It’s hard to imagine a world without continuous connectivity that puts work, entertainment, shopping, finances, navigation, translation and more at our fingertips. And it’s almost impossible to imagine any of this working without cloud providers that are building application hubs that enable businesses to deliver the information and content that enriches our daily lives.

Cloud providers play an essential role in our digital lives by delivering services and applications with immediate responsiveness. They, in turn, rely on sophisticated optical transport technology to connect us to data centers across the globe. But as demand for artificial intelligence (AI) workloads, real-time applications and massive data transfers grows, how are cloud providers evolving their optical networks to keep pace?

To answer this question, Nokia commissioned Heavy Reading to conduct in-depth interviews with leading cloud providers and carrier-neutral data center operators with the goal of better understanding their priorities, challenges and strategies for optical network automation. The new report, Network Automation for Cloud Providers: Interviews and Analysis, takes a deep dive into their:

• Motivations and use cases for investing in network automation. 
• Approaches to IP–optical convergence.
• Adoption of AI and machine learning (ML). 
• Concerns about larger-scale integration of automation into their operations support system (OSS) ecosystem. 

Here are three of the study’s key findings, which reveal how leading cloud providers are preparing for the next wave of cloud connectivity—and how their approaches compare with those of communications service providers uncovered in an earlier Heavy Reading study co-sponsored by Nokia.

1. Automation is critical for improving operational efficiency and driving new revenue 

One message was consistent across the interviews: Automation is a way to both reduce costs and an engine for growth. Cloud providers are using network automation to open new revenue opportunities and differentiate their offers, particularly in the case of network as a service (NaaS). 

Network automation is also driving a new category of AI exchanges (e.g., graphics processing unit as a service, or GPUaaS), which takes advantage of network assets in key locations and multi-tenant occupancies within data centers and campuses, along with the fiber-rich interconnects between them.

Subscribers can use NaaS to configure connectivity services on demand, using self-service portals similar to those offered by major cloud platforms. This on-demand model improves the customer experience and generates new revenue, such as with optical wavelength services. As Thomas King, CTO of DE-CIX, notes, “Automation enables the NaaS business model. Without it, offering self-service provisioning in seconds wouldn’t be possible.”

In addition to generating revenue and improving the customer experience, automation helps providers simplify operations, reduce human error, optimize infrastructure, enable open networking and accelerate service delivery. Compared to manual configurations, which are often time-consuming and error-prone, automated processes deliver faster turn-up times and more consistent service quality. Providers also emphasized that automation gives their valuable engineering resources more freedom to focus on innovation rather than routine tasks.

With its ability to unlock new premium services and expand addressable markets while streamlining existing operations, network automation has become a core strategic element for cloud providers, rather than just a tactical upgrade.

2. IP–optical convergence is accelerating, with coherent routing taking the lead

We talk about IP–optical convergence quite a bit here at Nokia, and for good reason: The study found cloud providers are rapidly adopting coherent routing, known within the industry as IP over DWDM (IPoDWDM), integrating coherent pluggable optics directly into routers and connecting them to dark fiber or optical line systems.

While many traditional telecom CSPs are still in the planning phases of coherent routing adoption, cloud providers are much more like hyperscalers and are well down the road in their coherent routing journey. They are embedding optical coherent transport directly into routers, thereby simplifying network design, reducing hardware complexity and lowering operating costs. As DE-CIX’s Thomas King says, “Managing optical transport directly from the router eliminates the need for separate DCI boxes, making the system less complex and easier to manage.”

Most cloud providers are transitioning from 100G to 400G connections now and considering 800G and beyond for future scalability and flexibility. The ability to operate 800G router ports using breakout cables at 2 x 400G to interoperate with existing equipment running at 400G or configure the ports to operate at 400G to achieve longer reaches will help cloud providers get maximum value from their capital investments. The need for higher-speed connectivity is driven by the rapid growth of east–west data center traffic and the rising demands of AI workloads.

This illustrates a clear trend: Cloud providers are not waiting for traditional telcos to set the standard. They are designing fully integrated, scalable and efficient optical transport that meets their specific needs now and will help grow their customer base.

3. AI and ML hold promise, but it’s still early days

AI and ML are often hyped as game-changers for optical network operations—and we at Nokia are big believers. However, the study found that most cloud providers are still in the early stages of integrating AI and ML into their networks. They’re focusing on education and evaluation before deploying at scale.

Nevertheless, cloud providers see strong potential for AI and ML to improve network troubleshooting, automate root-cause analysis, optimize traffic engineering and enable predictive analytics. By analyzing massive volumes of network data, AI can help them anticipate failures, optimize resource usage and ensure higher service quality.

But AI isn’t quite there yet. Cloud providers cited the lack of mature, open AI products and interoperability issues in multivendor environments as major barriers to adoption. As Scott Brookshire, CTO of OpenColo, explains, “A single-vendor AI solution isn’t enough. It needs to work across an open network, aggregating data from multiple vendors in a meaningful way.”

Cloud providers also raised issues with security and data sovereignty, particularly around sharing sensitive data with external platforms. While hyperscalers have the resources to build their own custom AI stacks, most cloud providers have competing priorities that limit their ability to undertake the necessary software and infrastructure development internally. 

But the foundation for AI-driven automation is being put in place today. As cloud providers evolve their operational models and address interoperability and security concerns, AI and ML will become more essential for automating their expanding transport networks. The implementation of these technologies will be prioritized by use case and return on investment.

Learn more: Download the full study

The Network Automation for Cloud Providers report provides insights into the priorities and strategies of today’s leading cloud connectivity providers. These insights can help guide your own network evolution.

Download the full study to explore all the findings and learn how you can stay ahead in an increasingly competitive cloud connectivity landscape. 

Download the infographic “Automation, IP–optical convergence and AI: Evolving cloud connectivity” to get a quick view of the report’s key findings.