09Mastered edge computing
Sensor-powered devices provide secured connectivity for the world (privacy, ai everywhere offline, decentralization)
Edge computing is bringing AI and other algorithms literally to the ‘edge’ of the high bandwidth world. It enables service providers to connect with their customers in-real time regardless of connectivity and physical location. As hardware continues to become cheaper and more powerful, edge computing has the potential to drastically improve the capabilities of our electronic infrastructure and tools.
What will change
As hardware becomes more capable and software becomes faster (yet more lightweight), artificial intelligence will move from the cloud into localised microprocessors capable of handling sophisticated algorithms. This will lead to even smarter vehicles and products, improved data security and zero latency decision making.
Edge AI will give devices the capability to make sophisticated, independent decisions in real time. This will enable smarter energy usage at lower cost with less data transmission. Local AI-powered decision making will also minimise latency and security issues, building better customer experiences. Cloud inventory will store transmitted data in one easy-accessible location, giving birth to new services that take advantage of both technologies.
Opportunities worth taking
Helping modern, widely-scaled IoT services move away from pure cloud connectivity will unburden centralised services and databases that struggle when devices require good connectivity to run as a raw pipeline to the cloud. Edge computing can also make devices more secure, particularly when handling sensitive data such as biometrics.
Manufacturing of goods
Modern computer vision applications will revolutionise predictive maintenance and quality control of assembly lines and big machinery.
Edge computing can be utilised in stores and warehouses to monitor products on shelves or analyse shopping behaviour, for example with indoor location data. In addition, edge computing can be leveraged for biometric identification of customers or employees.
The number of sensors and data points in Industrial IoT systems can be huge. These systems still require real-time control, which means cloud - and even wired - latency can be too high. Locally performed edge computing will be faster and more inexpensive than using cloud infrastructure.
Modern vehicles use increasing amounts of software to anticipate both surroundings and user actions. In most cases, it is crucial that latency stays low, making cloud services redundant. Also, the amount of data gathered by smart vehicles can be vast, so locally-performed analysis will be beneficial.
Edge computing can reduce the amount of data sent and received from customer electronics to remote services. It means more secure and quick devices like mobile phones. New kinds of algorithms such as computer vision can also be taken into use on devices previously not powerful enough to handle computation. This may lead to smarter cameras and home appliances, as well as devices like home assistant robots.
Smart home accessories can be helpful for residents, but they can also introduce irritation if they require constant connection to the cloud. Edge computing means appliances can work quickly and reliably - even on poor networks.
Deep-dive into this bet
Total market 2024
Edge computing $ 13,9 bn with 32% CAGR, consumer electronics $ 1,3 trillion.
Look at these 2024
Methodology: A high-level view gathered from various openly available data sources on market size & compound annual growth rates. To simplify comparison between analysts sources, we’ve removed differences in market details by calculating an estimated average for total market size 2024 and average CAGR for selected time period. This allows us to gain “correct enough estimates”. Unless linked directly to one source, each of our market estimate has included 2-8 sources e.g. WEF, CB Insights, Mordor Intelligence, Markets & Markets, Research & Markets, Grand View Research, PWC, Allied Market Research, McKinsey, Market Watch
Biometrics systems & tech $ 51,1 bn with 15% CAGR, Smart homes $ 180 bn with 18% CAGR, consumer robotics $ 19,2 bn with 20% CAGR, smart agriculture $ 20,6 bn with 15% CAGR, electric vehicles $ 667bn and 19% CAGR.
Top 18 unicorn sniffing smart money VCs, 1376 investments 2018-2020
Source: Crunchbase Pro, accessed: 03/20, research of company description terms between 01/2018-02/2020, investment types: early stage investments by top 18 unicorn sniffing smart money VCs, n=1376 investments
(1376 investments 2018-2020): edge computing 2%, Cloud 6%, security 8%, links to decentralised computing and might become an appealing pivot from blockchain if supported by use cases 3%.
Edge computing is a lucrative approach to ensure security & privacy issues, as well as operating offline in areas like automotive, robotic, maritime, offshore, logistics, and healthcare. Occasional connection to cloud in architecture design allows effective scaling & running updates. To get the most out of edge computing, software development, machine learning, cloud architecture & artificial intelligence are the enablers to make your hardware succeed.
Creating an edge-based solution takes time. With several decision points on the way, the development process requires creating the analytics model, deploying it and also running the model at the edge. One should also look into the processing capacity.
The processing power
Compared to cloud computing, which advantage is the centralized processing power able to perform complex AI-algorithms, the processing power of the edge devices might not be sufficient.
The maintenance cost for the systems
When more and more de-centralized touchpoints are connected to the service systems, it adds complexity for the product and service ecosystem and the maintenance cost of the complete system might go high.
A local AI platform to build beneficial and privacy preserving AI by Google Research
Enabling high performance compute solutions for computer vision applications at the embedded edge
Distributed Services Platform bringing the cloud and edge architecture together
Intelligent and connected hardware products and edge solutions.
Cheaper GPU and TPU hardware
Graphical processing units and tensor processing units, that enable systems to act using less power.
More powerful processors
Easy to use, productised versions of computing hardware such as embedded chips and cost effective small computers like raspberry pi.
AI/ML Artificial intelligence (AI)
Computer systems able to perform tasks normally requiring human intelligence by using a combination of Machine Learning (ML) toolbox of algorithms and learning rules from data.
A microcontroller unit (a small chip) is embedded with the ability to perform AI/ML computing offline based on sensor inputs without needing a cloud connection.
Neural Network (Deep Learning) is one of the most popular Machine Learning tools. Particularly effective with very large datasets.
The 5th generation of wireless technology that operates on a broader set of frequencies than 4G. This enables faster speed and latency in data transfer, as well as uninterrupted connectivity
Indicators of change
Get in touch with our team
Get in touch
You can find individual contact details for the friendly folk above.