When audiences discovered motion pictures in the late 19th century, they were enthralled. They couldn’t have known, however, that the earliest films — short, silent and colorless — would eventually spawn the even more entertaining blockbusters of today. When the Soviet Union launched the first satellite in 1957, onlookers were similarly captivated. But at that moment, they couldn’t have possibly conceived of GPS and its far-reaching impacts. And when they logged on to the internet for the first time in the 1990s, consumers once again were blown away. Never mind their inability to foresee things like social media, e-commerce and streaming entertainment.
Artificial intelligence is at its own exciting, unpredictable starting mark, according to Nutanix President and CEO Rajiv Ramaswami.
“Enterprises are just beginning their AI inferencing adoption,” he said at the second annual RAISE Summit, which took place July 8 and 9, 2025, at Carrousel du Louvre in Paris.
Nutanix President and Rajiv Ramaswami interviewed by the Cube following his RAISE Summit stage appearance.
At the summit — a cross-industry event for leaders who are driving AI innovation — Ramaswami spoke on stage with Serge Palaric, vice president of alliances, global cloud solution providers and independent software vendors for the EMEA region at NVIDIA. During a 20-minute fireside chat, they imagined AI’s future opportunities and detailed the IT infrastructure that will be required to move them from “theoretical” to “operational.”
The Future of AI is Agentic
If AI were a book, its current iteration would be merely the prologue, Ramaswami indicated.
“With any new technology, things start out being fairly simple,” he told Palaric, highlighting three basic, initial use cases:
Administrative tasks like document summarization and automatic language translation
Customer support tools like AI chatbots
Content creation like writing code, authoring articles, drafting contracts or creating videos.
“Those three use cases drive the bulk of enterprise inference solutions today.”
But these initial use cases are “becoming more complex over time,” continued Ramaswami, who foresees a future in which AI is dominated not by simple, generative tasks, but rather by “complex multi-agent workflows being done through agentic AI.”
Imagine a hospital, for example. If a patient falls and injures themselves, Ramaswami suggested, one AI agent can detect it by monitoring video surveillance footage, while another can analyze the same footage to determine the cause of the fall. Simultaneously, a third AI agent can determine the next steps — such as dispatching a paramedic — while a fourth can monitor the situation to ensure compliance with hospital policy and industry regulations.
“All of these can be automated,” Ramaswami said. “You can have multiple agents that are working together to make all this happen behind the scenes in a very effective way. That’s the world of the future.”
AI is Moving to the Edge
Enterprises that want to maximize the return on their AI investment must be able to use their AI models whenever and wherever they’re needed, Ramaswami told Palaric.
“AI, like almost every other application, will … be hybrid. It's all going to run wherever the data is, and the data is going to be everywhere,” he said. “It’s not going to be in one place. Therefore, you have to be able to make use of the data wherever it is.”
Consider an industry like manufacturing, where real-time latency can be a major challenge.
“In manufacturing … you have to run [the application] right there where the data is being produced,” continued Ramaswami, who cited as an example Nutanix customer Micron Technology.
“They’re one of the largest semiconductor manufacturers in the world, and they … rely on very sophisticated algorithms … [for] things like defect inspection of wafers,” he said. “There’s lots of data being generated at the edge that needs inferencing right there at the edge.”
He sees retail as another industry where edge AI will thrive.
“If you’re in retail … you’re looking at surveillance and you’re trying to figure out, for example, when do you need to go open a new counter to serve your customers? There’s a lot of those kinds of applications at the edge that need to be done locally,” he explained, offering data centers as yet another example.
“People have their own private data that they don’t want to put in the cloud. They want to be able to run those applications there.”
Infrastructure is the Missing Piece
AI applications are where enterprises will generate revenue and gain a competitive advantage. But applications without infrastructure are like cars without gas — they won’t go anywhere, Ramaswami suggested.
“If you think about what’s needed from an enterprise perspective, you have to do multiple things,” he told Palaric. “You’ve got to get an infrastructure in place. Then you need the AI stack on top of it. And then, on top of that, you can build your AI application.”
Enterprises that jumped straight into developing AI applications will soon realize the importance of infrastructure.
“It’s not going to be any different from how other technologies got adopted,” Ramaswami continued. “There was a time when virtual machines were new. There was a time when … Kubernetes was new. In each of these cases, you started out with individual teams trying to do their own projects, then realizing there’s a lot of stuff that needs to be done at the infrastructure level.”
When they realize how time-consuming it can be to establish the necessary infrastructure, those same teams tend to pool their resources.
“They’d rather be innovating on things that matter to the company from a business perspective,” Ramaswami observed. “So, as these things start to mature, what happens is that you create infrastructure teams … that can then create all of the infrastructure that’s necessary for people to run their applications. I think the same thing is going to happen with AI inferencing. We are in the early days of AI inferencing in the enterprise, and what you will see is people will ultimately turn to infrastructure teams to stand up the AI infrastructure that they need so they don’t have to worry about it.”
Simplicity is a Strategic Advantage
For enterprises that are laser-focused on AI applications for their customers and employees, infrastructure can be “a lot … to take on,” admitted Ramaswami.
“There are a handful of sophisticated enterprises who can do all of this by themselves. But for the vast majority, they need simple, turnkey solutions,” he said, citing as a key enabler Nutanix’s longstanding partnership with NVIDIA.
Nutanix, on the one hand, provides cloud-based software and infrastructure tools to help companies run applications and store data, whether it’s on their own servers, in public clouds or at the edge. NVIDIA, on the other hand, provides the optimized hardware and software that companies need to run their AI models.
It’s “the best of both worlds,” Ramaswami said.
“You need compute. You need storage. You need the network. All of these need to be put together. And by combining the NVIDIA AI stack on top of the Nutanix stack, you get exactly a turnkey infrastructure where you can start out in very simple terms,” he explained.
“For example, you can get an OpenAI-compatible API. We can download a model, instantiate it on the GPU, make it available, serve it up. You can pick your models of choice from the repositories of choice. The workflows are automated. And then — most importantly — from an enterprise perspective, you care about governance. You care about security. You want to protect your IP. So, we have things like role-based access control. We have the ability to protect IP and understand what data you have. All of that is put together in a package solution that’s easily usable by an enterprise.”
The net effect is “invisible” infrastructure, Ramaswami noted.
“Infrastructure … should be something that exists,” he concluded. “It’s there. You can use whatever you want. It’s flexible and fungible so people don’t have to worry about it. We make that happen.”
Editor’s note: Learn about Nutanix Enterprise AI solutions and how to build an IT foundation for enterprise AI success.
Ken Kaplan is Editor in Chief for The Forecast by Nutanix. Find him on X @kenekaplan and LinkedIn.
Matt Alderton contributed to this story. Find him on LinkedIn.
© 2025 Nutanix, Inc. All rights reserved. For additional information and important legal disclaimers, please go here.