AWS lock-in, should we be worried?

Amazon Web Services recently turned 10, and the impact this ground-breaking offering has had on the ability to build, manage, scale, and bring digital products to market cannot be overstated.

It is estimated that AWS runs more than 10 times the compute capacity of its nearest 15 competitors.  That's #1 with a bullet.    By some estimates, AWS is on track for over $6B in revenue in 2016, growing in the neighborhood of 80% YoY. Amazon has extended its dominance in e-commerce (where it is 6-7 times larger than its nearest online competitor : Walmart) to the arena of the cloud - a market it essentially invented.  

And while many view Amazon as a generous and benevolent steward of the state-of-the-art in cloud computing, some customers are developing an increasing sense of anxiety about what it means to put so many eggs in the AWS basket.  Enterprise lock-in is certainly nothing new for IT.  Oracle, IBM, Cisco, Microsoft and many others have reached levels of market dominance that presented once-zealous customers with tough decisions as costs rose, features stagnated, and competitors brought more compelling offerings to market.  

In the first act in this familiar play, a software product rapidly gains market share based on a more compelling feature set,  better pricing structure, or superior service level than its competitors.  Act two often involves customers doubling down on their investment as new proprietary features are unveiled that increase customer "stickiness" (a more marketing-friendly term than "lock-in").  When done well, this stickiness also entrenches customers via "data gravity" - the increasing share of customer data within the product and the rising cost of migrating it off.  Act three happens when a market shift, upstart competitor, technology "disruption", executive shift, or other stimulus throws the relationship into doubt and harried engineering and IT managers begin to assess just how entangled their operations have become in one specific vendor.

AWS, it would seem, is still enjoying a remarkable honeymoon period among engineers and IT execs.   It is hard to argue with incredible breadth of tools, services, and components they offer.  Building a new digital product on something other than Amazon would require some serious justification - the technology is just that good.  And by fully embracing open source and offering a public marketplace of services , Amazon has been able to help reduce the proprietary gag-reflex of IT customers.  While you may hear a few whispers at forums like AWS Reinvent about lock-in, and  a variety of efforts have been launched to reduce risk through standards (OpenStack, OVF, etc.), the reality is that Amazon's market share and staggering growth in cloud services is an irrefutable sign that an overwhelming number of us are "all in" on AWS.

Reflecting on my personal experience, in countless conversations with co-workers and colleagues (across a variety of verticals) over the past 4 years I can safely say I have not heard the words "lock-in" and "AWS" in the same sentence even once.   To be fair, some AWS customers are more carefully using EC2 in a capacity that thoughtfully avoids any of the proprietary services that would make migration difficult.  Products like Dynamo, RedShift, Kinesis, Lambda, Amazon IoT, Elastic Beanstalk, S3 (and more), however, create compelling options for customers in the AWS ecosystem and drive that "stickiness" Amazon executives adore.  It becomes increasingly harder to avoid these compelling products when the setup and integration is so clever and seamless.

It is hard to predict what event will push a critical mass of AWS customers towards migration.   Will the cost structure become too unfriendly to a meaningful segment of the customer base?  Will a disruption in cloud computing happen that Amazon fails to invest in or misses the boat on (a "Microsoft" moment?)  Will Google, Apple, Microsoft, or another major competitor get serious enough about cloud that they become appealing options?

Whatever the event, the Newtonian nature of our industry dictates that the turning point for AWS is not an "if", but a "when".  Are we doing enough to be ready for it?  How - if at all - is your organization balancing this risk?