What about AI?
AI was expected to change the game by providing a true differentiator for the major cloud players. It’s easy to believe that AWS, Azure, and Google Cloud are now as much AI companies as they are infrastructure providers, given their levels of investment and marketing enthusiasm. However, if you step back and examine the actual AI workloads being deployed in production, a pattern emerges. The necessary toolsets and infrastructure—GPU access, scalable data storage, major machine learning frameworks—are not only widespread but are also becoming increasingly similar across all public clouds, whether in the top tier or among the so-called “second tier” providers such as IBM Cloud and Oracle.
Additionally, access to AI is no longer genuinely exclusive. Open source AI solutions and prebuilt platforms can operate anywhere. Smaller public cloud providers, including sovereign clouds tailored to a country’s specific needs, are offering essentially similar AI and ML portfolios. For everyday enterprise use cases—fine-tuning models, running inference at scale, managing data lakes—there’s nothing particularly unique about what the major clouds provide in comparison to their smaller, often less expensive competitors.
Sticker shock
This brings us, inevitably, to cost, a topic no cloud conversation can avoid these days. The promise of “pay only for what you use” was initially a significant driver of public cloud adoption, but enterprises are waking up to a new reality: The larger you grow, the more you pay. Detailed invoices and cost analysis tools from the Big Three resemble tax documents—complicated, opaque, and often alarming. As organizations scale, cloud bills can quickly spiral out of control, blindsiding even the most prepared finance teams.