Qualcomm’s Snapdragon 8 Elite Chips Bring Powerful Local AI to Phones

Qualcomm unveiled two new flagship smartphone processors with a pronounced focus on running artificial intelligence models directly on the device. The Snapdragon 8 Elite Gen 6 and its higher-end variant, the Snapdragon 8 Elite Extreme Gen 6, were announced at Qualcomm’s annual Snapdragon Summit. The company emphasized improvements across AI inference, camera processing, video recording, and audio handling, positioning the new chips as the foundation for AI-first mobile experiences.

Pushing Local AI Model Capacity

The headline technical achievement is raw processing power for on-device AI. The Extreme variant can execute a 30-billion-parameter mixture-of-experts model on the smartphone chip itself. This represents a significant expansion of on-device AI capability compared to prior generations. In a mixture-of-experts architecture, while the overall model size is large, only a fraction of the total parameters activate for any given task. This design allows the processor to handle complex reasoning without activating every neuron simultaneously, reducing computational demand and power consumption while maintaining quality.

Comparing to Apple’s Approach

For context, Apple announced a 20-billion-parameter mixture-of-experts model in June at its Worldwide Developer Conference as part of its third generation of foundation models. Qualcomm’s 30-billion-parameter capability represents a 50 percent increase over Apple’s public specification. The significance lies in what it enables: more sophisticated language understanding, reasoning, and personalization on phones without sending data to external servers for processing.

Running larger models locally has privacy implications. When inference happens entirely on the device, conversation histories, personal preferences, and sensitive information never transmit across the internet. This addresses growing concerns about data collection and provides users with the option to interact with AI assistants without cloud uploads.

The Standard Gen 6 Architecture

The base Snapdragon 8 Elite Gen 6 includes a new accelerator element specifically designed to execute AI models with greater efficiency. For smartphone hardware, efficiency directly translates to practical benefits: longer battery life, less heat generation, faster task completion, and sustained performance during extended usage. The accelerator improves inference speed across a range of model sizes, not just the extreme 30-billion-parameter models, making AI features practical across different types of applications.

On-Device AI Assistants

Read also: AI’s Trust Problem: What Enterprises Need to Know at Disrupt 2026 · AI Safety Plan Gains Support, but Skeptics Question If Slowdown Is Real · AI Industry Promises Measured Pace Amid Pressure to Accelerate · Greece’s PM Pitches Tech Investment, Admits AI’s Unknowns

Both chips include a dedicated sensing hub capable of running neural network models up to 200 million parameters. This design creates a distinction between compute-heavy tasks that route to the main processor and lighter inference that stays on the sensing hub, preserving overall system efficiency and power consumption. The separation of concerns allows manufacturers to optimize power delivery and thermal management across different workloads.

The sensing hub enables a personal scribe feature that operates entirely on the device. This means a smartphone can record and transcribe conversations, meetings, or voice memos without uploading audio to the cloud. The chip can differentiate between speakers in a conversation, attributing statements to the correct person. Beyond transcription, Qualcomm stated the sensing hub can build memory based on user behavior and usage patterns, then leverage that memory to suggest automation opportunities for routine tasks. These suggestions could range from creating calendar events based on meeting discussions to drafting follow-up emails with appropriate context.

Qualcomm also confirmed the chips support a complete voice-input-to-voice-output AI agent, creating a conversational interface that doesn’t require text input and doesn’t require an internet connection for the exchange to function. This architecture enables always-available digital assistants that remain responsive even during network outages or connectivity issues.

Video and Camera Enhancements

The new processors offer pixel-level control over camera sensors, a capability previously found only in professional video equipment or high-end computational photography systems. This granular control enables smoother stabilization, more accurate tracking of motion, and new post-processing possibilities without external software or additional processing hardware.

The Extreme variant supports video recording at 8K resolution at 60 frames per second, and at 4K resolution at 240 frames per second for ultra-slow-motion sequences. To put these specifications in context, 8K60 recording was once limited to expensive cinema cameras, and 4K240 enables compelling slow-motion effects. Qualcomm introduced a new Advanced Professional Video codec called APV, which enables encoding that meets professional production standards. Professional videographers have traditionally relied on external hardware encoders; the inclusion of APV codec support on the smartphone processor narrows the gap between mobile and dedicated video production equipment.

Audio Processing

Audio capabilities received similar attention. The chips employ AI to enhance vocals and suppress background noise, relevant both for content creation and call quality. A new feature called voice bubble technology isolates the user’s voice during phone calls, preventing ambient noise from reaching the other party. This targets a long-standing mobile phone problem: background noise transmission during conversations in noisy environments.

Device Availability and Partner Adoption

Motorola announced the Motorola Signature 27 smartphone, powered by the Snapdragon 8 Elite Extreme Gen 6. The device will become available sometime in the current year, making it one of the first flagships to ship with the new processor. The Motorola partnership signals that established phone manufacturers view the new capabilities as essential for competitive positioning.

Beyond Motorola, Qualcomm stated it is working on more than 40 AI-focused devices using the new chips. The breadth suggests enthusiasm among manufacturers to differentiate on AI capabilities, signaling that the industry views on-device AI as a key selling point for the coming generation of phones and mobile devices. This level of adoption indicates that AI processing power has become a primary specification category, similar to how processor speed and RAM capacity were marketed in prior decades.

Smartphones as the Primary AI Platform

A recurring theme in the industry is whether consumers will adopt dedicated AI devices or use existing smartphones as their primary AI interface. Qualcomm’s announcement reinforces the smartphone-first perspective. Nothing co-founder Carl Pei has previously hinted that phones would be the center of gravity for consumer AI adoption. More recently, John Ternus, Apple’s newly appointed CEO, articulated the same view during the iPhone Duo launch earlier this month, stating that smartphones will remain the primary device for AI experiences.

The convergence of opinions across multiple companies suggests an industry consensus that the smartphone, not a dedicated AI appliance, will be where most consumers interact with artificial intelligence in the coming years. This consensus shapes product roadmaps and capital allocation across the industry.