Behind AI is a robust cloud structure

getty

With all the attention on artificial intelligence nowadays, we assume that the delivery mechanisms underneath are ready to handle the workloads. That’s not necessarily the case, a recent survey suggested.

Cloud has been on the scene for more than two decades now. However, just 14% of organizations have a fully leveraged cloud to manage key business initatives, a study by NTT Data finds. This has implications for readiness to take on the heavy lifting required for AI – supporting compute, inference, and data.

It’s also important to remember that AI itself is increasingly supporting cloud management and provisioning, In this case, the study suggests there may not be enough cloud resources to support AI – its execution layer. Almost all executives in the survey, 99%, say AI is increasing demand for cloud investment, but 88% agree current cloud investment levels are putting AI initiatives at risk, the survey of 2,335 senior decision-makers finds.

Of course, there are public cloud services with almost bottomless capacity to handle any AI workload – AWS, Microsoft, and Google top the list. But for AI, many companies are finding greater utility in internal private clouds, according to David Linthicum, writing in InfoWorld. “The workload profile of AI is different from the workload profile of ‘move my app server and my database,’” he said. ”AI workloads are spiky, GPU-hungry, and brutally sensitive to inefficient architecture. They also tend to multiply. A single assistant becomes dozens of specialized agents."

Private clouds are attractive as they enable enterprises to choose “where to standardize and where to differentiate,” Linthicum continued. “They can invest in a consistent GPU platform for inference, cache frequently used embeddings locally, and reduce the constant tax of per-request pricing. They can still use public cloud for experimentation and burst training, but they don’t have to treat every inference call like a metered microtransaction.”

Expectations for AI are accelerating, but cloud investment has stalled, the NTT survey’s authors caution. A majority, 84% of organizations report flat cloud spending over the past year, even as 99% say the rise of AI, including agentic AI, has increased their need for cloud investment. This disconnect raises concern. Nearly 9 in 10 organizations (88%) acknowledge that their current cloud investment levels put cloud-native, AI and modernization initiatives at risk.

At the same time, only 14% of executives responding to the NTT survey could say they were to the point where they were employing cloud-led innovation to accelerate business transformation, or had "cloud-native services embedded in core
strategies, offering advanced automation, Al and machine learning, and continuous delivery."

While more organizations view cloud as an engine of growth (64%) than as a mere tool for efficiency (55%), they are not convinced it’s delivering. Only 49% report
being fully satisfied with cloud’s impact on innovation, and just 44% say they’re
fully satisfied with their broader IT modernization progress. “Cloud is widely seen as essential, but its impact remains limited — not because of a lack of intent but because of how it has been adopted and integrated into business and operating models,” according to the survey’s authors.

The NTT study’s authors outline six rules for helping cloud catch up with AI’s torrid growth:

Develop cloud and AI strategies in tandem: “With cloud deployment choices now directly influencing cloud outcomes, organizations are increasingly adopting a mix of public, private, hybrid and sovereign cloud models.”

Apply finops and consolidated strategies: "As investments stall and environments become more complex, more than half cite cloud cost management challenges and organizations expect a threefold increase in fully managed cloud platforms."

Assure cloud security: Confidence in cloud security is low, with only 36% overall voicing confidence. "Define clear roles and responsibilities backed by regular audits, reinforcing the importance of the fundamentals as technology ecosystems grow more complex.”