Privacy, security, cost, speed and resilience
“Making forecasts is a difficult task, especially concerning the future,” used to say Pierre Dac, a French humorist from the past century.
Every year, Deloitte comes up with its TMT (Technology, Media & Telecommunications) predictions, presented by Duncan Stewart, at the beginning of the year. The presentation generally starts with a look in the rear view mirror to demonstrate past previsions came up true.
This year, Duncan Stewart chose to focus on Private 5G networks and Edge AI, considering they have a lot in common.
Private 5G networks can be seen as local area networks that use 5G technology to build and create a network. They allow large corporations and other organizations to own their communication network and use it for all their internal communications, while keeping the advantages of 5G, especially high speed, high throughput and low latency. As they are privately owned, private 5G networks bring privacy and security as communications do not travel over public networks. As they are privately owned, private 5G networks are expected to be less costly than the ones deployed by mobile network operators, while providing additional features. And having in mind the development of IoT and all kinds of remote operations, the high speed and low latency brought by 5G are paramount for all applications.
Although these 5G networks are private, they still need security, and especially a strong user authentication making them an opportunity for the secure transactions industry. Deloitte anticipates the main markets for private 5G networks to be in manufacturing, transport, construction, utilities and mining; in these markets, more than 100 companies worldwide will begin testing private 5G deployments by the end of 2020. Deloitte anticipates the global economic output enabled by private 5G networks to be over US$ 5 trillion (EUR 4.6 trillion) in 2035.
At the same time, Artificial Intelligence (AI), or machine learning as defined by Duncan Stewart, is becoming mission-critical for an increasing scope of activities. Until recently, AI was essentially accessed through the cloud as it was requiring huge computing power capabilities. On the other hand, computing power is dramatically increasing: any smartphone we have in hand is more powerful than the computers that succeeded in landing a man on the moon about 50 years ago. Nowadays, AI is present on smaller devices: in many smartphones, a part of the core processor such as the Samsung Exynos 9820, the Huawei Kirin 970 or the Apple A13 Bionic, is dedicated to AI; it is called the NPU or Neural Processing Unit, also called AI accelerator, optimized to execute machine learning algorithms. Most interesting is the direct cost of this NPU: between US$ 1.10 and 5.10 (EUR 1.00 to 4.65). Now, Syntiant, a California-based silicon designer, announced at CES a US$ 0.05 (EUR 0.045) NPU that consumes microwatts. In smartphones, these AI units are mainly used for biometrics, speech recognition and for photo improvement. Having an NPU aboard an edge device brings privacy, security, cost, speed and resilience advantages; in other words, it allows to make dumb devices smarter.
The NPUs are facing the same security needs as anything aboard a smartphone or more generally an edge device: they need to be properly secured in order not to lower the global security of the system. This represents a strong opportunity for the secure transactions industry to deploy its security expertise. According to Deloitte, edge AI chips are to be present in 750 million devices in 2020 and 1,600 million devices in 2024, representing at least a US$ 1.6 billion (EUR 1.45 billion) market, that may even be up to US$ 10 billion (EUR 9.1 billion).
To provide privacy, security, cost, speed and resilience, all technologies require security experts, creating new openings for the secure transactions industry.