The Evolution of Cloud | Greylock

Conversely, software written for cloud-only assumes logical disaggregation of compute, storage, and networking, meaning it can take advantage of the intrinsic elasticity of the cloud. This is why simply moving an open source database or project like Elastic or Kafka to the cloud doesn’t make it a cloud-only service. MongoDB realized this and launched Mongo Atlas, which began with the open source MongoDB project and was further augmented by significant code to make it a true cloud offering.

Likewise, Snowflake separated storage from computing, creating a truly cloud-only program that led to better workflows for developers. Other pioneering companies are using cloud and building new moats such as serverless infrastructures, which pushes data, compute functions, and apps to the edge.

"What needs to be built to fuel the next generation?

  • First, in a world where Big Cloud dominates the landscape, a startup must own a problem space and then own the solution awareness to that problem. In a cloud only world, your competition is not only just a click away but an API call away."
  • Second, cloud-only companies must leverage the underlying advantages of cloud architecture in terms of elasticity and scale.
  • Finally, to be truly enduring, these next generation of companies must build their own moats beyond the walls of the big cloud vendors. This could be extending their reach to the edge or home or building a deep IP moat around themselves.

(emphasis mine)

via David Crow