Carriers are drowning in AI traffic - but that’s a boatload of cash if they can figure it out
Cisco, Juniper and others are packing AI into networks
With approximately 400 startups engaged in native AI, a bubble is already inflating
AI is impacting carrier networks in two significant ways. First, it’s increasing demand for network capacity by adding huge volumes of traffic, presenting carriers with a bear of a problem attached to a long fluffy tail of revenue opportunity.
Carriers need to figure out how to service this exogenous bandwidth demand which involves money and work (boo!). But once they’ve accomplished this, they can capitalize on the revenue potential that comes with it (hoorah!).
The second effect is all positive. Vendors, including Arista Networks, Cisco, Ciena, Blue Planet, Huawei, Juniper Networks, Nokia and ZTE Corporation, are elevating their hardware and software offerings with wicked smart AI features to automate, optimize, troubleshoot and secure networks.
Oddly, our industry lacks an agreed terminology to describe both of these AI conditions. Analysts and media discuss “AI-driven” networks — but are they referring to AI traffic (the problem) or AI product capabilities (the solution)? To clarify the situation, I offer a suggestion: refer to the network impact of AI as ‘AI bandwidth’ and the integrated capabilities within products as ‘Native AI.’
The incumbents in this market have company. By my reckoning, there are around 400 (!) AI-focused startups worldwide. This includes areas such as RAN, core network functions, OSS/BSS automation, network analytics and orchestration, as well as AI tools for software-defined networks (SDNs), edge/cloud-native telecommunications and 5G network slicing optimization.
Regardless of the CEO of Palantir Technologies's opinion, this is very much a global movement (except in Gaza, which at one point had a burgeoning comms startup scene but now does not, as it’s challenging to focus on coding when two of the world’s most powerful nations are working together – with significant help from Palantir - to bomb you into the stone age).
Four hundred is a lot of startups. Too many for the market they are targeting to support. Clearly, a bubble is being inflated here. This is an inevitable factor of the software nature of AI, which gives it a remarkably low barrier to entry. Unlike traditional telecom products, most AI startups don’t need to develop bespoke hardware. (Mad software skillz are all you need, bruv).
The big question now is whether this vast startup ecosystem will lead to genuine innovation and value or if it will experience the same quality control challenges seen in other newly opened markets. By removing the need to cut down and pulp trees to share information, the internet largely destroyed the quality and value of trade journalism.
Similarly, the music industry turned to shit as vinyl was replaced by digital audio, enabling platforms like Spotify to co-opt the music scene, inflicting auto-tuned Taylor Swift and Travis Scott pap on the world and making it impossible for 99% of independent musicians to ply their trade.
It's all terribly interesting. And exciting! Welcome to a shiny new age of AI-native networking, brothers (and sisters).
And for more of this kind of thing but in fun-size video format, check out https://www.youtube.com/@fiercenetwork_tv
Op-eds from industry experts, analysts or our editorial staff are opinion pieces that do not represent the opinions of Fierce Network. Read all of our op-eds here.