CES 2026: A Comprehensive Field Guide

Beyond the Hype Cycle

The Consumer Electronics Show (CES) has historically functioned as the technology industry’s primary calibration point, a massive, chaotic metronome that sets the tempo for the ensuing twelve months of hardware releases, software integrations, and supply chain logistics. As the industry descends upon Las Vegas for CES 2026, the rhythm appears to be shifting. If the previous three years were defined by the explosive, often abstract arrival of Generative AI concepts—characterized by "what if" scenarios and cloud-based promises—2026 is shaping up to be the year of infrastructure. We are moving from the "Hype Phase" to the "Build Phase." The focus for the editorial team at RADAR, and our colleagues over at Configurine who track the granular component compatibility of these systems, is on "Physical AI" and the maturation of the hardware required to run these models locally, efficiently, and invisibly.¹

This report aggregates intelligence from early vendor briefings, supply chain leaks from manufacturing hubs in Taiwan and South Korea, and confirmed exhibition rosters to provide an exhaustive, expert-level analysis of what over 4,000 exhibitors will bring to the Las Vegas Convention Center (LVCC), the Venetian Expo, and the Wynn.³ The narrative emerging from the pre-show noise is one of optimization and industrialization. The questions have shifted from "Can AI do this?" to "what is the thermal design power (TDP) required to run this agent locally?" and "Is the NPU (Neural Processing Unit) capable of simultaneous translation and image generation without throttling?"

For the RADAR audience—predominantly composed of systems architects, high-end enthusiasts, and technical professionals—this pivot is welcome. It signals a return to measurable performance metrics, tangible hardware specifications, and the nuanced engineering challenges of thermal management and power delivery. We will dissect the silicon capabilities of NVIDIA’s Blackwell architecture, the architectural shifts in Intel’s Panther Lake 18A process, the pragmatic pivot of AMD’s RDNA 4 graphics strategy, and the confusing rebranding of display technologies by major Korean conglomerates.

The Temporal Landscape: Schedule and Logistics

Understanding the flow of CES is critical for interpreting the cadence of news. The event is structurally bifurcated into media-exclusive previews and the general exhibition, a distinction that often dictates when embargoes lift and when hands-on performance data becomes available.

The Pre-Show Sprint: Media Days

While the show floor officially opens on Tuesday, January 6, the narrative framework of CES 2026 is established 48 hours prior. The timeline kicks off on Sunday, January 4, traditionally reserved for "Media Day 1".⁴ This is when the largest consumer brands attempt to preempt the noise of the convention center.

• Samsung's First Look: Samsung continues its tradition of dominating Sunday evening with "The First Look," hosted at the Wynn rather than the LVCC. This event, led by TM Roh of the DX Division, will focus on the "Vision for Device eXperience," setting the stage for their display and AI-home announcements.⁴
• CES Unveiled: Running parallel on Sunday evening is CES Unveiled at Mandalay Bay ⁵, a tabletop exhibition that serves as a bellwether for startup trends and Innovation Award honorees before the massive halls open.

Monday, January 5, is the primary day for "heavy iron" announcements.⁶ This is the day the silicon giants take the stage:

• LG Electronics starts the morning with its "Innovation in Tune with You" press conference.⁴
• NVIDIA commands the midday slot (1:00 p.m. PT), where Jensen Huang is expected to unveil the RTX 50-series.⁶
• Intel follows at 3:00 p.m. PT, focusing on the Core Ultra Series 3 "Panther Lake".⁶
• AMD closes the major silicon announcements at 6:30 p.m. PT with Lisa Su’s keynote, likely covering Ryzen 9000 X3D and RDNA 4.⁶

The Exhibition: January 6–9

The LVCC doors open on Tuesday, January 6. This phase is characterized by the shift from slide decks to physical silicon. For Configurine users, this is when we begin to see the motherboard designs, the cooling solutions, and the chassis integrations that will define PC building for the year. The sheer scale of the event—spanning over 2.5 million net square feet across the LVCC (Central, West, North Halls), the Venetian Expo, and Aria ⁷—mandates a strategic approach to coverage. The "West Hall" has effectively become an auto show, while "North Hall" is increasingly dominated by robotics and enterprise AI.⁸

The Silicon Foundations: GPU Architectures

The central nervous system of the consumer technology industry remains the processor. For the demographic we serve—those who build, optimize, and maintain high-performance computing environments—the trajectory of the "Big Three" (NVIDIA, AMD, Intel) defines the capabilities of the coming year. 2026 sees a distinct divergence in strategy: NVIDIA is doubling down on dominance in the high-end discrete GPU market with massive dies and high power consumption, while AMD is pivoting to a calculated game of value and efficiency, and Intel fights for survival in the discrete market.

NVIDIA: The Blackwell Era Begins

The anticipation surrounding NVIDIA’s presence at CES 2026 is centered entirely on the GeForce RTX 50-series, codenamed "Blackwell." Jensen Huang’s keynote is expected to frame this not just as a gaming launch, but as the arrival of the "AI PC" engine.⁹ While the enterprise variants of Blackwell (B100/B200) have been public for some time, driving the company's valuation to stratospheric heights ¹⁰, the consumer implementation is expected to bring significant changes to the gaming and prosumer market.

The RTX 5090: The New Titan

The flagship RTX 5090 is rumored to be a massive piece of silicon, a true "halo product" designed to establish unassailable dominance. Leaks suggest a transistor count of roughly 92 billion, a significant jump over the AD102 die of the Ada Lovelace generation.¹¹ The critical architectural shift here is not just raw rasterization performance—though a 50-70% uplift is expected—but the integration of fifth-generation Tensor Cores tailored for local LLM (Large Language Model) inference.

• Memory Architecture: We expect the inclusion of GDDR7 memory. This is a pivotal upgrade from the GDDR6X used in the 40-series. GDDR7 utilizes PAM3 signaling (Pulse Amplitude Modulation with 3 levels) to achieve substantially higher bandwidth efficiency. This is crucial for driving 8K texture assets and, more importantly, for swapping large AI model weights in and out of VRAM quickly.¹²
• Power Consumption and Thermal Density: Early thermal profiles discussed in supply chain circles suggest the 5090 may retain or slightly exceed the 450W TDP (Thermal Design Power) of the 4090. However, the density of the heat generation due to the smaller process node may require more aggressive cooling solutions. We expect to see a wave of new AIB (Add-in Board) partner designs featuring vapor chambers and 4-slot coolers. This will necessitate robust PSU solutions—a topic we will cover extensively on Configurine’s power supply tier lists post-launch.
• Pricing Dynamics: Economic indicators from Korean manufacturing sources suggest a price hike is inevitable. The RTX 5090, which launched at $1,599 in the previous generation, may see a listing closer to $1,999 or even higher, driven by the high cost of GDDR7 and the insatiable industry demand for any silicon capable of AI workloads.¹²

The RTX 5080 and the "Super" Question

The RTX 5080 is expected to launch simultaneously or shortly after, potentially creating a large performance gap between the 90 and 80-class cards, similar to the 40-series launch dynamics.¹³ While some outlets suggest a mid-cycle "Super" refresh of the 50-series appearing at CES 2026 ¹⁰, the timing suggests a primary generation launch is more likely. The "Super" nomenclature is typically reserved for mid-generation corrections (12-18 months post-launch), making 50-series Super cards a conversation for CES 2027.

AMD: RDNA 4 and the Strategic Pivot

AMD’s strategy for CES 2026 appears fundamentally different from NVIDIA's. CEO Lisa Su will take the stage on the evening of January 5 ⁶, likely to clarify the company’s dual-track approach: dominating the mid-range GPU market and retaining the gaming CPU crown.

RDNA 4 (Radeon 8000 Series)

Contrary to NVIDIA’s push for the ultra-enthusiast segment, AMD’s RDNA 4 architecture (likely branded as the Radeon 8000 series, though rumors of a jump to the 9000 series to align with CPU branding persist ¹⁴) reportedly abandons the high-end battle entirely. There will likely be no competitor to the RTX 5090. Instead, AMD is targeting the volume segment—the $400 to $700 range—focusing on rasterization value and significantly improved ray tracing efficiency.¹⁵

The chip to watch is "Navi 48." This silicon aims to deliver RTX 4080-class performance at a significantly lower price point and power envelope.¹⁶ For RADAR readers building mid-tower systems where thermal management and budget are constraints, this could be the most relevant announcement of the show. The absence of a flagship "Navi 41" signals a maturity in AMD's strategy: acknowledging that the halo product war is costly and yields low volume, while the mid-range is where market share is won or lost.

Intel: The Battlemage Correction

Intel’s position in the discrete GPU market going into CES 2026 is precarious. Following the mixed reception and driver struggles of the Arc Alchemist (A-series), the company is rumored to unveil the Arc B-Series, codenamed "Battlemage." Specifically, the flagship Arc B770 is expected to debut.¹⁷

• Specifications: Leaks point to the "Big Battlemage" BMG-G31 die featuring 32 Xe2 cores and 16GB of GDDR6 VRAM on a 256-bit bus.¹⁷
• The Target: Intel is targeting the "performance segment," likely aiming to disrupt the RTX 4070 / RTX 5060 tier.
• The Configurine Perspective: For system builders, the key metric will be the maturity of the drivers on day one. Intel has spent two years refining its software stack, and Battlemage represents the test of whether that investment has paid off. A 300W TDP is rumored ¹⁸, which is high for this performance tier, suggesting Intel may be pushing clocks aggressively to remain competitive.

The CPU Landscape: Cache, Nodes, and NPUs

While GPUs capture the imagination, CPUs determine the platform's longevity. 2026 is a pivotal year for process node technology and cache architecture.

Intel: The 18A Era with Panther Lake

Intel is pivoting to Panther Lake (Core Ultra Series 3) for mobile, marking a critical milestone for its foundry business.⁹

• Process Node: This is the first mainstream consumer chip built on the Intel 18A node.¹⁹ 18A is Intel's attempt to regain process leadership from TSMC, utilizing RibbonFET (gate-all-around) transistors and PowerVia (backside power delivery).
• Mobile Focus: Leaks from Lenovo product roadmaps indicate "Core Ultra X" branding for high-performance mobile parts.²⁰ Expect to see these chips in the next generation of ThinkPad X1 Carbons and Dell XPS 13s. The focus is heavily on power efficiency to combat the rising threat of ARM-based Windows laptops.
• Integrated Graphics: Panther Lake will feature Xe3 integrated graphics ("Celestial"), which promises another leap in handheld and ultrabook gaming performance.¹⁹

AMD: Ryzen 9000 X3D Refinement

On the desktop front, the "X3D" technology remains AMD’s ace for gamers. We expect the full unveiling of the Ryzen 9 9950X3D2 and the Ryzen 7 9850X3D.²¹

• The 9950X3D2 Innovation: The suffix "2" is critical. Leaks suggest this chip will feature 3D V-Cache on both CCDs (Core Complex Dies), rather than just one.²² Previous generations (7950X3D) only had V-Cache on one of the two 8-core dies, requiring complex OS scheduling to ensure games ran on the correct cores. A dual-cache design yields a massive 192MB of L3 cache, potentially solving these scheduling headaches and creating a monster workstation/gaming hybrid chip.
• The 9850X3D: A frequency-bumped version of the highly successful 9800X3D, likely priced around $500, aimed at maximizing frame rates for competitive gaming where CPU bounds are most common.²¹

Display Technology: The Semantics of "Micro"

If silicon is the engine, the display is the windshield. CES is traditionally a TV show, and 2026 is no exception. However, a confusing nomenclature war is brewing between Samsung and LG, specifically regarding the term "Micro," which requires careful dissection for the informed buyer.

Micro RGB vs. Micro LED: A Necessary Distinction

Samsung and LG are both preparing to launch "Micro RGB" displays.⁹ It is vital for RADAR readers to distinguish this from Micro LED.

• Micro LED is a self-emissive technology where each pixel is a microscopic LED. It offers perfect blacks, infinite contrast, and high brightness, but remains prohibitively expensive (often exceeding $100,000 for large walls).
• Micro RGB (or RGB Mini-LED) is an evolution of Mini-LED technology. It still uses a backlight and an LCD layer. However, unlike standard Mini-LED which uses blue or white LEDs with a color conversion layer (Quantum Dots), Micro RGB uses an array of individual Red, Green, and Blue LEDs in the backlight.²³

Why it matters: Micro RGB promises significantly better color volume and spectral purity than standard Mini-LED without the astronomical manufacturing costs of true Micro LED. Samsung is expected to launch sizes ranging from 55 to 115 inches.²⁴ LG is countering with its own "Micro RGB Evo" panels featuring over 1,000 dimming zones. For the home theater enthusiast, this is the technology that will likely bridge the gap between the brightness of LCD and the contrast of OLED.

Transparent OLEDs: Moving to Production

LG remains the leader in transparency. The LG SIGNATURE OLED T, first teased previously, is expected to get a firm consumer release date.²⁵ This unit features a unique retractable contrast screen that rolls up to make the TV opaque for standard viewing (providing the necessary black levels for OLED contrast) and rolls down to make the panel transparent, blending into the room decor. While technologically impressive, it remains an architectural statement piece rather than a mass-market product, likely debuting with a price tag that reflects its novelty.

High-Refresh Monitors

For the PC gamer, the refresh rate wars continue. Samsung is rumored to showcase a monitor capable of 1,040 Hz.⁶ While arguably beyond the limits of human perception for most users, it pushes the requirements for DisplayPort 2.1 interfaces, reinforcing the need for the new GPU generation (RTX 50-series and Radeon 8000) which will support these ultra-high bandwidth standards natively.

Portable Computing: The Handheld Consolidation

The Windows/SteamOS handheld market has graduated from novelty to a legitimate product category. CES 2026 will showcase the "Second Generation" of these devices, addressing the primary complaint of the first wave: battery life and low-wattage performance.

The Chipsets: Ryzen Z2 Extreme vs. Lunar Lake

The battleground for handhelds is efficiency at the 15W TDP sweet spot.

• AMD Ryzen Z2 Extreme: Confirmed for early 2026 ²⁶, this chip features 8 cores (likely a mix of Zen 5 and Zen 5c) and 16 RDNA 3.5 GPU cores. It targets the 15W-35W power envelope, aiming to boost performance per watt significantly over the Z1 Extreme found in the ROG Ally. The inclusion of RDNA 3.5 graphics should offer a modest uplift in rasterization, but the main gain is expected in battery life management.
• Intel Core Ultra (Lunar Lake/Panther Lake): MSI is betting on Intel. The MSI Claw 8 AI+ uses the Core Ultra Series 2 (Lunar Lake).²⁷ Early metrics suggest Intel has finally caught up in integrated graphics driver stability, and the Xe2 architecture offers compelling performance at low wattages, potentially challenging AMD's stranglehold on this form factor.

The Devices: Second Gen Maturation

• Lenovo Legion Go 2: Rumors point to a split strategy. Lenovo is expected to debut a Legion Go 2 running Windows 11 for maximum compatibility, alongside a potential "Legion Go S" running SteamOS.²⁸ If confirmed, a major OEM shipping SteamOS hardware would be a watershed moment for Linux gaming, validating Valve’s OS as a viable third-party platform beyond the Steam Deck. The device is expected to retain the 8.8-inch screen size but move to a lighter chassis design.
• ASUS ROG Ally 2: Leaks from FCC filings suggest an iteration rather than a revolution—black and white colorways, dual USB-C ports (fixing the single port limitation of the original which hampered docking capabilities), and potentially 32GB of RAM as standard.²⁹ The RAM upgrade is crucial, as modern titles often starve the system when VRAM and system RAM are shared from a 16GB pool.
• MSI Claw 8 AI+: This device features an 8-inch screen and a significantly larger 80Wh battery ²⁷, directly addressing the poor battery life of the original Claw. This puts it in direct competition with the ROG Ally X in terms of endurance.

Laptops: The NPU Standard

Beyond handhelds, the laptop sector is fully embracing the "AI PC" designation. Razer is expected to update the Blade 16 and 18 with RTX 50-series GPUs.³⁰ Dell leaks suggest the XPS line will integrate Panther Lake processors in 2026.³¹ The overarching trend is the standardization of the NPU (Neural Processing Unit) across all tiers, enabling features like Microsoft Copilot to run locally, reducing latency and cloud subscription dependencies.

Automotive: The Software-Defined Pivot

The automotive hall (West Hall) at CES is no longer about horsepower; it is about FLOPs. The industry term remains "Software-Defined Vehicle" (SDV), but the reality is "AI-Defined Vehicle."

Sony Honda Mobility: Afeela 1

The joint venture between Sony and Honda will present the "pre-production" version of the Afeela 1.³² Deliveries are slated for California in 2026, so this is likely the final hardware revision we will see before launch.

• Integration: Expect deep integration of the Unreal Engine 5 for the HMI (Human Machine Interface). The dashboard is essentially a high-end gaming console, leveraging Sony's expertise in UI/UX to create a cabin experience that feels more like a PlayStation than a Honda Civic.
• Concept Model: A second concept vehicle, possibly an SUV form factor, is also expected to debut, signaling that Afeela is a brand, not just a single car product line.

Hyundai and Supernal: Ground and Air

Hyundai Motor Group continues to treat CES as its primary innovation showcase.

• Robotics Strategy: A "Group-level AI Robotics Strategy" will be unveiled ³³, emphasizing the "Software-Defined Factory" (SDF) concept where robots are adaptive co-workers.
• Boston Dynamics: The new, fully electric Atlas robot will make its stage debut.³³ Unlike the hydraulic version famous for parkour videos, this electric version is designed for commercial manufacturing applications—a true "cobot" (collaborative robot) that is quieter, stronger, and more precise.
• Supernal: Hyundai’s air mobility arm. While the S-A2 eVTOL debuted in 2024, 2026 will likely focus on the infrastructure ecosystem—vertiports and air traffic management software—rather than a new airframe. The industry is moving from "flying car" prototypes to the mundane but necessary work of certification and logistics.³⁴

Mercedes-Benz and the Virtual Assistant

Mercedes is pushing its MBUX Virtual Assistant, built on the MB.OS proprietary operating system.³⁵ The "Hyper-Personalization" features utilize generative AI to learn driver routines. For example, if you always call home at 5:30 PM, the car will preemptively suggest the call. Privacy advocates will likely have questions about where this training data resides—on the edge (in the car) or in the cloud—a critical distinction for security-conscious users.

Smart Home & Robotics: The Matter Settlement

The "Smart Home" is finally becoming the "Connected Home" thanks to the maturation of the Matter protocol. Version 1.4, released in late 2025, will be the standard for devices shown at CES 2026.

Matter 1.4: Infrastructure, Not Gadgets

The new standard adds support for heavy infrastructure: heat pumps, solar panels, and home battery storage.³⁶ This transforms the smart home from a collection of gimmicky colored lights to a cohesive energy management system.

• Nanoleaf: Known for decorative panels, Nanoleaf is launching Sense+ smart switches with a "Nala Learning Bridge".³⁷ This moves beyond simple on/off control to predictive automation based on user behavior, enabled by local AI processing.
• Yeelight: Is expected to debut new Matter-compatible bridges and lighting solutions ³⁸, further cementing the interoperability of the ecosystem.

Domestic Robotics: The Pivot to Utility

• Samsung Ballie: It has become a CES tradition to ask, "Where is Ballie?" The yellow tennis-ball robot, first shown in 2020 and refreshed in 2024, is rumored to finally get a price and release date.³⁹ It now features a built-in projector and onboard AI agent, aiming to be a mobile smart home hub rather than just a pet.
• LG CLOiD: LG is introducing a more anthropomorphic approach. CLOiD is a two-armed robot capable of fine manipulation tasks, like loading a dishwasher or picking up towels.⁴⁰ This represents a significant leap from simple "rolling tablet" robots, moving towards functional household labor.
• Roborock & Dreame: The battle for the floor continues. Roborock is teasing a robot capable of climbing stairs ⁴¹, a "holy grail" feature that would allow a single robot to clean multi-story homes. Dreame and Ecovacs (with the Deebot X11 OmniCyclone) are countering with maintenance-free base stations that plumb directly into home water lines and use advanced cyclone technology to reduce bag changes.⁴²

Startups and Innovation: The Emerging Fringe

Beyond the major halls, the startup ecosystem at CES often signals longer-term trends. The CES Innovation Awards provide a map of these emerging vectors.

Digital Health and Wellness

• BarunBio: This Korean startup is showcasing WE-STIM, a "battery-free energy-harvesting technology" embedded in clothing.⁴³ It converts body movement into micro-electric stimulation to aid muscle recovery. This represents a move away from rigid wearables (watches) to "smart textiles."
• Doublepoint: A startup focusing on gesture control, offering a hardware-agnostic platform that detects micro-gestures (pinches, taps) for controlling AR/VR and IoT devices.⁴⁴ This technology is becoming increasingly relevant as "spatial computing" (led by products like the Apple Vision Pro and Meta Quest) seeks input methods beyond bulky controllers.

The "Physical AI" Meta-Trend

If there is a single meta-narrative to CES 2026, it is the transition from Chatbots to Agents, or "Physical AI."

• Qualcomm is aggressively marketing the "AI Agent" concept—software that doesn't just answer questions but takes action (books the flight, orders the groceries).⁴⁵
• Algorized: A startup using ultra-wideband (UWB) radar for people sensing and vital sign monitoring without cameras ⁵, addressing the privacy concerns inherent in visual AI systems.

Conclusion

CES 2026 will likely be remembered not for a singular, world-changing gadget, but for the solidification of the AI ecosystem. The hardware is finally catching up to the software promises of the last three years. We have GPUs capable of running local models (RTX 50-series), NPUs in laptops ensuring efficiency (Panther Lake), and robots beginning to understand physical context (CLOiD, Atlas).

For the RADAR audience, the advice is clear: 2026 is a "Build Year." The components arriving in Q1 and Q2—whether it’s a Blackwell GPU, a Ryzen X3D CPU, or a Matter 1.4 energy system—represent a significant platform jump. It is time to update your Configurine wishlists, check your power supply headroom, and prepare for an era where intelligence is embedded in the silicon itself, not just rented from the cloud.

CES 2026: Watchlist Summary

Note: All specifications and release windows cited in this report are based on pre-show intelligence and are subject to change pending official keynotes.