Time and again, enterprises show that they are willing to put up with imperfect technology as long as they can get work done faster.
Why do enterprises so often choose to run applications in the cloud even when it may not be the cheapest option? Why do they turn to open source even if it’s not the most feature-complete choice? And why is generative AI so frothy hot even though, as my InfoWorld colleague David Linthicum argues, “Companies are good at spending money [on AI] but bad at building and deploying AI.”
The answer is speed. As an IT executive friend at a large financial services firm told me recently, it’s expensive to move slowly when market opportunities require fast execution. For him, it’s imperative to build with the cloud’s elastic infrastructure to eliminate the possibility that his company may fail to capitalize on windows of opportunity. This same general principle is driving a number of technology trends, including cloud computing, open source, and generative AI. Each is winning precisely because they help enterprises move faster.
The cost of failure
RedMonk’s Steve O’Grady talks a lot about the power of convenience in driving developer decisions. “For developers convenience trumps most other technology characteristics,” he has written, listing a number of open technologies that owe their early and ongoing success to making life easier for developers: Linux, MongoDB (my employer), Git, MySQL, and more. Each of these took off, to borrow from O’Grady, because of convenience, but also because of speed. Open source is primarily popular because it lowers barriers to using great software, so developers can focus on getting stuff done.
As I’ve noted, “Cloud perfects many of the reasons developers first embraced open source.” With cloud, developers not only get easy, speedy access to code but also to the hardware necessary for running it. As it turns out, this also plays well with enterprise line-of-business owners who are more focused on meeting evolving customer needs than counting pennies. It’s not that budgets don’t matter, it’s just that it doesn’t matter what a service costs to deliver if you’re too late.
That’s how my financial service executive friend sees it. For his company, it’s not an option to deploy their applications to private data centers because they can’t afford delays inherent in scaling private cloud resources. Cost is important but secondary.
Building for success
Nor is his experience atypical. Years ago, following a Gartner analysis of private cloud investments, I noted, “That company-changing app that will make your career? It’s running on AWS. Ditto all the other projects that promise to transform your business and, perhaps, your industry.” The more staid, non-transformational applications have tended to stick with private cloud. One of the primary voices for cloud repatriation, the practice of pulling back from public cloud to return applications to private data centers, is 37signals cofounder David Heinemeier Hansson. He’s spent the past few years trying to convince companies that “renting computers is (mostly) a bad deal for medium-sized companies like ours with stable growth.” That sounds reasonable until you ask, how many companies can realistically plan for predictable growth without any real upside (or downside)? Not many. So it makes sense to optimize for speed with cloud, open source, and now AI.
Just ask developers. AI continues to see more spending than success, to Linthicum’s point, but developers are actively, happily using it to help them build and test code faster. Although not a scientific sampling, Gergely Orosz’s informal survey results are representative of other polls I’ve seen from industry research firms. Most developers increasingly consult genAI tools such as GitHub Copilot to get answers to programming questions. The reason? Speed, convenience, or whatever you choose to call it.
Nor is this just a developer thing. As Eric Colson writes of enterprise misuse of data scientists, “The untapped potential of data scientists lies not in their ability to execute requirements or requests but in their ideas for transforming a business.” Too many enterprises constrain their data scientists to tactical analyses or operational dashboards, he says, which is non-differentiating and not likely to require or drive greater operational speed.
This brings us back to the original point. If an enterprise wants to play it safe, they’ll eschew open source and AI and stick to the comfortable routines of their private cloud infrastructure. That isn’t a recipe for winning. Enterprises that want to win will embrace those things that enable their developers to move more quickly.
