More than 300 million smartphones, or a fifth of units sold, will have neural network and machine-learning functions in 2017, according to Deloitte.
This artificial intelligence will be used to for everything from indoor navigation to image classification to augmented reality and speech recognition.
In addition, Deloitte said that these tools won't necessarily need to be connected: "On-board machine learning may allow tasks to be done better and faster, or with more privacy," said Deloitte in its Global Predictions 2017 report.
Deloitte's report covered a few bases, but the one revolving around machine learning is among the most certain. For instance, Huawei announced a smartphone with Amazon's Alexa. Android has machine learning built in. And digital assistants abound.
Some devices were capable of machine learning in 2016, but most of the computing was handled in the cloud. What will change in 2017 is that faster processors will enable machine learning natively. Deloitte said smartphones will be just the first wave of machine learning on the go.
"These capabilities are likely to be found over time in tens of millions of drones, tablets, cars, virtual and augmented reality devices, medical tools Internet of things devices and unforeseen new technologies."
Other predictions from Deloitte include:
- IT as a service, which rolls up a variety of cloud services, will nearly hit $550 billion as a market by end of 2018.
- The active base of fingerprint reader devices will top 1 billion in early 2017.
- Distributed Denial of service (DDoS) attacks will scale to terabits and IoT devices will be more vulnerable than ever.
- Autonomous driving will start with braking and automatic emergency braking will save 6,000 lives by 2022.
- 5G building blocks are being deployed and will build over time and provide more connectivity disruption than 4G did.
- By 2022, indoor spaces will have digital navigation similar to GPS.
- Flat is the new up for TV advertising.
- Tablet sales will continue to fall and hit 160 million units in 2017. "There is no dominant compelling use case for these devices," said Deloitte.