Penguin Solutions, formerly SMART Global Holdings or SGH, sees a new direction for itself that involves designing, building, deploying and managing next-generation enterprise infrastructure for AI and high performance computing. In Singapore recently to talk about it was Global Marketing Lead for Penguin Solutions’ Advanced Marketing segment, Stephen Greene who said, “We’re really taking the existing businesses and product lines, bringing them together and streamlining a bit, but focused on three core segments.”
Of these three segments – Cree LED, SMART Modular, and Advanced Computing – it is the combination of Penguin Computing and Stratus (now called Advanced Computing) that represents about 55% of the group’s revenue. This segment is focused on AI, high-performance computing (HPC), fault tolerance, and integrated management solutions.
Stephen wanted to highlight how Stratus for the past 45 years has really been innovating and pioneering in the world of fault tolerance, a capability that is valuable in this day and age of artificial intelligence and always-on computing.
“We helped lead the definition of high availability (HA) and fault tolerance in the edge computing space around 2018,” Stephen described, highlighting Penguin Solutions’ expertise and experience in protecting “high volume, high-risk data transactions in areas like financial services and telecommunications.”
This expanded into the industrial automation sector which Penguin Solutions is leading.
Stephen also detailed Penguin Solutions’ services and support pillar. “So, Stratus has had that for the past 45 years, and it is this service side that has really made us unique in the business… our ability really in and help customers as they lean in and put a lot of pressure on these systems.”
Trends that Penguin sees and wants to capitalise on
High availability computing is becoming increasingly critical in today’s era due to increasing reliance on digital platforms, cloud services, and real-time data by businesses and end users.
The rise of global connectivity, 24/7 operations, and demands for instant access mean that infrastructure must be resilient against uncertainties – from hardware malfunctions to cyberattacks and even natural disasters. By ensuring continuous service and rapid recovery, high availability computing helps organisations maintain their competitive advantage. Uninterrupted access to services is a baseline expectation these days.
Besides this rising expectation, Stephen also observed enterprises that are starting to virtualise their cybersecurity hardware. “Now, the availability of that hardware is going to be a lot more important.”
If we think about the edge as defined by where products are made or customers are served, you can see the role that AI inferencing and agentic AI would play.
Stephen Greene
Another trend that is developing if not already here, is workloads that are increasingly moving into hybrid environments. “By 2028, 75% of enterprise workloads will be run in hybrid environments.”
Stephen explained that at some level this extends workloads in enterprise/on-premise out to edge infrastructures. He called out the on-premise training and fine-tuning of large language models that is transitioning some of that effort towards the cloud.
We don’t play in commodity markets… [we focus on] very specific areas where data-intensive, critical workloads are going to run.
In fact, some enterprises that want to leverage LLMs in vertical markets find themselves moving to small LLMs in on-prem (or private) environments.
“If we think about the edge as defined by where products are made or customers are served, you can see the role that AI inferencing and agentic AI would play,” he explained.
This is driving Penguin’s focus on partnering with strategic solution providers or alliance partners, to go-to-market with across enterprise and edge infrastructures.
“We don’t play in commodity markets… [we focus on] very specific areas where data-intensive, critical workloads are going to run,” he emphasised.
Go-to-market
Ninety-five percent of Stratus business is traditionally through to the channel and this is expected to continue. However, Stephen added, “The AI and high performance computing business is more direct, however. We are not changing the fundamentals of the business model and how we will go-to-market.”
He also indicated that customers are increasingly looking at how to solve challenges from enterprise IT environments out to the edge.
“And we think we’re in a pretty good position, if not great position, to be able to solve that for them, and then do so directly or through select partnerships in the channel.”
