Hyperscale, Cloud Data Center, AI & Big Data | Corning

Hyperscale Data Center Frequently Asked Questions

Have additional questions about hyperscale? Contact us today!

• What is difference between hyperscale data centers and traditional data centers?

  The simplest way to compare hyperscale data centers vs. traditional data centers is by their size and scalability. Hyperscale data centers are designed with density and performance at a scale that can be up to 1000x greater than an enterprise-style data center. Hyperscale data centers benefit from economies of scale, despite their incredible appetite for power, cooling, and fiber, and therefore are able to offer services a traditional data center cannot. Traditional data centers rely more on manual processes, and are typically built with custom designs tailored to specific organizational needs.

• What effect has the rise of generative AI had on fiber optic data center networks?

  To meet the computational demands of Generative AI, big data analytics and other AI-like applications, customers now require a new fiber-rich second or “backend” network to connect GPUs. As a result, networks require over 10 times more optical fiber compared to legacy server racks.

• What is a backend network and why do data centers need to consider them in their planning?

  In the context of Generative AI, the backend network often refers to the server-side infrastructure that powers the algorithms to generate new content. This could include deep learning models, such as Generative Adversarial Networks (GANs), which learn to create new data with the same statistics as the training set, e.g., images, music, speech, or text.

  The backend network related to Generative AI involves several elements, including Data storage and management, processing power, model training, and deployment.

• What is a neural network?

  A neural network is a group of interconnected units that send signals to one another. In the field of AI machine learning, a model or program will use interconnected units called “neurons” or “nodes”, to process data in a way similar to the human brain.