Cloud adoption in an AI world

By Tiana Cline, Contributor
Johannesburg, 04 Apr 2024
Paul Spagnoletti, executive head of cloud and security at Synthesis Software Technologies
Paul Spagnoletti, executive head of cloud and security at Synthesis Software Technologies

As the hyperscalers entered the country, South Africa’s cloud spend, unsurprisingly, surged. It also changed user behaviour. “There was this massive uptake, which was really exciting,” says Paul Spagnoletti, executive head of cloud and security at Synthesis Software Technologies. “But for most cloud service providers, last year was tough.” Alongside an economic downturn, a lot of large organisations experienced changes in leadership, which put a pause on projects. “Cloud didn’t stop, but spend was curtailed quite dramatically, making it a hard year for most,” he says. “This year, I’m seeing quite a bit of uptake again in cloud spend and adoption.”

Spagnoletti has seen a definite shift in how organisations move to the cloud. He says that those businesses that have gone through the hype cycle, where moving to the cloud is what everyone is doing, have now developed a sense of maturity around what the hyperscalers offer beyond cost and scalability. “The days of having a data conversation and a [separate] cloud conversation are gone – they’re synonymous with each other and they intertwine,” he says. “Any service provider that cannot bring those two conversations together or provide that as a service is struggling. Just providing a cloud platform isn’t enough – organisations are wanting value from a data point as well.”

Lift and shift

For Spagnoletti, organisations that do a lift and shift without modernising or optimising their cloud environment end up like deer caught in headlights, and will be faced with bill shock. And while there’s a cloud repatriation trend happening in more mature cloud markets, on-premises in South Africa comes with additional challenges. “From datacentre running costs to power outages, what we’ve seen here, instead, is a drive to public cloud,” he says. “A public cloud environment is the next best option when looking to leverage sophisticated services without on-premises cost [such as buying hardware]. It makes sense when you need some scalability, but not an auto ramp-up, ramp-down scenario.”

Cloud maturity will come when a business can take a step back, look at specific workloads and recognise what works best in what environment. While Spagnoletti can see why businesses would consider public cloud, at Synthesis Software Technologies, he only plays in the world of hyperscalers. “If you look at Amazon or Microsoft or Google, the type and volume of workloads that organisations have trusted them with are in the thousands,” he says. “From a security perspective, there’s no question why you wouldn’t go with those three locally.”

I have conversations with CIOs who still tell me that they don’t see the business case.

Another reason Spagnoletti believes businesses would choose a hyperscaler for cloud has to do with AI. “AI is not new. It’s been around for decades, but they had a problem – they couldn’t run models because they didn’t have enough compute or scalability,” he says. What’s changed the cloud landscape is that hyperscalers provide organisations with the ability to deploy machine learning models at scale, “and it isn’t massively expensive and you can get really rich information out of it,” he adds.

When you look at the hyperscalers, there’s no doubt that they are all focusing on AI. Microsoft, for example, recently opened up access to its AI models to allow countries to build their own AI economies. Google’s machine learning (ML) platform, Vertex AI, once had a limited number of ML models, but expanded its scope to over 100 towards the end of 2023, including Meta’s Code Llama. “AI is a big focal point. When I talk to customers, 90% of the conversations we have include a data component. It’s not even about ML or generative AI; we talk about the cost of traditional databases – they’re incredibly expensive and hyperscalers are offering database freedom at a fraction of the cost,” he says.

Where is the value?

While certain industries – like financial services – have migrated to the cloud at a rapid pace, one of the most interesting things Spagnoletti is finding is that there is still a measure of reluctance when it comes to cloud adoption locally. “I have conversations with CIOs who still tell me that they don’t see the business case,” he adds. A major consumer goods company, for example, recently told him that they don’t have a cloud strategy in place and don’t see the value.

For those companies that are still in the process of migration, another difference Spagnoletti has noticed is that the lift and shift, which primarily happened two to three years ago, has been replaced by moving and modernisation occurring at the same time. “It’s all about acceleration and showing quick time to value. If you can’t, you’re in trouble,” he says.

One of the ways Spagnoletti recommends doing this is by having an overarching cloud adoption strategy that is broken down into incremental strategies per workload – and then having a business case for each one. “That’s powerful,” he says.

Something else that is shifting is that cloud adoption is no longer a case of simply picking one hyperscaler – organisations are beginning to realise that they can pick the right platform for the right application or workload. “It’s bringing a lot of complexity around cost. When you’ve got one horse, it’s a lot easier to manage because each hyperscaler has its own set of native tools, security and cost optimisation,” Spagnoletti adds. “Today, you need the skills to take an organisation on a journey that can leverage all that cloud offers, not just because it’s cool and you believe it’s going to solve all your business problems – cloud is more than that now.”


* Article first published on