/ 4 August 2010

Amazon’s cloud has a silver lining

Half a billion dollars. That’s what online retailer Amazon.com will earn from its cloud computing services this year — if you believe UBS Investment Research’s latest report. Not bad for a service that started off renting out Amazon.com’s spare computing capacity.

Even more startling are the profit and growth figures. Nearly half of that revenue is pure profit, and with the service growing at 50% every year, Amazon will be making a cool $2,5-billion from the business unit by 2014 — nearly 4% of its total revenue.

What makes the service so popular? Three things: technical flexibility, a highly granular pay-as-you-go billing model and a powerful self-service interface. Amazon Web Services (AWS) gives any company, from the tiniest start-up to the corporate behemoth, access to Amazon’s vast pool of computing power.

This means that instead of investing great chunks of cash in powerful servers that will sit idle at least some of the time, you can simply rent capacity — literally an hour at a time. It’s like being able to rent a delivery truck by the hour that shrinks (and therefore costs less) as you drop off each batch of goods.

When Amazon launched its commercial cloud offering back in 2006, most analysts were skeptical. What was a retailer doing playing in the IT infrastructure market? But Jeff Bezos, Amazon’s visionary founder, proved once again that he knows the market better than most.

What Bezos and his lieutenants realised was that by creating the world’s biggest ecommerce site they had consequently developed the expertise needed to set up and effectively manage enormous computing nodes (or “datacentres” as they are known). They saw the potential of the market long before many of their (supposedly) more technically inclined peers, and they have been at full steam ever since.

One of Amazon’s first foray into cloud computing services — Mechanical Turk — launched in 2005. Billed as a “marketplace for work”, it lets businesses parcel out chores (typically research or data capture tasks) to willing freelancers.

The service proved a hit, and Amazon was soon hard at work on AWS (much of which was developed in Cape Town by local talent). When it launched in late 2006 it was an instant success and has been growing in size and complexity ever since. It now operates as a separate business with it’s own infrastructure.

Amazon was neither the first, nor is it the only player in the market. The page you’re reading right now was served by the cloud of one of their more cost effective competitors. But while the other players in this new market have deep roots in IT and hosting (Google, IBM, Microsoft, HP), Amazon is the furthest outside of its comfort zone.

All the high falutin talk makes this seem like a totally new paradigm, in many ways it’s just a return to the good old 1960s when mainframes reigned supreme. Computing, like electricity generation and water purification, is most efficient when it’s done at scale. Hence the phrase “utility computing”, which is what they called cloud computing before IT had any cool buzzwords.

What the success of AWS (and its competitors) tells us is that the age of powerful personal computers is finally coming to an end. As utility computing expands and internet speeds increase, there will be little need to pack computers with heavy, expensive components.

And so the computers of the future (cue sci-fi music) will probably look more like iPads — portable and wireless with a great interface. If that’s the case, the future can’t come soon enough for me.

  • In a previous version of this column I implied that Amazon Web Services (AWS) relied on Amazon.com’s spare computing capacity. In fact AWS now has its own infrastructure and has not used Amazon.com’s excess capacity since 2006. We apologise for the error.