Raiyaan Serang (BGV Associate 2019), together with Yash Hemaraj, Anik Bose, Eric Buatois, and Amir Nayyerhabibi (BGV Partners), shares perspectives around how and what to invest within the edge computing space in this second of a two-part series on the topic.
How (and What) to Invest in the Edge
Because of the constantly evolving state of the edge ecosystem, this industry is being developed by everyone from horizontal platform providers like MacroMeta, intelligent computational storage producers like NGD Systems, and chip manufacturers like Graphcore that are use case agnostic, to vertical players that are creating end-to-end edge based solutions that are industry specific such as FogHorn Systems. In any scenario, investments should be made where value is created to address one or more of the four key aforementioned domain areas: connectivity, bandwidth, data synchronization & consistency, and latency, with security as an overarching consideration. Since these areas are interconnected (e.g., pushing processing to the edge reduces bandwidth constraints), the driving question behind any investment should be: Will this help facilitate real time decision making? By looking at opportunities through this straightforward lens, investors will be better equipped to separate the signal from the noise in this space.
Use Cases
Edge computing has the potential to impact many verticals and is already being used today to address problems governing the domain areas discussed earlier. Below are a few present and future use cases grouped by some of the industries most ripe for the edge:
| Industry | Use Case | Description | Impacted Domain Area(s) |
| Energy, Utilities, and Public Sector | Smart Cities & Homes | Streetlights, CCTV cameras, road conditions, traffic stops, and other aspects of residential areas will generate more data than ever before. Thanks to edge computing, large cities with established connectivity won’t be the only places where data can be gathered and used quickly to reduce congestion, decrease crime, and efficiently optimize energy demand inside and outside the home. Although the smart home space has already made several strides, smart cities will be a more forward-looking application as government and utilities agencies start to understand and adapt to this technology. | (✓) Bandwidth(✓) Connectivity( ) Data Synchronization & Consistency(✓) Latency |
| Predictive Maintenance | By actively monitoring the health of sensitive infrastructure in areas with potentially poor connectivity and proactively recommending repairs to prevent outages and unscheduled downtime, power plants and factories can improve cost efficiencies and extend the life of expensive machinery. Companies today are already investing in this area, and this investment will continue to grow. | (✓) Bandwidth(✓) Connectivity(✓) Data Synchronization & Consistency( ) Latency | |
| Travel & Transportation | Autonomous Vehicles | AV’s are set to generate 4,000 GB of data per vehicle per day. Although likely 5+ years away from becoming mainstream, autonomous vehicles will be edge devices that will conduct much of the data gathering, processing, and storage at the vehicle itself with no direct need to communicate in real time with the core to make decisions, and will instead leverage connectivity for routine software updates. However, as the technology develops, instances may arise where the vehicles themselves will want to communicate with each other and with a central core at real time speeds to make the most accurate decisions possible, factoring in road conditions and weather forecasts along with car model-specific information. Aggregate information could then be consumed and used by customers to automatically schedule maintenance and by manufacturers to improve overall automobile design. Full use of edge computing for the AV space will require further research and advancements in technological capabilities. | (✓) Bandwidth(✓) Connectivity(✓) Data Synchronization & Consistency(✓) Latency |
| Insurance | As owning a car becomes less common and ride sharing becomes even more mainstream around the world, insurance agencies may explore new business models and continue to leverage driver performance information from on-vehicle sensors to better predict premiums and deductibles at an individual driver level. These systems will have to work across geographies and signal strengths and may be used to gather information during accidents. Companies are already doing some of this today, but this activity is only expected to increase as the technology develops. | (✓) Bandwidth(✓) Connectivity( ) Data Synchronization & Consistency( ) Latency | |
| Supply Chain | Inventory Tracking | By collecting information on thousands of assets without creating strain on network bandwidth and by utilizing advanced activity and temperature monitoring, products can last longer and stay fresher. This use case is being explored today by several organizations and will continue to evolve as consumers’ expectations shift to fresher ingredients and reduced waste. | (✓) Bandwidth(✓) Connectivity( ) Data Synchronization & Consistency(✓) Latency |
| Demand Prediction | Closer integration between producer and consumer through user-specific customer experiences with rapid feedback incorporation in the back end can drive organizations to accommodate shifting consumer tastes in near real time. This is a use case where edge computing can be leveraged to solve problems in the present. | (✓) Bandwidth(✓) Connectivity(✓) Data Synchronization & Consistency(✓) Latency |
These use cases represent a subset of how edge computing can solve problems previously unanswered by current cloud capabilities. However, it’s important to note that new use cases will develop as AI, IoT, and 5G unlock new opportunities for edge technology to take off and a dominant design is established.
Investment Opportunities
In the hardware space, investments should be focused on companies who are building low cost, low power, and high-performance solutions that bring connectivity and processing to the edge at low latency. Venture funding should consider companies that are trying to build the components that will enable this industry by bringing machine learning and artificial intelligence capabilities to the chips that will make up the edge devices of the future. The primary public cloud providers have introduced their own specific chip sets that enable this functionality, but the trade-offs involving power, performance, interoperability, and cost imply that different solutions will be preferred for different customer segments. Areas also exist within the telecommunications and networking space, where companies are working to tackle multi-access edge computing (MEC) through network slicing and network strings. Other opportunities also exist for companies focusing on building out edge infrastructure (data centers located closer to the edge).
In the software realm, investments should be made in platforms that are designed to allow developers to build the new set of streaming applications of the future that can process data bidirectionally without latency. These platforms should bring siloed applications together to allow for insights to be gathered in real time. Other opportunities exist for companies focused on the management and intelligence layer that will exist across different edge devices and the central cloud. Expect edge hosting, edge infrastructure as a service, and edge platform as a service models to develop.
While some companies are developing general purpose solutions, others are targeting specific industry verticals to create a one stop shop for organizations most ripe for the edge. Due to the fragmented nature of the edge ecosystem and the different needs of different verticals, expect companies with a more vertically focused, end-to-end approach to succeed in this space in the near term. Smart teams with deep domain expertise will be better equipped to create and capture value using edge-based technologies and will optimize for the use cases that matter most to the industries that stand to benefit, which will be informed by the four key edge domain areas. General platforms can be expected to evolve out of the vertical solutions developed as organizations tackle adjacent use cases and leverage professional services teams and partners to expand their footprint.
Conclusion
The edge computing landscape offers promise for those who can effectively differentiate between where the benefits of the edge are a must-have vs. a nice-to-have. Edge computing will not completely replace cloud computing across all scenarios and should be looked to as an extension of the cloud’s current capabilities. Although many high-profile use cases may be 5+ years away, several enabling trends and current needs warrant investment in this nascent space to solve problems governing connectivity, bandwidth, data synchronization & consistency, and latency. With every investment, the focus should be on enabling real time decision making.
