Despite Serverless has problems and challenges at present, in many use cases, Serverless actually used at the backend. Lack of standards, tooling and vendor lock-in are problems and risks for those who are not the cloud service providers. Rather the mass consumers. However, the cloud service providers of IBM, Microsoft, HP sized use serverless for delivering their services. We simply do not know each and every detail of backend architecture. But, different machine learning and IoT services do appear to involve Serverless. The interest of IBM towards OpenWhisk is not only for raw delivery of Serverless, but they have Watson services and enterprise-grade IoT services. As software consultancy, they will want to deliver maximum scalability and security to the larger clients.
Complex web applications with a significant shelf-life require (disgusting) maintenance and updates for years after their original creation. It is never easy to maintain WordPress without avoiding breaking changes. The upgrades are mostly done for changes in the operating system or for the need of increasing security.
Before the arrival of serverless, to run a function in the cloud, we had only way to execute dynamic function was to create a virtual machine, configure and manage it. PHP was an indispensable tool, simply because it was easy to handle common dynamic works without the need for much training. Serverless removes that requirement. The program becomes a conduit to send API requests. The success of the cloud is the key driver of serverless. Peoples actually are happy with various cloud storage services.
Serverless is likely to hurt middleware, DevOps, and maybe the concept of IDEs. More serverless adoption will invite more APIs in the cloud which in-turn will create new kind of services.
Serverless provides financial benefits via reduced development cost, reduced total cost of ownership. So it provides agility. Technically, started with scalability and high availability. Cloud-based other models, such as PaaS was never built in that way.
Serverless architectures have the potential benefits as a direct consequence of the microservices approach, such as high availability, failure isolation, and better maintainability and flexibility and the concept of decoupled components.
Presently, a lack of tooling with support for serverless development and debugging is a challenge. Latency and vendor lock-in are present concerns but probably not the problems of growth. Basically, the FaaS systems are newer concepts which play with our imagination. We can assume that tools will catch up with the required growth, and serverless will be more common within the next five years.