Ganzi believes that we are on a 10-year journey to truly enable the infrastructure to support Gen AI and we must acknowledge what the challenges and opportunities are and navigate it as a sector together.
At the keynote address of Metro Connect USA Ganzi said, “We have a once in lifetime opportunity to build AI infrastructure – much like public cloud was 10 years ago. Every 10 years, we get an opportunity. Making sound investment decisions is the most important thing to do as a sector. What we do in this room is vital and important.
“We are in the 11th year of public cloud today. Think about what it took to build public cloud and we are now about 60% of the way. We are now focused on hyper edge. But none of these businesses are doing anything with AI and many are still focusing on the cloud. But the backbone of AI infrastructure is cloud enabled. We are five or six years away from Gen AI. But don’t discount importance of wireless in terms of where AI goes.”
Ganzi believes that we are on a 10-year journey to truly enable the infrastructure to support Gen AI and we are around 3-4 years away from true Gen AI. He told the audience we must focus on re-architecting networks to enable those applications, much of Gen AI will run on software defined networks.
He said “The opportunity to partner with hyperscalers with a shared infrastructure model sits in front of us. There is a reason why the shared infrastructure model works. This is the discipline we must think about in building Gen AI. Gen AI is a big opportunity. The demand for what we do across all businesses has doubled in 12 months. We accept the macro environment today is incredibly difficult.
“Ultimately AI feeds itself. It is a self-fulfilling prophecy in terms of amount of data it consumes. As these data centres and the impact on power increases it puts more pressure on us as landlords. 80% of all DC power will be consumed by AI. Today it represents less than 10%. This is changing very fast.
“We have another 8 – 9 years to go from language-based models to micro edge AI, it requires a lot of connectivity. If you own long haul and metro networks, you are seeing this demand with customers. And you need to build metro networks. The final frontier is edge. Gen AI to the edge is only enabled at the cell tower model.”
Ganzi believes we sit at a unique intersection of supply and demand with many questioning where to invest in AI but patience is key. “It is important to take your time. AI isn’t going anywhere; it isn’t a boat leaving the dock. It is a multi-decade journey, just like the cloud.
This is a big sandbox, and this sandbox is going nowhere. All these opportunities are here to execute as a sector.”
