Can Edge Computing Exist Without the Edge? Part 3: The Economics of Edge Computing
Do the math. For the past decade, digital businesses have been investing in digital transformation initiatives. The promise is to increase top-line growth while maximizing customer lifetime value. As infrastructure improves, IT spending is shifting from the cloud to the edge. But while edge computing makes headlines, can you expect a meaningful return?
In my last blog, I suggested a simple model: If you derive value from taking immediate action on short-lived data, consider edge computing. But edge computing takes two forms -- cloud-out and edge-in -- so which is right for you?
Cloud-out capabilities are horizontal, and apply across many industries, applications, and use cases. Edge-in capabilities are industry-specific, and address very specific needs at the digital touchpoint. Yet depending on the use case, the edge could be in the core or at the digital touchpoint.
Consider the following edges:
Wide area network (WAN) edge: Database, networking, and secure access vendors focus on the edge between on-premises infrastructure and the cloud. Zero Trust discussions focus on the WAN edge because of the value derived from secure access to data and intellectual property.
Cloud edge: Cloud providers focus on maximizing value from their centralized deployments. They derive value from broad use case applicability that scales up heavy computing.
Data center edge: Hardware, colocation providers, and internet service providers (ISPs) focus on regional edge data centers. These smaller deployments aim to create an edge that is 1-100 miles from consumers. Here, the value is from creating bandwidth and scale that capture the promise of 5G.
Gateway edge: Mobile network operators (MNOs) and vertical solution providers are investing here. The aim is to filter, reduce, and process data less than a mile from a device. The value is in fast and secure decision-making for fleets and specialty devices.
Device edge: Hardware and licensed software vendors are investing here. The value is to execute business logic and apply security controls on the device itself. A critical need is to apply edge computing even when the device is not connected to the internet.
Cloud-out capabilities are common from the WAN to the data center. Edge-in use cases dominate between the gateway and the device.
So where is the money going today? Vendors are racing to establish footholds across many technology areas. The rapid switch to remote work in 2020 accelerated investment in secure access service edge (SASE). Gartner projects that 40% of enterprises will have SASE adoption strategies in place by 2024. The future of work requires secure, scalable solutions deployed at the WAN edge. Components of this approach, including secure web gateway, Zero Trust Network Access, and firewall as a service, are all growing at double-digit rates.
Cloud providers are investing to add edge computing capabilities to their platforms. Microsoft announced they are investing $5 billion in IoT and edge computing. HPE announced it will fund edge computing tech with $4 billion.
Infrastructure providers are spending to remain relevant for 5G and augmented reality (AR) and virtual reality (VR) use cases. PWC projects that the global edge data center market will triple, growing to $13.5 billion by 2024. Small, local data centers promise to reduce latency by bringing storage and compute close to the end user.
Meanwhile, the gateway edge is more fragmented. Investments in this area are vertical-specific. Expect nuanced approaches and rampant spending until individual verticals determine consistent architectural approaches.
Businesses are shifting investments to technology that improves customer loyalty and lifetime value. Chief among them are perimeter security and customer experience optimizations. Edge platforms provide solutions that deliver on both. Analysts estimate that security spending will continue growing by double digits through 2024. Meanwhile, over 90% of web performance growth will come from new accounts and use cases.
Cloud-out use cases are where business-to-business companies that monetize through websites and apps will invest first. Edge computing holds the promise of accelerating their digital transformation. Cloud-out is the natural next step because it applies to apps built for the cloud. These apps are always connected to the internet, and data needs to flow between the cloud and the edge. For these companies, scalable, centralized cloud infrastructure is critical. They need the cloud to manage customer and IP data, enrich user interactions, and mine data to derive value. These heavy compute use cases promise to unlock insights that improve customer loyalty while decreasing customer acquisition and retention costs. Edge computing helps by eliminating latency while reducing cloud costs when those insights are applied. This is the primary area where Akamai has been investing in our edge computing capabilities.
Edge-in use cases are most relevant for industry-specific use cases. There is clear value in eliminating costly on-premises computing deployments. Like cloud-out, real-time insights obviate the cloud due to long round-trip times. As Gartner's Bob Gill says, "The rise of edge-in cloud architectures results from the need to pull cloud services to the edge and use them selectively in edge-specific ways, rather than push public cloud architecture to the edge as a complete platform solution." This is an area where Akamai has more recently invested in our Internet of Things (IoT) edge cloud solution lines for over-the-air updates and publisher/subscriber messaging.
The challenge for most businesses is understanding where to shift their investments. Remember when Google Glass debuted in 2013? The technology was ahead of its time. The quirky $1,500 smart glasses failed to provide enough value to connect with a broader audience. Initial image and video capture use cases were better suited to mobile device interactions. One could argue that, years later, ByteDance captured that value via TikTok. Capturing video, editing and enriching it, and then sharing it via social media is a cloud-out use case. TikTok wins because the wearable or mobile device does not have to perform a specialized function. It needs to capture and share video.
There is an edge-in use case for Google Glass as well. From manufacturing to medicine, there is a massive AR market opportunity for skilled workers. Avesha delivers a fast and scalable deep learning platform based on artificial intelligence (AI) being used to assist doctors in detecting polyps during routine colonoscopies. Regular screening identifies cancers early and can help prevent new cancers from developing. In both cases, these tests can improve survival. The Avesha system is like a second set of eyes that can identify anomalies during an exam in less than 30 ms. The investment provides a clear return on investment (ROI) when you consider that, in the U.S. alone, there are over 20 million colonoscopies done per year. Improving efficiency and accuracy can save lives.
For most businesses, determining the ROI for edge computing use cases proves challenging. Many have begun experimenting with edge computing, but production workloads are rare. This is due, in large part, to "edge washing." When every technology vendor claims to have an edge computing solution, which should you choose? For example, Amazon touts how Wavelength provides the infrastructure that enables Avesha. The solution provides value by enabling AI inference (a form of computing) at the 5G mobile edge.
Most infrastructure providers cannot afford to build out edge data centers that can get to within one mile (or 20 ms) of consumers. There is value in embedding compute devices in MNOs. But who should make that investment? Hyperscale cloud providers have the breadth of technology that these use cases demand. Global edge platforms like Akamai have already created the infrastructure to get to within 20 ms of most consumers. In this example, partnering to marry the cloud to the edge makes more sense than proliferating edge deployments.
While edge computing hype intensifies, expect vendors to continue establishing self-serving reference architectures. Analysts project that repeatable implementation architectures will emerge over the next three years. Consolidation will occur over that time and create a daunting challenge for early adopters -- alignment to those standards.
Akamai has been engaging in cloud-out and edge-in use cases. In the next blog, we will cover use cases in each investment area, and highlight the business value for each.