed it will support Socket AM5 through 2029.
What did Samsung Display show at Computex 2026?
Samsung Display brought close to 16 gaming OLED and QD-OLED panels, led by a 31.5-inch QD-OLED it calls the world’s first self-emissive panel to pair 4K resolution with a 360Hz refresh rate. The lineup also included an 8.8-inch handheld panel, a 49-inch ultrawide and a thinner laptop OLED.
