Qualcomm Inc. unveiled its new AI stack that unifies the software approach to incorporate AI across its products and services. The company is on its path to transforming into a “full-stack” company that provides intelligent solutions for cloud-to-edge technologies. Cristiano Amon, Qualcomm’s Chief Executive, explains that the company follows several semiconductor firms with a similar vision to aid developers in accelerating AI integration.
According to the corporation, the Qualcomm AI Stack supports a wide range of intelligent devices with extensive AI software, access, and interoperability. It also supports a variety of AI frameworks, including open-source TensorFlow, PyTorch and ONNX.
The Qualcomm AI Stack portfolio gives developers direct access to the Qualcomm AI Engine, an AI library to delegate and deploy existing models directly to AI accelerators. The stack also provides specialized access to AI cores on Qualcomm Cloud AI 100. Further, it allows OEMs and developers to design one feature and then move the same model across other products and tiers.
Ziad Asghar, Qualcomm’s Vice President of Product Management, said, “We think with the Qualcomm AI stack, we are enabling developers and OEMs [Original Equipment Manufacturers] to be able to do so much more with the AI capability that we’re baking into our devices.”
The AI Stack portfolio also includes the AI Model Efficiency Toolkit, a specific AI development graphical user interface, model analyzers, and neural architecture search. The company announced its partnership with Google on Neural Architecture Search, to expand its neural network technology across the platform. Asghar added, “I think hardware is of course, critical, but increasingly AI software is absolutely critical. We’ve made very large investments to be able to make this AI stack, the best in class, AI stack for the intelligent edge.”