GenAI has become a top priority for companies and Qualcomm is pushing the envelope by developing solutions for on-device AI PCs. At Collision Toronto 2024, Emily Herbert, Research Analyst at Counterpoint Research, had a quick chat with Qualcomm’s Durga Malladi, SVP and GM of technology planning and edge solutions. He spoke about the new Snapdragon X Elite SoC line that allows on-device GenAI features to run directly on PCs with companies like Lenovo and HP that have already launched AI PCs. Some applications include live transcription, new studio effects using the web camera and Windows Studio Effects, and ‘paint’ to bring drawn images to life.
The interview:
Key takeaways from the discussion:
How Qualcomm defines AI PCs:
• According to Qualcomm, AI applications are critical to define AI PCs as they provide a personalized experience for the user with the privacy and security of keeping data stored in the device itself.
Benefits of on-device AI:
• Privacy and security are major benefits for running on-device AI as data stays on the device as opposed to cloud processing.
• With on-device AI, users can access these features without an internet connection, including scenarios like on transatlantic flights or in rural areas where the device may be used in the enterprise space with unreliable connectivity.
Qualcomm’s efforts in helping developers push on-device GenAI:
• Qualcomm AI Hub is a great tool to help developers access pre-trained models, write their applications, and test them remotely on real devices.
• Developers can now take the use cases to their desecration.
• People who did not have access to this technology across various fields like medicine, logistics, agriculture, and various other sectors can now use AI in ways that tech developers may not have been able to think of as the technology becomes democratized into the wider enterprise and consumer space.
Analyst takeaways:
• On-device GenAI has seen an excellent push within the enterprise space and now we are seeing real-life applications for the average consumer. The initial integration is expected to be slow as people learn about this technology and begin integrating it into day-to-day life. Increased knowledge and advertising in the market can help with wider consumer penetration.
• Now is a good time for on-device GenAI as hardware is reaching a point where it can support these higher processing applications and still reach the consumer price band, granted it is still within that ‘ultra-premium’ price point.
• The ultimate immersion of these AI technologies across all consumer segments will become easier and more cost-effective when the devices can be manufactured for and sold to more average users.
