- Lumen is reshaping internet architecture through strategic “coopetition” with hyperscalers
- Lumen and Google Cloud are building new 400G cloud on-ramps to replace outdated internet infrastructure
- A single Lumen port connects to multiple clouds, cutting costs and complexity
When you think about the cloud, Lumen’s name is probably not the first that comes to mind. It’s all the more surprising, then, to know that the company came up in a surprising number of conversations at Google Cloud Next earlier this month. That – apparently – was no coincidence.
Lumen’s name was in the news at the show thanks to its collaboration with Google Cloud on the latter’s new Cloud WAN product. But Lumen’s ambitions are much, MUCH broader, CTO Dave Ward told Fierce.
Ward first articulated his desire to reimagine the U.S.’s internet infrastructure in an exclusive interview with Fierce in May 2024, arguing that existing infrastructure isn’t optimized for cloud and treats cloud traffic as second class. He expressed a desire to reweave the fabric of the internet to enable not just point-to-point connections through colocation faciltiies, but multi-cloud access through a virtualized network exchange.
Catalyst for internet infrastructure change
At the time, it sounded like a feat that would take years to tackle, but it seems AI has provided the push needed to get customers on the bandwagon. And by customers we mean major names like Microsoft and Google Cloud.
Which brings us back to Google Cloud WAN. Ward told Fierce that when he sat down with Muninder Sambi, the hyperscaler’s VP and GM for Network and Security, to talk through the pair’s partnership potential, he told Sambi that connecting enterprises to Google’s backbone fiber network via existing on ramp locations wouldn’t work. Ward said that’s because these carrier neutral facilities don’t have enough power or space and lack the 400G connectivity modern AI applications need.
His solution? “Let’s rebuild the U.S. national internet architecture away from those facilities and into new location.”
So that’s what they’re doing, 400G links and all. “These new on ramps will be at all new, high-speed locations distributed throughout the country,” Ward said. They’re being built to hook into both existing Google data centers as well as those still in the planning and construction phases.
But these links won’t be the traditional point to point kind that only connect to Google, Ward said. Buying a port on Lumen’s connectivity fabric means enterprises will also have horizontal access to other clouds as well.
That’s a big deal because it means “I just took out over 50% of the ports and a double digit percentage of the cross connects required. So, it’s much more valuable from a cost perspective but also for managing your bandwidth to workloads,” Ward said. “It also reduced the friction of getting to multi-cloud.”
Connectivity theory versus reality
How does Ward know this will work? Lumen, he said, is customer zero for the fabric he’s talking about.
Lumen already has buy-in from Google Cloud and Ward said the company is also in talks with other hyperscalers as well as specialty clouds (storage, SaaS, UCC, etc.) to replicate this model.
“It sounds quite egotistical to say I’m taking on the internet architecture, but in fact you have to if you want to build a modern enterprise,” Ward said. “The cloud will continue to expand the faster I can get customers there with higher bandwidth”
He continued: “We see massive expansion of AI data centers and data centers in general. I’m building to meet that but I’m building it with the ability to make it easier, lower friction, lower cost and API driven.”
Cooper-tition
For the record, Google isn’t the only hyperscaler to have the bright idea to offer its backbone network as a product. Microsoft launched its Azure Virtual WAN (an overlay on its global backbone) in 2018. AWS followed suit with its iteration of Cloud WAN (yes, same name) in 2021.
But as AvidThink Founder and Principal Roy Chua noted, Google could get a little help gaining traction thanks to timing and the present AI boom.
As for why Lumen is teaming up with Google Cloud on this rather than, say, going it alone, Chua said it’s all about “cooper-tition.”
Yes, it’s true that there could be some overlap between the company’s backbone networks. But hyperscalers lack an access network, which means last and middle mile connectivity is valuable for them, Chua explained. And for Lumen, Google offers a global scale partner. Voila, coopertition.
Judging from Ward’s comments, it looks like there are a few more deals with rivals/partners (riners? We tried.) up its sleeve. Stay tuned.