Qualcomm’s New Venture Fund to Invest $100M in On-Device AI Startups
Last week, Qualcomm announced the Qualcomm Ventures AI Fund, a $100 million fund that will invest in startups working on artificial intelligence technologies. The fund has already made its first investment, in the Series A round of AnyVision. A face, body, and object recognition startup, AnyVision also uses on-device AI. The amount invested was not disclosed, but Qualcomm Ventures did say investments would range from $1 million to $10 million per deal and the total AnyVision funding round was $28 million according to PitchBook. Robert Bosch led the round and Elysian Park Ventures also participated.
The Qualcomm Ventures AI Fund is specifically looking for companies finding ways to do intensive data processing on devices like smartphones and security cameras instead of in remote cloud systems; Qualcomm wants on-device intelligence to become pervasive. Quinn Li, head of Qualcomm Ventures commented,
If you look at the smartphone today, this is one of the biggest AI platforms. We want to make edge AI a ubiquitous platform.
An On-Device AI Focus
Photo Credit: Qualcomm
On-device AI runs on the device, like a smartphone or vehicle, rather than the cloud. It doesn’t require cloud connectivity to operate and add value but could supplement its features through cloud access. Autonomous cars, robotics, computer vision, and IoT are all target applications of on-device AI where Qualcomm Ventures plans to focus its investments.
The AI powering key voice UI components such as automatic speech recognition and natural language processing, has traditionally run in the cloud due to computing, storage, and power constraints (this is how Alexa’s AI works). However, on-device processing provides unique benefits, such as a much faster response, increased reliability, and better privacy protection says Qualcomm. In addition, having AI locally not only saves power on devices but also provides an increase in network bandwidth. Not having to regularly send data back and forth to the cloud is incredibly beneficial.
Managing Power and Memory Constraints
According to Qualcomm, the key challenge in implementing on-device AI is to run the complex voice UI functionality within the power and thermal envelope of a mobile device. Sensory and Snips.ai are examples of software companies that focus today on local AI solutions with low power and memory requirements. In May, Sensory launched a new low power wake word solution that cut power requirements by 80%. However, Qualcomm believes new innovations are needed to push on-device AI capabilities even further. Albert Wang, Qualcomm investment director, says on-device AI is the future, telling TechCrunch:
Today’s AI processing is very computationally intensive. When you’re talking to Alexa, nothing is processed on your device, it gets taken to the cloud and gets scrunched there. There are a few problems with that — performance deteriorates, it consumes a lot of bandwidth and there are privacy issues. Imagine you have an Alexa that is more private and user-friendly, you ask the questions and can get the answers instantly. It doesn’t take the round trip all the way to the cloud.
Qualcomm will be able to gain insight into the future of AI and what’s needed from its processors with this fund. Low power processing and connectivity are essential for artificial intelligence. New AI chips that can rapidly process information and return results to users are a big focus of many in the semiconductor industry, including Qualcomm. What this means in regards to the voice industry is that we could be seeing better performing voice assistants in the future and on many different devices ranging from appliances and IoT to automobiles.
Qualcomm is only the latest company to direct new investment to artificial intelligence. Google has long focused on AI and actually does incorporate some on-device AI in the Pixel 3. Samsung announced in June that Samsung NEXT had launched a fund to invest in early-stage startups devoted to “solving AI problems, as well as those using AI to solve computer science problems.” Intel Corp, Micron Technology Inc, and Nvidia Corp have all invested in AI as well. In October, Micron said it was planning on investing up to $100 million in startups focusing on artificial intelligence, and in recent years Intel’s venture capital arm has funneled more than $1 billion into AI.
