
During the recent Oracle AI World conference in Las Vegas, Bob Evans, Founder of Cloud Wars, sat down with Mahesh Thiagarajan, Executive Vice President of Oracle Cloud Infrastructure (OCI). Drawing on his extensive knowledge of Oracle’s infrastructure business, Thiagarajan provided Evans with an overview of how OCI is adapting to meet customer needs in the AI Revolution.
Software and Hardware Excellence
Thiagarajan has spent close to a decade working with OCI and, during that time, has overseen some of the significant advancements in the technology that continue to this day. “You just continuously innovate on what I think is foundational, and even until today, we’re still inventing new things in infrastructure,” says Thiagarajan.
One of the major leaps that Oracle has taken in recent months is its substantial investment in data centers to support the meteoric growth of the AI sector. How will this combination of highly advanced software with state-of-the-art steel and concrete infrastructure play out, asks Evans?
“If you think about what actually enables us to do infrastructure the way we do it, it’s at every level,” says Thiagarajan. “Today I announced Dynamic Resource VM, a fundamental technology [for] efficient planning of memory, which is one of the highest cost components in infrastructure.
“But — I couldn’t have attempted — or even thought about, doing something like that if I didn’t actually do Linux, [or] have a team that really does that.”
Thiagarajan explains that integrating hardware and software, particularly within Oracle’s AI infrastructure, is crucial. Ultimately, given the investments made by Oracle and its customers, the large clusters that power the company’s AI products must be exceptionally reliable and deliver a greater return on investment (ROI) than smaller alternatives. This reliability is achieved through a combination of computing power and the software layer that runs on top.

AI Agent & Copilot Summit is an AI-first event to define opportunities, impact, and outcomes with Microsoft Copilot and agents. Building on its 2025 success, the 2026 event takes place March 17-19 in San Diego. Get more details.
Latest Developments: Products and Partnerships
One of Thiagarajan’s major announcements was the launch of Oracle Acceleron. “Oracle Acceleron, for us, is a foundational networking concept that has three key pillars: the off-box virtualization device … foundational network architectures that offer reliability for very mission critical workloads … [and a] fabric accelerator, which is foundationally a suite of software that has the intelligence to be able to say, “I know where you’re going. I’m going to take you there very securely.”
Oracle has extraordinary strategic partnerships with NVIDIA and AMD, which are supporting its growth in the AI sector. Evans asks, what is the differentiator that enables Oracle to forge these critical alliances?
“We’re in the business of making sure we bring customers that flexibility and choice,” explains Thiagarajan. “The market is so big and there are tons of opportunities in terms of what our hardware partners can accomplish and what we can help them accomplish, and what we can accomplish for our customers. And so, for us, the goal is: put our customers first.”
Thiagarajan continues the discussion by documenting the progress of Oracle’s massive data center project in Abilene, Texas. “I think it was an audacious goal,” says Thiagarajan. “We basically said: “Look, we want to really get compute accessible really fast, and what are some ways in which we can actually get that in the hands of customers in less than a year?”
“So, what we’re doing at Abilene, it’s 1,000 acres, eight buildings it’s a pretty sizable, large deployment. I’m very excited to bring it online, but also, we’re doing much bigger things, which we announced as well. So, it’s going be fun.”
Ask Cloud Wars AI Agent about this analysis






