Signpost: Flexibility beats brilliance in the AI gold rush

Artificial intelligence may be the most glamorous story in technology, but many companies have a more ordinary frustration: they cannot always get the hardware they want, when they want it, at a price they can justify.

For many IT teams, the problem has become brutally practical. They are making long-term infrastructure decisions in conditions that feel anything but stable.

Multi-cloud software company Nutanix, which serves more than 30,000 businesses worldwide, presented a simple strategy to address this conundrum at its annual. NEXT conference in Chicago this week. In a market shaped by supply constraints and rising complexity, it said, flexibility may prove more useful than the brilliance AI offers.

Rajiv Ramaswami, CEO of Nutanix, set the tone in his opening keynote: “Today, it’s more important for you to have the flexibility to run anything, all your applications anywhere, having the flexibility to run on premise, run in the public cloud, pick the hardware of your choice, be able to select your hardware assets, given the situation right now with respect to supply shortages on hardware driving up the prices.”

That reflects a shift in how companies are approaching AI. Much of the discussion still focuses on graphics processing units, or GPUs, the specialised chips widely used to train and run AI models.

In practice, the constraint is broader. In a media briefing, Ramaswami said: “You want to get the most out of their investments. You don’t want to have resources being used inefficiently. You don’t want them sitting there idle, and in this world of GPU inferencing and agenting, you’re going to want to maximise the utilisation of that as a shared resource.”

That focus on utilisation points to the pressure companies are under. Investment in AI is being approved, but there is little tolerance for waste. Systems that are underused or duplicated quickly come under scrutiny. The question is no longer only how to access advanced computing power, but how to use it efficiently once it is in place.

Nutanix is positioning itself around that problem. The company first made its name in hyperconverged infrastructure (HCI), a model that combines computing, storage and networking into a single system managed through software, rather than as separate layers. That approach reduces complexity and makes it easier to deploy and scale systems.

The company is now extending that approach across public cloud services and existing storage systems.

“Five years ago, we were HCI company,” Ramaswami told Gadget. “Largely, many customers were using us for that HCI platform, to be able to handle all the workloads that they run. But the same time we recognised that the world is changing around us. Our customers want to operate in multiple public clouds, and so we focused on evolving from this HCI company to a hybrid cloud company.”

Asked how the company might evolve over the next five years as a result of AI, he said: “Fundamentally, it’s based on what we can do more for our customers. Where are our customers today, and where are they going? And that, of course, has a direct impact on the market opportunities that we will see. And therefore what we should do.

“The next change that we see here for the next few years is around the evolution of modern cloud native applications and AI and even before that, this whole notion of external storage.”

That reflects a wider pattern. Companies are running systems across their own data centres and multiple cloud providers at the same time. They are maintaining older applications while building newer ones. They want to avoid being tied to a single platform or supplier, particularly when costs and availability can shift.

The same pattern appears in how AI is being adopted.

“AI adoption is fairly early in the enterprise today,” he said. “If you look at the use cases today, those uses are practical: searching documents, summarising content, transcribing speech, or assisting with routine workflows. They do not require large-scale system changes, but they do require infrastructure that can expand gradually.”

That pace suggests a steady build rather than a sudden shift.

The largest cloud providers will continue to dominate on scale, but the current phase of the market is highlighting a different strength: the ability to keep systems running and evolving under imperfect conditions.

* Arthur Goldstuck is CEO of World Wide Worx, editor-in-chief of Gadget.co.za, and author of “The Hitchhiker’s Guide to AI – The African Edge”.