Friday, May 02, 2014

Why Microsoft should buy Canonical

When Satya Nadella became the CEO of MSFT there was oodles of advice for what he should do - the majority of which I consider extremely poor. Alas during today's lunchtime conversation with a financial analyst, the topic came up and my friend announced that he thought MSFT should buy Red Hat. [Apparently a view he shares with Barb Darrow on GigaOm which is worth reading to get a different perspective]

I tried to resist but my friend needled the point. My response was that there is a reason why he plays a poor game of chess, he rarely looks ahead and fails to understand positioning. This is why he thinks that RHT would be a good acquisition whereas the far better play would be to buy Canonical.

This was followed by the usual guffaws and references to open source and Wall Street articles. So, I explained my reasoning and thought I'd write down my opinions here on how I see the world.

First, I'm not saying that Microsoft will do this and Satya would have had a game plan before he took over. However, I find the topic interesting because in any gameplay it's important to understand the landscape and the potential plays.

The Landscape.
Red Hat has done extremely well in the past. It's a large organisation with a handsome revenue but its problem is the majority of this is associated with past models of working and naturally past models create inertia to change. Whilst Red Hat might be the giant in the linux server world, it's a minnow compared to Ubuntu in the public cloud computing space. The acqui-hire of CentOS seemed more about buying Red Hat relevance (RHEL was probably in the 3-5% MaSh space) in the cloud but even with CentOS the combined efforts are still a fraction of Ubuntu. You can argue over just how much dominance Ubuntu has, whether it's 60% or 70% of the public space is irrelevant. What matters is Ubuntu is by far the most commonly used guest operating system on Amazon and I would be surprised if this was not the case for Linux on Azure. 

Yes, it's true that Red Hat has cried 'Charge!' and stormed into the valley with an all out commitment to OpenStack. But there are several dangers here, not least of which is that OpenStack is focused on a private cloud market (a likely transitional space), that OpenStack suffers from a collective prisoner dilemma and has followed what I consider a daft policy of differentiation rather than co-option. Red Hat has also acquired Ceph (at what seems to be an extremely high multiplier and btw congrats to Ceph) and Gluster but as with the acqui-hire of CentOS then in my view of the world, these aren't great strategic moves but instead smack of desperation for relevance in the future. Even OpenShift looks a fundamentally flawed effort compared to Cloud Foundry - a system which Red Hat should have co-opted. 

On the other hand whilst Ubuntu has a growing OpenStack business it also has many paths to future success from dominance in public cloud use, to Ubuntu phone to its relationship with Cloud Foundry.

When it comes to positioning, then Canonical (a small and nimble organisation) occupies the future space and Red Hat (a large organisation) dominates the past. How Canonical went from minnow on the server to dominating the cloud whilst Red Hat went from dominating the server to minnow on the cloud is testament to the level of strategic play between the companies. To put it bluntly, IMHO the Canonical execs outclass the Red Hat execs at every turn. Whilst Red Hat might have great engineers, their execs can't play chess for toffee in my opinion. Having once run strategy at Canonical, I can say this from experience as well. For almost two years, I watched Red Hat fail to react as we positioned Ubuntu in the cloud.

Buying Red Hat seems likely an expensive proposition at any price because you're buying into a lot of legacy use (servers) with poor future use (cloud) and a large organisational structure (with all the liabilities that incurs). If I wanted to acquire the engineering talent at Red Hat (which is considerable) then I'd swoop in and hire them with large cheque books. I'd simply gut Red Hat for talent and leave the less savoury cuts behind. Buying Canonical however gives you future positioning and there's a lot that can be done with this.

The Play
Let us assume that you're the CEO of MSFT and you've just bought Canonical. What could you do with this? Loads.

First, the Nokia (now MSFT) Android phone is in my opinion a mistake and possibly a last gasp politically driven effort by Nokia to show its innovative prowess. At one point I would have said it made sense but this should now be reversed. There is a far better way of playing the game. 

Whilst mothballing the Android phone, I'd launch a Nokia phone built on Ubuntu. There is already a telecoms / carrier group building around Ubuntu and lets face it, the Android market is dominated by Samsung. There are many others who want a more level playing field and to get in on the action.  By not having an Android phone then MSFT can still exert its Android tax which becomes difficult if you provide an Android phone due to IP clauses in the license agreements. Along with exerting the tax, it could encourage the formation of a market around an open source Ubuntu phone and provide a real and open alternative to Android. There's a lot that can be done here but the obvious question is why not Tizen or Mozilla? Well, first Tizen appears more about buyer / supplier games between Samsung and Google. Secondly, Ubuntu provides leverage which Mozilla does not.

How do I mean leverage? Well Google wants in on the cloud market, hence kicking off the price war with Amazon. But Google came out with RHEL and bizarrely there was little noise about the lack of Ubuntu support though this noise is slowly growing. This is bizarre because Ubuntu dominates AWS guest operating usage and you'd expect many of these would be users of GCE and hence to kick up a fuss. So what does this say? Well, it probably indicates there's not a lot of volume around GCE. 

Whether Google likes it or not it will probably need to support Ubuntu to make a dent in this space, you can't simply ignore the dominant operating system for long.

However, if Google ends up needing Ubuntu to play its cloud game against Amazon (where the majority of guest operating system is Ubuntu) and if Microsoft owns Canonical then it's going to be more difficult for Google to counter Microsoft providing an Ubuntu phone whilst MSFT exerts an Android tax. Ubuntu provides Microsoft with leverage. If Microsoft owned Red Hat then there's little or no leverage as Google can simply adopt Ubuntu.

But what of Apple? Well, what of it? Apple needs Microsoft Office on its iPads to compete against Android tablets. Apple needs to promote itself as the more superior 'enterprise class' environment. It'll be tough for Apple to launch a patent attack against a Microsoft Ubuntu phone if Apple needs Microsoft Office on iPads to compete against Android tablets and in any case it'll be in Apple's interest for an Ubuntu phone to undermine Android's growing dominance in the space.

So, how about Amazon? Well, the most popular OS on Amazon is Ubuntu and Amazon doesn't mind competing. In fact, whilst Amazon will want a big piece of the pie, it won't want all the pie because of monopoly commission issues. So if Microsoft buys Canonical and launches a Microsoft Ubuntu phone then it can still exert the Android tax, it can provide a way for carriers to compete against Samsung's dominance in the Android market, it can leverage Ubuntu's necessity for Google in its GCE race against Amazon, it can leverage Apple's need for Microsoft Office and Amazon EC2 users need for Ubuntu and hence MSFT can muscle itself a more respectable and powerful position in cloud. 

But there's more.

Ubuntu is supported on Azure but Ubuntu is also tight with Cloud Foundry. IBM and others seem to have realised that they've lost the IaaS game for now and need to build a market in PaaS and play substitution games to build their own positions. Cloud Foundry is the route for this game, it has strong strategic leadership in James Watters and if Ubuntu is supported on Azure, Amazon and eventually GCE then it makes the process of building Cloud Foundry environments across all of these that bit more easier - especially with technology like Ubuntu's Juju.

IBM can mix in its own IaaS effort and buy itself a space at the table with PaaS. The same goes for Dell, HP and others but Microsoft will gain influence in this space. It doesn't have to rely on the success of Azure's own platform play because Ubuntu's relationship with Cloud Foundry (the only viable route for these other companies) buys MSFT some positional power.

But what of OpenStack, isn't that going to change the world? Well, with a differentiation ploy, a focus on private (a transitional market) and collective prisoner dilemma then it has problems. Of course, things can change.

For example, what if Ericsson and Reliance Technologies get together with Canonical and some other players and fork the codebase, create a benevolent dictatorship, co-opt Amazon's APIs and with large enough wallets enable service providers to create a competitive market around an AWS clone (which is what Openstack was originally supposed to do). They could counteract Amazon's ecosystem advantage to some extent to provide an opportunity for service providers to survive. 

But wouldn't Amazon react? Well, Amazon likes to compete, APIs are principles and not copyrightable and it would be difficult for Amazon to take on such an effort if Canonical is involved given the dominance of Ubuntu on Amazon. In almost all scenarios, Amazon is more likely to settle for owning a huge bit of the pie (equate to $30 billion in annual revenues) and live with a competitive market rather than face potential monopoly issues. By owning Canonical, Microsoft can influence this whole game.

At the IaaS layer you'd end up with Amazon vs Google vs Azure with a market of clones around an AWS compatible OpenStack driven by Ericsson and Reliance Technologies. This would naturally push the market to consolidate on AWS APIs which after all is what almost every major company I've spoken to wants  - a market of AWS clones. 

At the platform layer, Cloud Foundry runs across all these environments and would enable IBM, HP and Dell to play a competitive platform game with substitution plays.

In the phone space, the telcos and carriers would get an alternative to playing second fiddle to Android / Samsung - a Microsoft Ubuntu phone. Google will need Ubuntu to be a player against Amazon and Apple will need Office to play against Android tablets. With the ability to exert an Android tax, provision of Ubuntu and Office then Microsoft gets to create highly competitive and open markets in all these spaces and take a chunk of the pie.

By owning Canonical then Microsoft gets influence and leverage across all these future fields. Yes, it would mean massive transformations within MSFT and learning how to use open source as a competitive weapon but MSFT is changing in any case. By owning Red Hat then Microsoft gets a chunk of the past.

For me, it's the obvious play but then I don't work as a financial analyst ... I do however play a mean game of chess.

--- Added 4th May 2014

I wrote the above post in response to the question 'Should MSFT buy RedHat?' for which I strongly view that Canonical is the better option for reasons of positioning and leverage. However, I've subsequently been asked whether 'MSFT buying Canonical is the best of all options?'.

My own view is that the best tactical choice for MSFT is probably not to buy either company but instead invest heavily in Canonical. The reason for this is to do with issues of stewardship and community response. Hence if I was running the game at MSFT, I'd be looking to invest $2-3 billion in Canonical in return for a 20-25% stake in the company. From a tactical stand point, I'd be wanting to demonstrate commitment and involvement in Open Source prior to any all out acquisition. However, that wasn't the purpose of this post. If you're going to buy one then buy the future ... which means Canonical.