Intel previews Computex 2026 lineup across handhelds, desktops, and servers as 18A process becomes casting calling card

Intel will arrive Computex 2026 on June 2 in Taipei with something that hasn’t happened in a decade: a product in every computing category built on a single production story. Launched at CES in January, laptop chip Panther Lake is expanding into handhelds with the Arc G3 and Arc G3 Extreme processors, aimed at the gaming handheld market. Nova Lake, a 52-core desktop chip with a new socket and new CPU architecture, will be introduced in the second half. Clearwater Forest, a 288-core server processor shipped at MWC in March, completes the Xeon lineup for data centers and cloud computing. They’re all built on or around the Intel 18A, a 1.8-nanometer process node that combines RibbonFET gate and all-in-one transistors with a PowerVia rear power supply and represents the most advanced semiconductor manufacturing capability made entirely in the USA. CEO Lip-Bu Tan will deliver the keynote address. The location is 40 kilometers from TSMC’s headquarters. The message is not subtle.

Products

Panther Lake was introduced at CES in January as the Core Ultra Series 3 and is already shipping in more than 200 laptop designs. The chip delivers 180 total platform TOPS, combining 120 TOPS from the Xe3 integrated GPU with 50 TOPS from the NPU 5 neural processing unit, and claims a 60 percent improvement in multi-threaded performance over its equivalently powered predecessor. The Computex expansion brings Panther Lake to gaming handhelds via the Arc G3 platform: a 14-core design with two performance cores, eight efficiency cores, and four low-power cores paired with a 10- or 12-core Xe3 GPU in a power envelope configurable between 25 and 80 watts. MSI, OneXPlayer, GPD and Acer are expected to showcase handhelds powered by Arc G3 chips at the event, with reports suggesting a Microsoft Xbox-branded handheld could also make an appearance.

The Core Ultra Series 4-branded Nova Lake is Intel’s next desktop platform and will be previewed at Computex ahead of its launch in late 2026. The chip has 8 to 52 cores using the new Coyote Cove performance cores and Arctic Wolf efficiency cores, features an LGA 1954 socket, and integrates Xe3 graphics, Thunderbolt 5 and Wi-Fi 7. The power range is between 35 and 175 watts, featuring a design that covers both basic duty and high voltage duty. Nova Lake adopts what Intel calls a “large end-level cache” architecture, a design approach inspired by AMD’s success with large L3 caches that prioritize data storage close to the CPU cores. Intel’s first-quarter earnings revealed that demand for AI-powered CPUs is real: Data center and AI revenue grew 22 percent to $5.1 billion for the year as the agency shifted the processing demands of AI workloads back to CPUs and away from a GPU-only model that defined the training cycle.

Server

💜 of EU technology

The latest rumblings from the EU tech scene, a story from our wise founder Boris and some questionable AI art. Free in your inbox every week. Register now!

Officially unveiled at MWC in March as the Xeon 6+, the Clearwater Forest is Intel’s most architecturally ambitious server processor. It combines 288 Darkmont efficient cores in 12 compute chiplets manufactured at 18A assembled with Foveros Direct 3D stack on Intel 3-built baseplates. The IPC increase is 17 percent over the previous generation, and the chip is targeted at cloud computing and compute-intensive workloads. The agency’s shift to artificial intelligence is increasing the demand for outcome calculations among all the major cloud providers: Meta has committed more than $140 billion to sourcing chips from Nvidia, AMD, and Amazon, and the inference workloads served by these chips are increasingly demanding the processor resources required by autonomous AI agents for orchestration, memory management, and real-time decision-making.

Intel’s server story at Computex also included updates on Crescent Island, its dedicated inference accelerator, and Jaguar Shores, a rack-scale computing platform slated for an AI data center in the late 2020s. Neither product has been officially launched, but both are expected to receive architectural details at Tan’s keynote. The inference accelerator is Intel’s attempt to directly compete with Nvidia’s inference-optimized products, rather than taking over the AI ​​accelerator market entirely. Whether Intel can build a chip that can deliver a competitive bottom line while simultaneously expanding its foundry business and launching three client platforms is an operational question that Computex won’t answer, but can’t raise.

Process

The wire connecting every product at Computex is 18A. Panther Lake is the first consumer chip built on a node. Clearwater Forest is the first server chip. Arc G3 handheld processors are the first gaming-focused silicon. Nova Lake will be the first desktop chip, although reports suggest that more than 90 percent of Nova Lake’s compute wafers will be manufactured by TSMC on the N2 process rather than in Intel’s own fabs, a concession to the fact that Intel’s foundry capacity is not yet sufficient to satisfy both domestic demand and external customers simultaneously.

This concession is important because the 18A node is not just a manufacturing process. It’s the product Intel sells to Apple, Amazon, Musk’s Terafab, and every other company that has signed or is negotiating a foundry deal. Intel recently hired Qualcomm veteran Alex Katouzia to lead its new Client Computing and Physical AI group.a signal that the company is looking at the result of local AI running on PCs, handhelds and peripherals rather than in cloud data centers as the next wave of chip demand. The Computex product line is a proof of concept: if 18A can produce competitive chips across notebooks, handhelds, desktops, and servers, the casting process for foreign customers becomes significantly more reliable than a roadmap slide.

market

GitHub suspended new Copilot registrations after agent AI workflows consumed more computing than users paid foran early signal that the agent AI economy will shift processing to local hardware. If AI agents continuously run on cloud infrastructure, costs scale linearly with usage and eventually become unsustainable with fixed subscription prices. If these agents run locally, on a laptop with 180 TOPS of AI processing power or a desktop with 52 cores and a large cache, the economics shift from per-request cloud payments to one-time hardware purchases. Intel’s claim is that AI is an architectural requirement, not a PC marketing label: the agent cycle needs native computing, and Intel’s chips are designed to provide it.

The competition is real. Qualcomm’s Snapdragon X Elite has established itself in the thin and light Windows market with high energy efficiency. AMD’s Ryzen AI 300 and 400 series compete directly with Panther Lake in laptops and Arrow Lake Refresh in desktops. Apple’s M series processors remain the benchmark for integrated performance in the consumer market. Nvidia’s server GPUs sell for a million dollars each in China despite export controls, this reflects a level of demand that Intel’s data center products have never generated. Intel’s advantage at Computex isn’t that its chips are the best in any category. It has chips in every category, all produced at the same time in a process hub that is the core of the foundry business, and the foundry business is the reason Apple is negotiating, Musk is building a $25 billion factory, and the US government owns 10 percent of the company.

stakes

Computex has been an Intel event for decades. The show is held in Taipei, the heart of the global semiconductor supply chain, and Intel has traditionally used it to announce products that define each generation of computing. The difference in 2026 is that Intel is no longer a chip designer delivering products manufactured in its own fabs. The host is a foundry operator competing with the nation’s most valuable company for the right to manufacture other people’s chips. Tan’s keynote will be watched not only for what Intel announces about its products, but also for whether those products are ready to serve 18A’s external foundry customers. Every Panther Lake laptop shipped flawlessly, every Clearwater Forest server that meets performance requirements, every Arc G3 handheld running in a thermal envelope is a data point for every company evaluating whether to trust Apple, Amazon, and Intel with their silicon.

In 2016, Intel was the world’s largest semiconductor company by revenue. By 2024, Nvidia has fallen to eighth place behind TSMC, Samsung, Broadcom, Qualcomm, AMD and Texas Instruments. Its manufacturing process was two generations behind TSMC. Its CEO was fired by a board that had lost faith in the turnaround. The stock crossed $18. Fourteen months later, Intel is at an all-time high, its foundry has customers including Apple and Musk’s Terafab, and it’s heading to Computex with a product in every category for the first time in a decade. The shift is real, but it’s also incomplete: The foundry is losing $2.4 billion per quarter, foreign revenue is $174 million versus TSMC’s $20 billion, and 90 percent of the desktop chip Intel previewed at Computex will be made by the rival it’s trying to displace. The 18A node is Intel’s answer to all these problems. Computex is where the answer starts to prove it works.