Key points

• Established secure, private interconnectivity between global resources in AWS, Google Cloud, Alibaba Cloud, and Tencent Cloud
• Accelerated AI model training and inferencing via high-speed cross-border dedicated lines
• Simplified expansion projects and improved scalability via Zenlayer’s one-stop shop and self-service management portal

 

About the customer

Industry: AI
Needs: Global scalability, cross-border connectivity, direct cloud connections
Zenlayer services used: IP Transit, Cloud Networking, Cloud Connect

The customer is an AI startup with valuations at over $2.5 billion that specializes in natural language processing models. Their AI systems integrate various modalities that include text, voice, and visual inputs for applications across diverse customer sectors like office automation, social media, and healthcare.

 

The challenge: ensuring a reliable, flexible, and scalable network with global coverage

As global user demand soared for AI services, the startup wanted to enhance their AI systems to ensure superior service delivery and user experiences. In order to improve the efficiency of their AI model training processes, they needed a scalable and high-speed network to rapidly integrate GPU and Tensor Processing Unit (TPU) resources. This network has to be flexible and reliable enough to support enhanced training and customized models across various locations worldwide.

Further, to ensure smooth operations and consistent service quality across the globe, the customer also needed secure, compliant cross-border connectivity to eliminate bottlenecks in interconnecting domestic and international resources.

 

Securely interconnecting global cloud resources

We provided the AI startup with a solution that privately and securely connected their resources in AWS, Google Cloud, Alibaba Cloud, and Tencent Cloud within and between the US, China, and Asia-Pacific using our hyperconnected global backbone, guaranteeing high availability and low latency networking critical to the startup’s AI model training and inferencing.

 

Using Zenlayer’s direct connectivity to all major public clouds, the customer was able to rapidly deploy compute and start generating output. Each cloud connection was established with a redundant path to ensure uptime and easy access to their global cloud resources.

Zenlayer’s premium network blend and data center services enabled the startup to rapidly integrate of their GPU and TPU resources, enhancing the speed and cost-efficiency of training processes while improving service reliability and performance in key global regions.

 

Accelerating training and inferencing

To ensure consistently high-quality connections between domestic and international resources, we leveraged our partnerships with the top local carriers and ISPs throughout the Asia-Pacific and China to provide the customer with premium IP transit and flexible, reliable cross-border dedicated lines. Zenlayer’s networking services helped the customer to seamlessly interconnect global training and inferencing clusters, giving the AI company additional agility to quickly and cost-effectively conduct enhanced training and customized models across global locations to accelerate development.

 

Simplifying expansion with a one-stop shop

Zenlayer’s one-stop shop approach and fully integrated self-service platform, zenConsole, allowed the startup to monitor network usage and latency changes, set custom threshold alerts, and maintain constant awareness of overall network status. This ability to monitor and adjust in real time enables the customer to provide smooth cloud-native application integrations to their B2B clients, ensuring high-quality service delivery to end-users.

With zenConsole at their fingertips, the AI company can now take full control of their network on the fly. As long as there’s an internet connection, the customer can manage and optimize their network resources quickly and effortlessly anywhere according to real-time demands.

 

Looking ahead

We plan to collaborate with the AI startup to continuously enhance their infrastructure leveraging Zenlayer’s global network and end-to-end connectivity solutions. In addition, we will provide ongoing support and guidance to help drive the customer’s growth and success across the world.

 

Empower your AI applications with Zenlayer

Our global coverage, lightning-fast global backbone network, and direct connections to all major public clouds helped the customer expand to new locations and improve the quality of their services worldwide.

If you’re looking to expand your business’s global footprint, streamline operations, or accelerate apps or services to improve user experience, talk to a Zenlayer solution expert today!

For instant provisioning, check out zenConsole — our self-service platform that lets you deploy around the world in minutes.