Qualcomm Acquired Scyfer to Push AI-on-a-Chip
Qualcomm acquired Scyfer, the Amsterdam-based machine learning startup, to further its AI-on-a-chip ambitions. Most of the artificial intelligence (AI) solutions you hear about are cloud-based. Whether its NLP-based voice assistants like Amazon Alexa or big-data virtual advisors like IBM Watson, the compute power is all in the cloud. There are a few outliers that want to keep the processing on the local device. Mycroft is an example of software solution that focuses on local NLP for its voice assistant in the name of privacy. A Qualcomm statement focuses on other reasons.
Many companies focus on the execution of AI workloads in the cloud, but Qualcomm Technologies is focused on the implementation of AI on end devices – smartphones, cars, robotics, and the like – to ensure that processing can be done with or without a network or Wi-Fi connection.
The key distinctions are about independence, control and flexibility. If you run your machine learning or other AI functions locally, then you are not dependent on your cloud hosting or platform provider for your workload. You also have more control over the data and flexibility to operate in environments that are not hospitable to always-on network connectivity.
Qualcomm is Simply Implementing its Broader Strategy
It is worthwhile to note that Qualcomm doesn’t see the on-chip AI as replacing cloud-based AI. Instead, the company focuses on the concept of complementing cloud resources. Qualcomm acquired Scyfer to push that strategy forward and presumably add to its local processing for machine learning tasks.
You can see from the Qualcomm timeline that it sees Scyfer is another step on its decade-long quest to implementing AI-on-a-chip. When people talk about AI leaders, Qualcomm is not often in the discussion, but it should be. There are many instances where chip-level integration is either critical or helpful in executing voice and AI-related tasks. Sensory’s Todd Mozer brought this up in a Voicebot interview earlier this year. It’s not always about the cloud. Sometimes the chip will be the real star of the AI show.