As TinyML vendors continue to democratize machine learning (ML) at a rapid pace, global technology intelligence firm ABI Research forecasts that TinyML Software-as-a-Service (SaaS) revenue will reach $220 million in 2022 exceed and will become an important component from 2025 onwards.
While overall revenue will be dominated by chipset sales as TinyML device shipments continue to grow, the TinyML market for SaaS and professional services has the potential to become a billion-dollar market by 2030, adds ABI Research.
ABI Research is a global technology intelligence firm providing actionable research and strategic guidance to technology leaders, innovators and decision makers worldwide. The above results come from ABI Research’s TinyML: A Market Update application analysis report, which is part of the company’s AI and ML research service.
The TinyML market has come a long way since ABI Research first analyzed this market in 2020.
The TinyML Foundation, which gathers most of the prominent providers in this space, has grown significantly in recent years.
There was a similar expansion in TinyML’s applications; with wildfire detection, shape detection and seizure detection to some of the most spectacular use cases.
Furthermore, given the central importance of environmental sensors to TinyML, the possibilities are extensive.
Nonetheless, environmental detection and audio processing remain the most common uses in TinyML, with sound architectures holding a nearly 50% market share in 2022. Most of these applications use either a microcontroller (MCU) or an application specific integrated circuit (ASIC).
The personal and work equipment sector will see the biggest increase in the near future.
“An ML model can probably be applied to any sensory data from an environment. Some of the most common applications are word recognition, object recognition, object counting, and audio or speech recognition,” explains David Lobina, artificial intelligence and machine learning research analyst at ABI Research.
With so many options, there are potential pitfalls, but ABI Research believes there are well-identified solutions.
“The physical limitations of TinyML devices are real. These devices prefer small and compact ML models that require innovations at the level of software solutions for specific use cases. And software vendors will be the most active in the TinyML market,” says Lobina. Software providers in this space include Edge Impulse, SensiML, Neuton, Nota and Deeplite.
Additionally, given the variety of use cases, vendors need to focus on those applications for which TinyML has worked out a clear value proposition prior to production.
“The role of software is crucial, and vendors need to develop software tools to automate TinyML themselves. Eventually, new technologies will be required to produce ever more sophisticated TinyML models. Neuromorphic computing and chips along with the corresponding spiking neural network technology bode well for the future,” adds Lobina.