Earlier this year, the venture capital firm Andreessen Horowitz published a post on its website called ‘The Cost of Cloud, a Trillion Dollar Paradox’, which seemed at odds with prevailing opinion. The writers, Sarah Wang and Martin Casado (who are both Andreessen partners, and Casado is one of the inventors of software-defined networking) say cloud is one of the most important platform shifts in computing history, and its extraordinary growth is due to it being able to deliver the exact resources needed by the business, when the business needs it.

So far, so uncontroversial, but then the writers say as this model matures, it’s becoming evident that while cloud delivers on its promise for early stage businesses, this isn’t necessarily true as a company scales, and growth slows. These companies, which will probably be quite large, will of course by now have spent years running parts of their business in the cloud, and may not be focussed on infrastructure optimisation. Many will also not be able to bring off any useful restructuring that will improve efficiencies, or at least not without a great deal of effort.