How we built Cloud WAN to support businesses amid the rise of AI
When a suspected undersea avalanche damaged four subsea cables along the West African coast in March 2024, major internet outages were reported from Côte d'Ivoire to South Africa. As the continent scrambled to reconnect, Google provided a 9,000+ mile lifeline.
The Equiano subsea cable — which connects Portugal to South Africa, with landing points in Togo, Nigeria and more along the way — stayed live during the disruption. Businesses using Equiano stayed live, too, while others impacted by the outages turned to us to get back online, buying up fiber and capacity. “We were able to protect Google Cloud customers’ traffic and aid in the resilience of the global internet,” says Subhasree Mandal, a lead on the Global Network Technology team. “It showed how far our network has come.”
Our global wide area network (WAN) — which supports Google applications and powers Google Cloud — has indeed come a long way over 25 years. Today, it spans 2 million miles of lit fiber, with 33 subsea cables and 202 network edge locations connecting 42 cloud regions and 127 cloud zones. It’s highly reliable, with teams working round the clock to ensure the network is live and functioning, and designed to meet the demands of the AI era.
Our global network today. Blue lines represent fiber-optic cables, while blue, red and yellow dots represent network edge locations and cloud regions.
At Next 25, we announced that we’re making Google’s global network available to external customers for the first time with a new offering called Google Cloud WAN.
“Connectivity has become increasingly complex for enterprises,” says Group Product Manager Satish Kondalam. “Many have developed multiple solutions for site-to-site connectivity that can be costly, and different solutions to connect to the internet and cloud services. It’s led to fragmented networks with inconsistent security and a constant need to balance reliability, speed and cost. The rise of AI — and the data demands that have come with it — has only further complicated things. We’re excited to help our customers increase network speed, decrease TCO and simplify their networks with Cloud WAN.”
Customers have long come to Google Cloud to take advantage of the same compute technology that powers Search, Gmail and more. Now, they’re also able to take advantage of the same network underpinning it all. With Cloud WAN, Cloud customers can enjoy our scale and reliability, and access state-of-the-art technology like the Premium Tier network, which ensures traffic connects to Cloud WAN through the geographically closest network edge location, for lower latency and a better user experience.
Cloud WAN enables customers to securely connect branch offices and campuses to public and private applications in Google Cloud and other clouds; Service as a Software (Saas) applications; and the internet, through capabilities like Cloud Interconnect — which provides private and dedicated low-latency connections between Google Cloud regions and data centers — and Cross-Cloud Interconnect, for connectivity between cloud environments.
The network they’ll use has evolved over the decades to meet dramatic shifts in technology.
“In the early days, we built the network for applications like Search and Ads, and our focus was reliability, scale and efficiency,” Subhasree says. “When we acquired YouTube and moved into streaming, we adapted the network to ensure we delivered high-quality video. With the rise of cloud computing and Google Cloud — when suddenly we weren’t just supporting our own apps, but our customers’ — we focused on greater resiliency, security and regional reliability.”
The AI era brought new challenges, starting with the scale of traffic AI-powered apps and model training sends to our network. “We introduced a multi-shard horizontal network architecture to swiftly grow capacity,” Subhasree says. “Here, each shard is essentially a different instance of the network that exists independently, and we can scale the network within each shard as well as increase the number of shards as demand increases. It’s like we’re offering capacity from multiple ISPs, which ensures redundancy, too.”
With this new architecture, we increased our WAN bandwidth by 7x between 2020 and 2025. And as the network’s architecture evolved, so too did its physical scale, with new data centers, cloud regions and cables creating greater reach and capacity.
Customers have already been using Cloud WAN in the lead up to launch, including Nestlé, the world’s largest food and drink company. They connected with Google after deciding to shift to a cloud-based network backbone. Following the deployment of Cloud WAN, including using the global Premium Tier network to connect their branches to their cloud ecosystem, Nestlé’s global head of IT platforms shared that their app performance is up 40% — and costs are down.
“It’s gratifying to hear those customer stories,” Subhasree says. “We were initially building Cloud WAN just for Google and now we’re building it for everyone. With the diversity of applications we’re running and the volume of data, that’s a huge responsibility — it’s, dare I say, uncomfortably exciting. But we’ve put an incredible amount of work into this network, and we’re ready to help the world connect.”
