AI might be moving to the edge, instead of the cloud

27.08.2020 1,777 11

Martin Deshev

The edge is starting to give the cloud a run for its money in yet another segment. According to a forecast by ABI Research, the edge is set to become a prime market for AI chipsets, DataCenterNews reports.

Currently, the cloud is the main home for artificial intelligence. It’s where most AI projects are being trained. They also gather and store data and more. AI algorithms already work with a lot of services making user experience better. Now AI is preparing for the next big step in its development – the edge.

Why the edge?

It’s a valid question. Especially since, we’ve grown used to the fact that AI consumes a lot of everything. As such, it doesn’t’ really seem suited for the edge where quickness and lightness are key. Strangely, that’s also the reason why AI is looking towards the Edge.

The industry has been changing. Cybersecurity, privacy and flexibility are top priorities along with low latency. AI needs to cover all of these bases. It also needs to become more adaptable and “light”. It has to be able to work with smaller devices and sensors and thus it has to be closer to them.

“As enterprises start to look for AI solutions in the areas of image and object recognition, autonomous material handling, predictive maintenance, and human-machine interface for end devices, they need to resolve concerns around data privacy, power efficiency, low latency, and strong on-device computing performance,” says ABI Research principal analyst Lian Jye Su.

“Edge AI will be the answer to this. By integrating an AI chipset designed to perform high-speed inference and quantised federated learning or collaborative learning models, edge AI brings task automation and augmentation to device and sensor levels across various sectors. So much that it will grow and surpass the cloud AI chipset market in 2025.”

ABI Research says that by 2025 the cloud AI chipset market will reach $11.9 billion. Meanwhile, the edge AI chipset market will grow to $12.2 billion in revenue for the same period.

So who will benefit?

In theory almost everyone. Chip manufacturers will have yet another market opportunity. They will have to help with the implementation in various industries, manufacturing, retail and more. All of these will benefit from AI services and the edge will be of prime help. This will be good for consumers and clients, too.

Even data center providers will be on board. AI moving to the edge doesn’t mean it will leave data centers. Quite the opposite. It will need even more power and capabilities to help crunch all that extra data and analysis gathered by the edge. As such they will experience a strong demand for chips, too.

COVID-19 slowed down this trend, but not by much. It’s expected that it will fully rebound by 2022. New and emerging technologies like Wi-Fi 6, autonomous vehicles and so on will also play a vital role in the increased market demand going forward.