Qualcomm has introduced its new Snapdragon Reality Elite platform at the Augmented World Expo, marking a major step forward for mixed reality, virtual reality and AI-powered wearable devices. The announcement is not just about a new chip. It also signals a broader change in how Qualcomm wants to position its extended reality portfolio for the next generation of headsets, glasses and connected AI devices.

With Snapdragon Reality Elite, Qualcomm is moving away from the older XR1, XR2 and XR2 Gen 2 naming system. The company is now bringing its immersive technology products under the “Reality” branding, with Reality Elite placed at the top of the lineup. In simple terms, this is Qualcomm’s most powerful platform for devices that blend the physical and digital worlds.

The new platform has been designed for both traditional all-in-one headsets and a newer category of lighter tethered devices. In an all-in-one headset, the display, processor, battery and sensors are built into the headset itself. In a tethered setup, some of the heavier computing and battery components can be moved into a separate puck that is worn on the body. This allows the headset or glasses to become lighter and more comfortable, which has been one of the biggest challenges for mixed reality hardware.

Snapdragon Reality Elite performance

Performance is one of the biggest talking points around Snapdragon Reality Elite. Qualcomm claims the platform delivers up to 60 percent higher GPU performance, 30 percent higher CPU performance and 160 percent better NPU performance compared to the XR2 Plus Gen 2. The improved Hexagon processor is rated at 48 TOPS, giving the platform much more room to handle AI and perception workloads directly on the device.

For users, this means future headsets and glasses could offer sharper visuals, smoother performance and more intelligent real-time experiences. The platform supports visuals up to 4.4K per eye at 90 frames per second, while also handling video see-through graphics and spatial awareness tasks. This is important because mixed reality devices need to constantly understand the user’s surroundings while rendering digital content on top of the real world.

Until now, much of the processing power in XR devices was already being used for core features such as head tracking, hand tracking, depth mapping and environmental understanding. These tasks run continuously in the background and leave limited space for heavier AI features. With Reality Elite, Qualcomm says developers will have enough headroom to run advanced AI models directly on the device.

On-device AI gets a major push

At AWE, Qualcomm demonstrated this by showing an offline device running a photorealistic 3D avatar powered by a large language model-based chat agent. The company also showed stable diffusion image generation completed in around 1.7 seconds without using a cloud connection. This is a significant development because on-device AI can reduce latency, improve privacy and make devices more useful even when internet connectivity is limited.

Another major focus area is perception. Qualcomm has upgraded its EVA, or Engine for Visual Analytics, block. This dedicated technology helps with 3D reconstruction, depth estimation and tracking. The updated version includes new accelerators that are meant to improve accuracy and reduce latency. In mixed reality, even small delays can break the illusion of digital objects existing naturally in the real world. Faster and more accurate perception can make interactions feel more stable and believable.

Better video see-through and visual quality

Qualcomm also says video see-through latency has improved by around 10 percent compared to the previous generation. Video see-through is the technology used in devices where cameras capture the outside world and display it on screens inside the headset, allowing digital elements to be placed into the user’s view. Better latency and image processing can make the experience feel less artificial and more comfortable over longer sessions.

The platform also brings improvements in visual quality through denoising and foveated processing. Foveated processing focuses high-quality rendering on the area where the user is looking, while reducing detail in the peripheral view. This helps save processing power without making the experience feel noticeably worse. Such features are important for creating high-quality visuals while keeping heat and battery consumption under control.

Improved battery life and cooler performance

Despite the large jump in performance, Qualcomm says Snapdragon Reality Elite can run up to 20 percent longer on battery at similar workloads. It also runs up to 12 degrees Celsius cooler under load compared to the previous generation. These gains could help device makers build slimmer and lighter products, since heat management and battery size are two of the biggest factors affecting headset design.

The support for tethered or disaggregated designs could be one of the most important parts of the platform. Moving compute and battery into a separate puck allows manufacturers to design glasses that feel closer to regular eyewear. However, this is technically complex because the headset and puck must communicate instantly while handling graphics, sensor data, tracking and AI processing. Qualcomm says Reality Elite has been built to manage this full data pipeline between the two components.

First Snapdragon Reality Elite devices

The first device based on Snapdragon Reality Elite will be XREAL’s Project Aura. It is a pair of optical see-through glasses connected to a tethered compute puck running Android XR. The product has been developed through a collaboration between Qualcomm, Google and XREAL. Project Aura was shown at AWE, but XREAL is expected to share more details about pricing and availability separately.

Another device using Reality Elite is also being developed by Chinese OEM Play4Dream. More details about this product are expected to be revealed around AWE. Together, these early devices show that Qualcomm is trying to push the market beyond bulky headsets and towards lighter, more wearable mixed reality products.

Snapdragon START for AI glasses

Alongside Reality Elite, Qualcomm also announced Snapdragon START, which stands for Scalable Turnkey AI Ready Toolkit. This programme is aimed at brands that want to enter the AI glasses market without building every part of the product from scratch. It includes a turnkey module based on the AR1 Plus platform, companion apps for iOS and Android, an AI cloud solution and white-label glasses designs.

These designs can range from simple camera-and-speaker glasses to models with monocular or binocular displays. This could help more brands bring AI glasses to market faster, especially companies that do not have deep experience in hardware manufacturing or supply chain management.

Second-generation smart ring reference platform

Qualcomm also revealed a second-generation smart ring reference platform, developed with KeyWear and based on the Snapdragon S7 Plus. The ring now adds micropower Wi-Fi along with Bluetooth Low Energy, allowing it to control multiple devices across a personal AI ecosystem. This could include AI glasses, smart TVs, tablets and even connected cars.

A microphone has also been added to the ring, allowing users to whisper commands instead of speaking out loud. This hints at Qualcomm’s broader vision where glasses, rings and other wearables work together as part of one connected AI system.

What it means for the future of mixed reality

Snapdragon Reality Elite is clearly more than a routine chip upgrade. It is Qualcomm’s attempt to create the foundation for a new wave of mixed reality and AI wearable devices. With stronger graphics, faster on-device AI, improved perception, better battery life and support for lighter tethered designs, the platform could play a major role in shaping how future headsets and smart glasses are built.