When Google took the wrapper off its Amazon-like Compute Engine service last week, investors in rival cloud companies – including Amazon – panicked.
Shares in Amazon and the OpenStack-backer Rackspace fell as investors feared Mountainview would translate its dominance of internet search and ads to cloud hosting.
There's certainly a new chapter opening in the battle to own the cloud, signified by Amazon, Google and others announcing new features and strategies.
They all want one thing: to rent your data to you by becoming your outsourced data centre provider.
But Amazon isn't the one likely to feel the heat - not any time soon. And this isn't a battle for number one - at least not yet.
Currently, we're looking at a battle for the number two, and it's a battle that'll be waged between Google and Microsoft, based on the customers they can leverage and their level of drive and determination. OpenStack, Rackspace and those pitching the "open cloud" aren't in the race for number two.
Based on size and growth, Amazon's AWS is the one to beat. Its S3 storage at last count in April held 2 trillion objects - that was up from 1.3 trillion five months before 1 trillion 10 months before that. S3 is growing like a hockey stick: it took six years to hit 1 trillion objects, but now it's going straight up.
Meanwhile, Microsoft claimed 8.5 trillion objects in rival Windows Azure cloud in June, but such claims are worth taking with a pinch of salt. They came out after Amazon's statistic and Windows Azure is several years younger than AWS.
Judging by the places where cloud companies are making changes by introducing new features, the target demographic for rapid growth is the business customer.
Business customers have have warehouses full of data, most of which is considered mission-critical to something or other. Once you've got them on your platform, it's difficult for them to move because data is sticky and heavy.
Amazon has been adding more business-class features to AWS for the years and in the last month it added PC virtualisation, app streaming, high-throughput storage, identity management and data stream processing options to EC2.
Not so fast, Google... Since when were you the biz bods' choice?
Now Google is gunning for business customers.
Business customers are a ripe target: they like the idea of outsourcing their servers but in truth want something less Amazon-y, especially the Microsoft shops. Microsoft shops want to re-use their existing skills and investments. Also, they want the kind of security and SLAs that come with Microsoft. Redmond is big in business, so it's a natural fit.
Google's has business customers - just not so many. Its customer base is mostly consumer thanks to internet search and ads, but it has been making a business play. Also, it has been signing up more companies to Google Apps.
If you don't think Google can do this, just remember Amazon was an online book shop before it became a supplier of enterprise IT.
If you don't like Microsoft or Amazon and don't trust Google, there is another option: OpenStack. This sprung to life in 2010 as a project at NASA with Rackspace – one provided the compute while the other contributed storage components. OpenStack was supposed to sow the seeds for an open cloud, with a project like Linux - hence it was described as Linux for the cloud. A thousand OpenStack clouds were supposed to bloom.
Problem is, OpenStack has not found its launchpad moment.
There has been plenty of enthusiasm and developer interest: More than 517 contributors and 230 new features at last count in April. Industry support has poured in from AT&T, IBM, Cisco, Hewlett-Packard, Dell, Juniper, Rackspace and Red Hat.
But we're not seeing the public OpenStack cloud deployments. The biggest we have is HP's Public Cloud service in September 2011, but two years on HP is not talking business, revenue or customers.
And now, even NASA has dumped OpenStack for Amazon, as the consultants who built its OpenSack implementation left long ago to go into consulting and development.
The problem seems to have be OpenStack's customisability: There's too much choice and too much focus on features.
The result has been a handful of OpenStack consultants and specialists working the cloud scene on a number of customer implementations. That's great for the consultants but bad for mass-market, public adoption of OpenStack because it remains too rocket-sciency for the average adopter.
HP faced problems getting OpenStack up and running thanks to the paucity of knowledge and its own lack of internal experience.
Until last week, the battle for biz cloud appeared to be a comfortable two-way fight between Amazon and Microsoft. The former had a huge lead on customers and features, while the latter kept launching new features and price packages to catch up. Amazon would respond with more new features and lower prices.
Google opens up a new front for Amazon. You're unlikely to see Amazon flinch, not yet at least. Not until Microsoft and Google have finished punching each other up. ®