The Future of Cloud Computing: Why Some Firms Are Keeping Their Feet on the Ground

There's no denying the momentum behind cloud computing. In a span of a year, spending on external cloud services shot up from under $100bn to almost $230bn. Titans of the industry like Amazon Web Services (AWS), Google Cloud Platform, and Microsoft Azure have seen their revenues surge by over 30% annually. With these giants rolling out advanced artificial intelligence tools, it's tempting to believe the traditional on-premises company data center is fading.

But the reality paints a nuanced picture. Despite cloud budgets surpassing in-house data centre spending a few years back, companies are still pouring billions into their own infrastructures. Why? For many, especially in the industrial sector, on-premises computing offers unique advantages. A significant chunk of data from connected factories is expected to remain on-site, despite the allure of the cloud.

While cloud services offer convenience and cost benefits, there's a catch. Transferring data to distant data centers and back can introduce delays. In many industries, like manufacturing, immediate data processing is crucial. When creating virtual replicas of factories or developing new products using real-time data, delays and inconsistencies - still frequent in the cloud - aren't acceptable.

Manufacturers like Volkswagen, Caterpillar, and Fanuc have adopted a balanced approach. They maintain crucial real-time data on-site and delegate less critical data to the cloud giants. As data sovereignty concerns grow, housing a data centre close to production sites can be a strategic move, especially in regions with stringent data regulations, like China and India.

Companies choosing to sidestep complete reliance on cloud giants have a buffet of options. From building their own centers, large or petite, to renting servers and infrastructure, the possibilities are diverse. Take, for example, modular data centers from Vertiv or Schneider Electric, which can be conveniently placed near industrial hubs and connected via 5G networks.

Cost calculations for whether to build or rent are intricate. Factors like rising prices for power, land, and labour, combined with construction delays and the high cost of AI-enabled servers, add complexity. It's noteworthy that AI servers can cost up to 30 times more than regular ones, and with cloud giants like AWS monopolizing AI chip purchases, others might find themselves at a disadvantage.

But, the cloud powerhouses are evolving, too. They're launching data centers in diverse locations, from Saudi Arabia to Thailand, and are even venturing into prefab data centers. Notably, Toyota's software arm in the US uses AWS's compact Outpost prefabs, and the Pentagon has opted for larger AWS equipment.

Still, many companies remain cautious about going all-in on the cloud. In this rapidly changing landscape, businesses are weighing their options carefully, understanding that sometimes, it's wise to keep both their head in the clouds and their feet firmly on the ground.