There’s a belief that the future of Cloud is compute, and there’s new evidence that the future has arrived. We are now moving away from the server age, into an era of compute. In an interview with John Furrier and Dave Vellante for theCUBE at this year’s HP Discover in Las Vegas, Antonio Neri, Senior VP & GM for HP Servers and Network, discussed the rising era of compute and how it’s set to work with hyperscale trends in the enterprise.

You Need Compute to Run Cloud

Those who believe in open compute think that servers are dead and will all be integrated into one big “God Box” or some kind of engineered system. To discuss the role of old blade models in this supposed future, Furrier asked Neri to offer up his thoughts on the old blade models and what innovations are currently happening with servers.

Without hesitation, Neri responded, “You’re not going to buy boxes. You’re going to buy compute.”

He further explained that compute is needed to run the Cloud, which is why it’s not about the box. Rather, it’s about “optimizing the compute resources for the right workloads.” Neri said that the HP server strategy is to basically “provide the right compute for the right workloads at the right economics, and that means you will employ several different architectures,” he added.

Giving an example, Neri said if you think about the workloads in the traditional enterprise, HP has been leading with the right compute platform in that space with ProLient, its rack and tower business. Mission critical workloads are about scalability, data integrity and performance, which are all about availability. That’s why we HP has a broad portfolio of mission critical platforms, like the DL580, mission critical x86 and NonStop, which Neri mentioned has new innovation.

How architecture is going to evolve

Vellante admitted that he and Furrier previously joked about how the future of Cloud was compute. It’s really not. He added that, essentially, there is the persistent storage layer coming together with the compute.

“The first instantiation of that, believe it or not, was Moonshot because Moonshot was the first instance where we brought that concept into the architecture,” responded Neri. He then said that the next steps are going to be system on a chip (SOC), then universal memory, then photonics. According to Neri, this is how architecture is going to evolve over time.

• Compute and hyperscale

Hyperscale is bleeding into the enterprise. But, how much of the existing market is right for hyperscale systems? Neri believes that, in terms of total addressable market (TAM), the enterprise market for hyperscale is going to double over the next five years. Today, the growth of that market is growing between 20 and 35 percent. Some of that growth is driven due to large service providers doing big buys, dependent on the seasonality of the business. It’s also tied to their services in the Cloud, which is a different type of economics.

Ultimately, it’s about being able to provide a completed set of solutions, customizing those solutions, then adapting infrastructure to the app. Most apps are designed with new architectures, allowing the ability to drive the economy between management, the app and the infrastructure to a whole new level.

“That’s why I said we are moving away from a server era to a compute era,” said Neri. He explained that the compute era is going to be dominated by a workload centric approach, and the hyper scale will be dominated by those apps going over on the Cloud. Neri added, “Those service providers are going to optimize that app to the infrastructure to drive the maximum possible performance, power, efficiency and space.”

You Have to Out Compute

Concluding the interview, Furrier asked Neri to offer up his opinion on why this moment in time for the world and for technology so important. “Well, first of all, it is almost a privilege to be at this point in time in this industry,” replied Neri. He then said that the speed of change is impressive and that you have to move fast. However, speed “matters in a context of a strategy.”

In closing, Neri said, “to out-compete, you have to out-compute, which means that you have to be ahead of the time or at least one, two steps ahead all the time.”