Sky darkens on cloud computing
Are cracks appearing in the seemingly eternal reign of cloud providers? 37Signals, a prominent software company, has made headlines by pulling its entire IT infrastructure out of the cloud, highlighting significant potential savings. This move, coupled with other similar shifts, poses a critical question: could the much-vaunted cloud revolution be undergoing a significant reassessment, prompting more businesses to look back towards their own servers?
Posting on LinkedIn, 37Signals’ chief technology officer David Heinemeier Hansson said that the company would make savings of $1.5 million a year (€890,000) by running its own IT infrastructure.
“Cloud can be a good choice in certain circumstances, but the industry pulled a fast one convincing everyone it’s the only way,” he wrote.
Indeed. Hansson, by bringing 37Signals’ operations back in-house, has not only saved money but also challenged an article of faith in the tech world: that the cloud is both inevitable and optimal.
37Signals is one of the highest profile businesses to ditch the cloud, but it is not alone. Cloud storage provider Dropbox, for one, has significantly reduced its dependency on AWS, meaning it is at least less reliant on subletting space. US insurance giant Geico, owned by Warren Buffett’s investment outfit Berkshire Hathaway, pulled out in 2024, claiming it would make significant savings by building and operating its own infrastructure.
The Geico case is perhaps the most interesting. Rebecca Weekly, Geico’s vice-president of platform and infrastructure engineerin, explained that given the company’s operations are grounded in data, the cost of using the cloud is extraordinary.
“We are a very large storage company, we have a LOT of data (most insurers do)… we do a lot of predictive analytics to understand risk,” she told The Stack.
She’s not wrong. What is interesting is that so many others are.
Outage outrage
Given that, today, it is virtually impossible to see the lips of any politician or business person moving without also hearing a canned speech about data, artificial intelligence (AI) and digital transformation, it is a wonder that more businesses aren’t doing the maths on cloud expenditures.
It’s not just about cost, though. There are even deeper reasons to build capability in-house.
When businesses move to the cloud, they surrender direct control over their infrastructure. This might seem like a reasonable trade-off for smaller operations, but for enterprises whose competitive advantage depends on their technical capabilities, this loss of control can be catastrophic.
Cloud outages are a real thing, with each incident affecting thousands of businesses simultaneously. This is also true of on-premise infrastructure, of course, but when AWS, Azure, or Google Cloud goes down, there’s nothing their customers can do other than wait – and watch as their own customers grow frustrated.
This centralisation creates systemic risk. As more businesses depend on the same handful of cloud providers, we’re building an increasingly fragile digital economy where a single point of failure can cascade through countless unrelated services.
Perhaps more concerning is the institutional knowledge drain that occurs when companies outsource their infrastructure needs. These skills aren’t interchangeable. Companies that maintain their own infrastructure preserve critical expertise that, once lost, is difficult to rebuild.
It is true to say that the cloud is important, and reinforces a view long held by late, lamented network and distributed computing pioneers such as Sun and DEC. Clearly, then, there is more to computing than a bunch of PCs strung together on a network, just as, indeed, there is more to computing than information technology. Nevertheless, we should be slow to abandon the great leap forward that the microcomputer helped us to make in the late 1970s: the right to use our own machines as we see fit.
Subscribers 0
Fans 0
Followers 0
Followers