Twelve months is a long time in cloud computing (just think back on all the new developments 2017 brought us) but I\u2019ll willingly uphold this time-honored tradition by asking my own crystal ball what the future holds. Without further ado, here are my 2018 Cloud Predictions.\nCloud 2.0 goes mainstream\nFor all the incredible innovation occurring in the cloud today, the vast majority of what we do is still basic compute and storage. Yes 16+ years after the creation of the type 1 hypervisor, over 85% of the cloud is really just better virtualization\u2014 that still runs 24x7x365 \u2013 in someone else\u2019s data center.\nThe good news is we're just following the repeated pattern of all disruptive innovations. When confronted with a disruption, consumers initially seek to use it like its predecessor technology. Remember the introduction of digital photography, when we consumed this technology through digital cameras whose form and functions look suspiciously like film-based cameras? Now digital photography is embedded in everything we do, from our smartphones to laptops to our glasses and even to our home assistants ("Alexa, can you take a creepy picture of me?").\nThe same is happening in the cloud, where the true disruptive innovation - what I call Cloud 2.0 - is finally going mainstream. Cloud 2.0 is about platform services, serverless, and composable applications leveraging heterogeneous cloud services that only run when required. It's about our break with three decades of variations of the client-server model that has put persistent servers and operating systems at the foundation of all our applications. While the transformation will take over a decade, 2018 will be the year we reach a tipping point in which 2.0 will finally be accepted as the future of application architecture.\nLet's start the long goodbye to virtual servers. Don't let the door hit you on the way out.\nMachines save the day\nThe pace of change in the cloud is unparalleled. A reasonably sized cloud customer can drive hundreds of thousands of changes to cloud environments every day. Each of these changes has the potential to impact cost, security, performance, availability, and even compliance for your organization.\nWhile infrastructure as code has enabled us to launch all these applications and infrastructure with incredible efficiency, now it's up to the humans to manage them. The good news is we're all doing a great job trying the make all this work. The bad news is: I read the last chapter of the book and the humans lose.\nYes, 2018 will be the first year where machine learning and AI plays its first meaningful role in our ability to manage the cloud complexity we have unleashed. Just like high speed algorithmic trading transformed stock markets, so will high speed algorithmic automation transformation management. We will start to see machine-driven knowledge and automation driving our monitoring, incident management, cost management and configuration management. The end result: reduced costs, higher security, improved SLAs and better performance.\nDon't feel bad: we still need the humans. At least for now.\nGoogle gets real\nI have been accused of being too hard on Google Cloud in the past. It's not that I don't like Google. I admire the company, appreciate their cloud innovation, and really, really don't like being evil. But I've also watched a company that by right should own the cloud be outmaneuvered and outperformed by a rival that started in the cloud later and with fewer leveragable assets. But I predict Google will become the #2 public cloud in 2018, outpacing Amazon and Microsoft in customer growth, and beating Azure in actual cloud revenue.\nNote: by actual cloud revenue I mean revenue related to IaaS and PaaS (hint: that doesn't include Office365 or Windows Server and an assortment of other high growth products that have nothing to do with the cloud). Something is going on in Mountain View these days.\nThe year 2018 is when we will see Google's innovation shine forth, and their enterprise execution achieve critical mass.\nHeterogeneous cloud happens\nA hybrid cloud is when an application spans the public and private cloud. A heterogeneous cloud is when an application spans multiple cloud providers. While the hybrid cloud has been steadily on the rise, the heterogeneous cloud has been nascent for two reasons: (1) the cost of data transfer, and (2) the gravity of data.\nA heterogeneous cloud presumes the diversity of services being offered continues to grow and differentiate, making public cloud providers more different than alike as you look up the stack. To take advantage of the added diversity, new application architectures will evolve to exploit this opportunity. A heterogeneous cloud application might choose to use AWS ECS, AWS DynamoDB, Google BigQuery, Google DNS, AWS S3 and Azure Search. In 2018 the combination of reduced data egress costs, container technology, microservices and serverless will finally result in us seeing the emergence of applications built on the heterogeneous cloud.\nGovernance as code\nThe management of all infrastructure in the cloud exists in feedback loops, in which there is constant optimization required to maintain the optimum cost, availability, performance, security and usage of infrastructure and applications.\nFor example, to maintain the security of an internet-facing web cluster, an organization needs to regularly analyze security, identify opportunities, make changes, and then repeat this process continually at an interval consistent with the business need. Pre-cloud, these "feedback loops" had sufficiently long intervals that often went unnoticed (e.g. quarters, years). But post-cloud, the intervals are steadily trending toward real-time, making it increasingly difficult for organizations to apply people instead of software to this management.\nDue to the rapid change that occurs in a cloud environment, it is no longer possible to manually govern the health of infrastructure and applications. Governing cloud infrastructure today requires software - ideally software that allows an organization to capture and drive management from their internal policies, best practices and reference architectures. In the same way that DevOps found the solution to the impedance mismatch between engineering and ops was through "infrastructure as code", the solution for maintaining governance without sacrificing the agility enabled by the cloud is what we call "governance as code." It involves capturing your best practices, policies and external compliance requirements as rules that can be monitored, reported, and where applicable automated.\nIn 2018 we will see the emergence of roles, processes and technology supporting governance at cloud scale.\nBlockchain transforms the cloud\nThe blockchain will transform everything in the cloud. I mean everything.\nWell, in truth I don't actually believe this, but how can anyone take my 2018 predictions seriously if I don't talk about how blockchain will increase cloud efficiency, reduce costs, solve world hunger, and generally be really, really awesome? Maybe blockchain will solve the problem of consuming 500+ kWh of power to validate a transaction in 2018.\nDo you take me seriously now? Didn't think so.