Intel (NASDAQ:INTC) used its CES 2026 event to outline how the company is positioning its client and edge roadmap around what executives described as a strategic inflection point for AI computing. Speakers highlighted progress on Intel’s 18A manufacturing technology, provided details on the upcoming Core Ultra Series 3 platform, and emphasized a broader push to scale AI capabilities across PCs and edge devices through hardware, software, and ecosystem partnerships.
Intel highlights 18A progress and Panther Lake ramp
Opening remarks framed AI as reshaping “every workflow, every industry, and every device,” with Intel’s stated mission to make intelligence “accessible, efficient, and ubiquitous.” Intel said it has met its commitment to ship its first 18A products by the end of 2025, adding that it has “over-delivered.” The company also said it is ramping “all three Core Ultra Series 3 die packages” and that the day’s announcements were enabled by tight integration of silicon design, process technology, and packaging.
Johnson said RibbonFET is Intel’s gate-all-around transistor technology designed for precise current control to improve performance and energy efficiency, while PowerVia is a backside power delivery approach intended to enhance power flow and signaling. Intel said these advances enable up to 15% better performance per watt and more than 30% better chip density.
Core Ultra Series 3 positioned as an AI-first PC platform
Johnson described Core Ultra Series 3 as the first processor built on 18A and said Intel improved “virtually every piece of the chip design.” He outlined key silicon elements including new E-cores, new P-cores, a larger GPU with built-in ray tracing, and an NPU designed to deliver more low-power AI capability in a denser footprint than prior generations. He also cited next-generation memory, I/O, and connectivity enhancements.
Intel said it used the power-efficiency foundation of Lunar Lake as the architectural basis for Series 3 and its roadmap going forward. Among the changes, Intel said it moved the GPU tile into its own chiplet, enabling larger or smaller GPU chiplets to be paired with the CPU depending on market segment and device needs. Johnson also said Intel redesigned cores for 18A to operate at lower voltage for higher per-core performance and efficiency, and improved multi-threaded performance by adding up to eight E-cores.
Johnson highlighted a redesigned “low-power island” featuring dedicated low-power E-cores with their own cache, aimed at maximizing battery life for workloads such as web browsing and video conferencing. He also said Intel consolidated I/O onto a single chip, which he framed as a request from customers seeking broader design flexibility.
At the platform level, Johnson pointed to features including Intel’s “intelligent display technology,” context-aware charging, low-power video conferencing using an image processor, and Wi-Fi-related connectivity improvements. He said Intel’s Foveros packaging supports a range of Series 3 configurations using a single package type, enabling multiple memory and power delivery options to support OEM designs.
On performance, Johnson said Core Ultra Series 3 delivers 60% more performance than Lunar Lake Series 2. He also cited power efficiency improvements, stating that streaming 4K video on Series 3 draws “one-third the power than prior generations,” which he said supports battery-life expectations “in days, not in hours.”
Intel introduces Arc B390 integrated graphics and modern rendering features
Dan Rogers, Intel’s General Manager of PC products, focused on Intel’s graphics progress and the Arc brand. He said Intel has re-architected its software stack and strategy with “game day testing” and “day zero driver support,” noting that Intel engaged 300 developers on pre-release titles and supported 50 day-zero driver releases over the last year. He also referenced Battlemage as “one of the best-reviewed cards of 2025.”
Intel announced a next-generation integrated GPU for Panther Lake: Intel Arc B390 graphics (previously referenced as “12Xe”). Rogers said Arc B390 includes:
- 50% more graphics cores
- Twice the cache
- 96 built-in XMX AI accelerators for 120 GPU TOPS
Rogers said Panther Lake delivers 70% more gaming performance and 50% more AI inference performance versus Lunar Lake. He also said Intel testing across APIs from DX9 to DX12 showed Arc B390 delivering smooth gameplay and, compared to “the latest from AMD Radeon with similar power and similar memory,” up to 70% higher average frame rate and 2x faster performance on select titles. As examples, he cited Painkiller and Delta Force running at well over 100 frames per second at 1080p high settings with XeSS super resolution.
Rogers also said the Arc graphics in Panther Lake will be the first integrated graphics solution to ship on day one with AI-based multi-frame generation, delivering “three AI-generated frames for every one rendered frame.” He used Battlefield 6 as an example, stating that with multi-frame generation enabled, Arc B390 could reach over 120 frames per second and be nearly three times faster than AMD under the scenario described.
EA technology partnerships executive Jeff Skelton discussed collaboration with Intel on Battlefield performance, citing optimization efforts for frame rate, stutter reduction, and stability across desktops and mobile processors like Panther Lake. Rogers said Intel worked with EA from early Panther Lake silicon to ensure robust driver support and to implement XeSS technologies, and Skelton said EA is working on “a native integration of XeSS 3.”
Rogers added that Intel plans to launch “an entire handheld gaming platform” with Panther Lake, with more details expected from partners later in the year.
AI software ecosystem, hybrid AI, and edge expansion
Johnson said delivering “new and better experiences” for AI PCs depends on significant software investment, describing the AI PC category as a partnership between hardware and the software ecosystem. He said Intel supports hundreds of ISVs with tools to optimize AI applications across CPU, GPU, and NPU. He cited Adobe Premiere Pro using the Arc GPU for media search via natural language prompts and Zoom using the NPU for a “virtual ring light” effect that brightens the subject while dimming the background.
Johnson also said Intel provided day-zero support in October for Alibaba’s Qwen 3 large language models on Core Ultra, and noted support for deployment paths including llama.cpp and PyTorch. He said Intel’s OpenVINO tools provide production-ready optimizations across Intel engines, and claimed Intel is “the only silicon vendor supporting APIs on all three processors: CPU, GPU, and NPU” in Windows ML. He also said Core Ultra Series 3 systems will ship with Copilot+ support across all SKUs.
On AI performance, Johnson said Core Ultra Series 3 provides up to 180 total platform TOPS, including 120 GPU TOPS and a 50 TOPS “always on” low-power NPU. He also said that with OpenVINO and access to 96GB of memory, the SoC can handle a 70-billion-parameter model with a 32K context locally.
Intel described hybrid AI—splitting tasks between local and cloud—as a major direction for the industry. Johnson cited ByteDance’s use case where a cloud-based CapCut feature, AI Clipper, was straining cloud capacity, leading ByteDance to pair cloud processing with Intel AI PCs for “same quality, better performance, lower cloud cost.” He also introduced Intel’s “AI Super Builder” platform, described as enabling local AI for secure on-device execution while cloud AI handles global reasoning and orchestration.
Perplexity CEO Arvin described localized compute as increasingly important for performance, privacy and security, economics, and enterprise control, and said Perplexity will launch Comet for enterprise next month.
Johnson said Intel is bringing Series 3 to the edge, and due to growing demand, the company is accelerating the launch of 18A products for edge to align with the PC launch. He highlighted edge requirements including temperature validation, 24/7 reliability, and extended life support, and cited form factors spanning smart cities, factories, healthcare, and automation. Intel also referenced a robotics-focused reference board and developer kit with Intel’s robotics suite, intended to be available at hardware launch.
In closing, Intel said partners have “over 200 designs” in development and that consumer designs would be available to order starting “tomorrow,” with more designs rolling out throughout 2026. Johnson characterized Core Ultra Series 3 as Intel’s most broadly adopted and globally available AI PC platform to date, emphasizing improvements in battery life, graphics, overall performance, and AI capability.
About Intel (NASDAQ:INTC)
Intel Corporation, founded in 1968 by Robert Noyce and Gordon E. Moore and headquartered in Santa Clara, California, is a leading global designer and manufacturer of semiconductor products. The company is historically notable for introducing the first commercial microprocessor and for driving the x86 architecture that underpins many personal computers and servers. Intel’s core business spans the design, fabrication and marketing of processors, chipsets and related components for a wide range of computing applications.
Intel’s product portfolio includes client and mobile processors marketed under brands such as Intel Core and Pentium, as well as high-performance Xeon processors for data centers and cloud infrastructure.
