Edge Computing in the world of AI/ML

By: Joe Madden, Founder and Lead Analyst, Mobile Experts July 26, 2024

Collected at: https://www.rcrwireless.com/20240726/analyst-angle/edge-computing-in-the-world-of-ai-ml-analyst-angle

Mobile Experts has been watching the Edge Computing market develop, with investments by AWS, AtlasEdge, EdgeConnex, Verizon and others brewing up a rich soup of business models. Investment is all over the place: The money comes from hyperscalers, from telcos, from neutral hosts and enterprises. Some data centers are regional. Others are local, or even on-premises. Every world region has some investment going on, even Europe, where it’s difficult to get bankers to cough up the cash.

The situation is fluid; there is no single direction.

Amazon has established 41 Local Zones, and they have numerous success stories in gaming, manufacturing, IT, retail and other markets.   

Communications Service Providers (CSPs) deployed more than 50 edge data centers in the last year, targeting specific enterprise customers with local data centers in some places, and covering an entire country with data centers every hundred miles in other cases.

And Neutral hosts deployed 38 data centers in the last year, taking advantage of digital sovereignty laws in Europe and a fluid business model in the USA.

So, almost equal investment is coming from cloud players, connectivity players and real estate players. And there’s no clear correlation yet to show that a specific vertical market is tied to each business model. Will it be a free-for-all? Or will a preferred business model emerge here?

AI will push the players to make up their minds. Today, the investment climate for hyperscalers is heavily weighted toward huge data centers with billion-dollar racks of GPUs. There’s not much oxygen left for edge data centers. At the same time, enterprises would like to invest in AI/ML for their local operations. This will lead to upgrades in on-prem servers in Manufacturing, Logistics, Oil &Gas and Financial Services. AI will also lead to interesting local and regional data center activity for Utilities, Transportation, Education, Gaming, Drones, Retail and Healthcare.    

Industrial markets will benefit from enhanced operational efficiencies. Carpeted enterprises will benefit from quicker response times and AI chatbots. And the telcos themselves want to use AI to optimize their networks. The opportunity to get in front of these opportunities is coming soon.

The operators and neutral hosts from the telecom market have an opportunity now. They should move while the hyperscalers are distracted. They should blanket the developed world with data centers every 400 miles and partner with the big hyperscalers by offering the connectivity and real estate solutions. Private cellular networks and local data centers to run the AI/ML and Edge workloads will become a significant business area. Very few verticals want to move their computing stacks away from the hyperscalers. But 25-40% of the revenue flows to the connectivity and physical hosting of the data center, so the telecom industry players should take the lead, to make it possible to serve AI at the Edge.