In five years, you wouldn’t want the cloud data to hit the brakes for a self-driving car.
As various industries move towards digitalization, our storage and computing increasingly rely on the cloud, which has become the hub of the information society. However, a collaborative effort from both industry and academia is attempting to create a new architecture called edge computing, outside of the cloud computing industry.
This new architecture will integrate networking, computing, storage, and applications closer to the objects or data sources, providing edge intelligence services nearby.
Do not underestimate the potential of this change. The industry is still unclear about how many application scenarios will utilize cloud computing in the evolution towards the Internet of Everything, and how many scenarios will distribute computing to the edge. With the potential complementarity and competitiveness of cloud computing, who will dominate under the new technological architecture and new play? Anything is possible.
Not All Computing Needs the Cloud
The wave of digitalization is sweeping through almost all traditional industries, changing our ways of communication, transportation, education, healthcare, and various traditional services… The underlying technology behind digitalization is cloud computing, where almost all data needs to connect to the cloud, and then be stored and computed in the cloud, connecting with each other through the network. However, cloud computing may not be the only path for various industries to move towards digitalization in the future.
In fact, as the PC internet and mobile internet evolve towards the Internet of Everything, how to transmit and process massive amounts of data has posed significant challenges to the current cloud computing technology architecture.
By 2020, it is expected that 50 billion devices will be connected to the network. According to the latest report from the ITU, the amount of data created per person per second will reach 1.7M in 2020. “In the face of such excessive data, we need to consider whether to adopt the traditional cloud computing approach or consider other methods?” said Xu Heyuan, Chief Engineer of the Technical Standards Research Institute of the China Academy of Information and Communications Technology.
“Our original idea was to send it (the Internet of Things) to the cloud, but compared to the current internet data nodes, the Internet of Things will see an increase of three orders of magnitude. Relying entirely on this (sending to the cloud) vertical connection is actually very difficult to achieve, so we need to consider processing at the edge level,” said Yu Haibin, Director of the Shenyang Institute of Automation, Chinese Academy of Sciences, who is engaged in industrial automation research.
Cloud computing unified the internet, but in the industrial field and OT (Operational Technology) field, there may be an urgent need for another computing system.
“Traditional cloud computing can respond to you after processing in the background with the click of a button. However, in edge computing, many applications need to respond in real-time within milliseconds. If we want to achieve real-time, we cannot send everything to the cloud because the cloud may be thousands of kilometers away, and latency and jitter are uncontrollable,” said Liu Shaowei, President of Huawei’s Network R&D Department.
However, edge computing is not entirely opposed to cloud computing; some experts believe that in certain application scenarios, analysis can be done at the edge while also conducting in-depth analysis in the cloud, allowing both to showcase their strengths. After all, the computing power of the cloud is still unmatched by the edge.
“I recently went to Boston and visited the site of a previous explosion. There, they performed intelligent analysis at the front end and then transmitted effective data after simple analysis of the suspect to the back end. This is a great example of how to use cloud and edge collaboration complementarily,” said Fang Fahe, Executive Vice President of Softcom Power.
Tech Giants Collaborate to Open New Market Spaces
“What will the future edge look like? Sometimes it is hard to imagine. But at least according to our judgment, the demand in the manufacturing industry has reached a stage where product lifecycles are getting shorter, personalized demands are increasing, and the trend of full lifecycle management and service-oriented approaches is becoming more apparent,” Yu Haibin said.
It is precisely the pull of the industry’s digitalization demands that drives the acceleration of fundamental technological changes. “We always say that being one step ahead is advanced, but being three steps ahead makes you a martyr,” Liu Shaowei said.
Some large companies have already begun their attempts in edge computing within the Internet of Things field. For example, Huawei has entered industries such as electricity and elevators, while Intel has also entered the industrial and IoT fields through embedded computing.
“There are over 15 million elevators globally, but a portion of them do not operate normally every day. By utilizing the Internet of Things, their efficiency can be improved by over 50%, while significantly reducing the failure rate, bringing actual economic and social benefits,” Liu Shaowei said. He also cited the example of Nigeria’s electricity sector, stating that through digital transformation, it has helped this African country improve its electricity production efficiency by 30%.
Chip companies like Intel and ARM also hope to expand their business in the IoT field through edge computing. “The next step is whether IT and OT can be integrated, which is an exciting opportunity,” said Chen Wei, General Manager of Intel’s IoT Division in China.
In the future, the entire edge computing industry could be a market worth trillions, but no one can accurately estimate its market capacity because the market is too large, and the industries involved and the enterprises needing digitalization are too many.
“The electricity IoT market didn’t seem very large before, especially in the communication sector. But we specifically conducted research and found that the communication network in the electricity IoT field has nearly $5 billion in space. This is just the network market in the electricity segment, and if we extend it, there are four segments in the electricity industry: generation, transmission, distribution, and consumption,” Liu Shaowei said in an interview.
“We hope not only to have new points but also to turn points into lines, lines into surfaces, and surfaces into volumes,” Yu Haibin said in an interview. He is unsure how much scale expansion can be achieved for existing chips, switches, and computing output, but he noted, “For transformative technology, linear extrapolation may not be valid. We often mention ‘exponential growth,’ but after the reserve period of ‘exponential growth,’ it becomes difficult to predict growth accurately; we can only say it will grow rapidly.”
Liu Shaowei stated that the potential huge market in the future cannot be monopolized by one company or a few companies because the digitalization of numerous industries requires deep participation from many traditional fields. “It will not be like the past where one company could do everything, especially in the context of industrial digitalization. It involves different fields and requires support from information technology and communication technology. In fact, everyone needs to collaborate and cooperate to innovate solutions.”
At the same time, large technology companies also hope to accelerate market maturity through alliances and find more suitable positions for themselves. Recently, companies like Huawei, Intel, and ARM have formed such an alliance. According to some technical experts, the various technological attempts made by many companies independently will be further standardized through the alliance, thus accelerating the maturity of the industry.
For the future, there is a consensus that although edge computing is named edge, its technological influence and potential market scale may not be marginal at all.