All the fascination about terminal hardware applications will be over in the near future. The "Cloud" and SAS will rock the hardware and software world and make access to technology easier for vast populations. Devices to do so will cost pennies on the current dollar or they will be free.
Like the PC makers, the sun is already setting on cell phone devices, associated applications, OTS packaged software and related products. Even though these products are enjoying current popularity They are expensive and will be rapidly overtaken by tight economics and services competition.
Smart,strategic planners are pointing to the future and it is not a hardware and licensed software market - it is service oriented with low cost access and rates. Volume, free products, advertising and shareware will drive it all.
Possible exceptions for a bit longer period of time are the high-end hardware and software technologies in government contracting, which for security reasons must be cloistered, protected and safeguarded. Your friendly government agency will be the last to boot its PC out the window.
How critical cloud computing is to *any* business - startups or otherwise - depends entirely on the business itself - on its needs and goals and on its policies and strategies. It is far from certain that all businesses need cloud services.
"Cloud" technologies are much misunderstood, much misrepresented and poorly understood even amongst those who work in IT: amongst the issues which are now poorly presented are
a] there's no such thing as "the cloud" - there are many many different cloud-like deployments of systems and services; each offering different levels and types of service. Some are entirely private, some are entirely public; and others are a mix of the two. Some clouds are entirely on-premises; some clouds are remote, and some may be a mix of the two. There is no "one-size fits all" cloud deployment. There is no cloud. There are merely collections of distributed services which are *described* as being a cloud - or as something or other as a service.
b] moving to a cloud deployment is not significantly different to deploying any other fail-safe high-resilience deployment of technology. The difference is that one has moved the complexity further away - outside one's direct control, and increased the fragility and the number of dependencies unless suitable risk and impact analysis has been done prior to the design/deployment phase - and done to appropriate standards of due diligence.
c] the reduction of costs is largely an illusion; we ourselves may see reduced capex/opex costs, but meanwhile the energy costs and Carbon footprints of the global data warehousing and cloud industry and all of its NOCs have spiralled exponentially so that they now significantly exceed those of all the worlds air traffic and are well on track to exceed those of air and road transport combined by circa 2020. Cloud doesn't reduce those wider social and environmental "costs" it merely moves them elsewhere - out of our sight - leaving "us" with the illusion that we have reduced our capex/opex.
d] moving to a cloud deployment is all very well but it increases a critical risk which has been with us since the dawn of the internet; the wire limit - how much data one can move between locations in a given time period. We are now creating [and using] data at a rate that vastly exceeds our capacity to move it.
We are also creating a gigantic single point of failure for all businesses which make themselves entirely dependent on the cloud; if all their comms fail so does the business. If their data movement time exceeds their risk recovery window then the business fails.
Cloud technology can be very useful; but only when all parties involved truly understand its risks and its rewards. Startups need to make informed choices when determining how critical a cloud deployment may [or may not] be for them. Appearances are often deceiving.
You can easily find out what a cloud network can do for your business ... by requesting a comparison of available providers including free quotes here: