MSPs need to move beyond infrastructure hosting by offering customized, high-margin services to their customers. GPUs are getting a lot of attention as a means to accelerate application performance, but GPUs are financially out of reach for many companies. GPU-as-a-service is one potential source of new MSP revenue and a new level of customer satisfaction.

Let’s first take a look at why GPU interest is rising. Artificial Intelligence (AI) and Machine Learning technologies are quickly becoming commonplace today and are shaping our experiences in computing like no other time in history. Interactive speech (e.g. Alexa, Google Home, etc.), Visual Search and recommendation engines are just a few of the consumer applications that are available today on our phones, websites and e-commerce platforms, and which leverage AI and Machine Learning.

The impact of Machine Learning is getting broader with enterprise applications in health sciences (e.g. Dr. Watson), finance, security, data centers and cyber surveillance. General-purpose CPUs cannot deliver the user responsiveness and inference latency required by complex deep learning and AI workloads. That’s because – unlike GPUs built for this purpose – general-purpose CPUs are not designed to rapidly perform parallel operations on large amounts of data, e.g., multiplying matrices of tens or hundreds of thousands of numbers. Processing large data sets through the same hypothesized algorithm for learning and for intelligent inference is a fairly common operation in Machine Learning and deep learning applications, and GPU processing is an infrastructure requirement for such processing.

But GPUs are an expensive acquisition for most companies. Here’s one example. With GPU cards that sit in a server costing thousands of dollars, companies are looking for alternatives to an outright purchase. Even though customers are looking at accelerating a range of applications beyond just AI and Machine Learning, they will want to test their GPU thinking, many of them without making a large up-front investment.

So, MSPs can offer GPU-as-a-service and allow companies to dial up GPU performance on demand and without a big Capex investment. This not only brings new revenue for the MSP from existing customers, but it’s also a way to attract new customers. MSPs can carve up GPU costs and spread them out across a lot of customers.

ZeroStack’s GPU-as-a-service capability gives customers powerful features to automatically detect GPUs and make them available in the ZeroStack environment. ZeroStack helps MSPs maximize investments in GPUs because admins can configure, scale, and allow fine-grained access control of GPU resources to end customers. End customers can enable GPU acceleration, deploy new Machine Learning and deep learning workloads with tools such as TensorFlow, Caffe, etc., and give those apps dedicated access to multiple GPU resources for order-of-magnitude faster inference latency and user responsiveness. End customers can access these applications with one-click provisioning through the ZeroStack AppStore.

AI and Machine Learning are hot new market spaces. MSPs can be heroes to their customers by enabling these applications in a secure, cost-controlled, and high-performance environment. GPU-as-a-service gives customers an easy and cost-effective entry to high-performance processing on demand and opens up compelling new revenue sources for MSPs.

For more details on GPU-as-a-Service, go here.


Leave a Reply