‘Matrix Integration’ And Its Effects On The IT Industry

Matrix Integration Analysis Highlights

• ITCandor uses the concept of ‘Matrix Integration’ to describe the change in vendor strategies over the last few years
• The vertical integration stage of the industry was ended by anti-trust cases against IBM and others and a genuine change in behaviour
• Horizontal integration gave customers many advantages in the choice of vendor and cost of IT
• With the new style of supplier strategy, changing supplier will be more difficult and the availability of Open Source software will be less
• Vendors are more likely to take responsibility for their solutions if they supply the whole system
• ITCandor hopes that we are not returning to the dark days of vendor lock-in associated with vertical integration

For a long time I’ve been thinking about the development of large supplier strategies and how they appear to becoming more ‘vertically integrated’ – acquiring new companies, filling out their portfolios and offering customers solutions, as opposed to point products. I wanted to write about it to see if anyone else has views or concerns about these developments. I believe these changes represent a real shift in the industry requiring new thinking and analysis. First of all let me explain what I mean…

1970-1980 Vertical Integration Restricted Choice And Made Users Feel They Were Paying Too Much

When I first entered the industry in the early 1980s it was a time when IBM had been undergoing dramatic internal changes as a result of the successful anti-trust cases against it. In particular the executives I spoke to then got very nervous just at the mention of the words ‘dominance’ or ‘monopoly’. I believe the period before that change was one best described as ‘vertical integration’. A vendor would build complete solutions (typically based around servers, or ‘multi-user computers’, as we called them then). IBM was the market leader with more than 50% of the market and there was no need to break the market into the sub-categories and subjects we do today.
IBM and its competitors didn’t need to distinguish hardware, software and service offerings for customers. Their products were referred to as ‘proprietary’ and engineers would often lock the customer’s data centre door before making modifications to the equipment. Other vendors competed by offering entire solutions, which was expensive for them and restrictive for end-users. To swap suppliers would involve a complete ‘forklift upgrade’ and take many months (if not years) to complete. You may have heard the term ‘BUNCH’, which was an acronym based on the main US competitors Buroughs, Univac, NCR, CDC and Honeywell. There were, of course, regional systems companies such as Siemens, ICL, Olivetti and Bull in Europe and Fujitsu, Hitachi and NEC in Japan. I don’t want to get too bogged down in ‘historicism’ here and there were certainly many examples of companies who would build additional and/or alternative equipment at the time. I wasn’t there at the time, but I’ve heard tell that the shift from IBM’s 360 to 370 architectures caused users a lot of consternation as the two systems were incompatible, leading users to demand easier upgrade roadmaps and migration strategies. The major outcry about vertical integration, however, was that it restricted choice and there was a major suspicion that vendors were overcharging by bundling everything into a single ‘solution’.
In all my years in the industry I have very rarely heard anyone proposing the advantages to customers of vertical integration and legislative bodies such as the EU and US federal authorities continue to regulate and prosecute large vendors where they feel their acquisitions and product strategies restrict user choice. I wonder how the current changes will be seen in years to come and whether they will instill violently negative reactions as before.

1980-2000 – Horizontal Integration Brings Vendor Choice And Open Systems

There’s a basic rule of first movers that states that the original supplier is never overtaken by its competitors, which held true for many years in the next stage of development – a period I call ‘horizontal integration’. It was a telling moment when EMC overtook IBM in the mainframe disk systems market, which demonstrated that the rule wasn’t going to hold fast in the IT market. The rapid takeoff of the IBM PC from 1978 onward demonstrated that a product made up of components manufactured and developed by multiple vendors typified this age. For the first time suppliers such as Intel and Microsoft created massive revenues from other vendors, as opposed to gaining them exclusively from customers. Suppliers would often make complete machines for other suppliers, something we described at the time as Original Equipment Manufacturing (OEM). Even today OEM revenues from manufacturers and first tier channels account for around 7.5% of the total user spend on IT ($255b/$3,395T in my market estimates for 2009).
In the systems market the difficulty and expense of changing from one supplier to another (or often even from changing from one platform to another from the same supplier) led to a massive debate about Open Systems. I don’t know if anyone else remembers the long arguments we used to have about whether Pick was better than Unix – the fact that Unix was clearly going to be the winner, although Pick was thought to be technically better, underlined the fact that ‘better products’ are often beaten by ‘better marketing’ in our industry.

Initially many ‘Open’ products were by no means interoperable – there were many different variations of Unix, for instance and fibre channel cabling was almost always plugged in with proprietary interfaces, making it impossible to mix and match products from different companies. A defining moment for the horizontal integration came when IBM announced the PS/2 in 1987. It attempted to win back control of a market in which Compaq had already overtaken it by incorporating its proprietary Microchannel architecture and promoting the use of its OS/2 operating system, originally co-developed by (but competing against) Microsoft’s MS-DOS, which was called PC-DOS on IBM’s machines. In is interesting that Apple launched its Macintosh computer in 1984 – a proprietary architecture from the outset. The success of Compaq and the PC compatible suppliers also led to a vast expansion of indirect distribution channels, especially in Europe, although we should not forget that minicomputers such as IBM’s System/34, DEC’s Microvax and Olivetti’s Linea Uno also led to an expansion of Value Added Resellers, dealers and agents. I believe that the success of indirect channels is based in large part on the inter-changeability of equipment from one supplier to another. Even today PC companies can come and go without it effecting customer choice dramatically; things are different in the server market, where a supplier’s kit will still need to be maintained by someone long after the supplier itself has gone out of existence. There are probably still submarines with Prime minicomputers in them being maintained by someone.
As horizontal integration evolved, some vendors began to base their ‘Go-To-Market’ strategies on ‘ecosystem’ assumptions rather than detailed planning and investments in channels. There was also an interesting development of Open Source software, where it looked as if users, large and small, would work together with vendors to design ‘free to use’ software – including operating systems (Linux), web servers (Apache), virtual machine monitors (Xen) and databases (MySQL). Here the step away from proprietary products was extreme as Open Source software was owned by the community in which it was developed – you might even argue that it wasn’t ‘owned’ in a conventional way at all. The internet and the worldwide web were also vital in the development of horizontal integration as they made information and software immediately transportable around the globe, and for the first time allowing vendors operate from a single location without needing a physical presence in every country.
For a customer the ideal of horizontal integration is that there should be a choice of multiple offerings and vendors on each of the rows (simplified in Figure 1) – Intel, AMD on the chip level, Microsoft, Linux and Novell on the operating system level, etc.

2000-2020 The Development Of Matrix Integration Changes The IT Industry

Since the turn of the millennium there have been distinct ways in which the industry has turned against the more customer-choice horizontal integration period. In particular:

• Open Source companies have almost all been bought by commercial organisations; while most retain ‘free to use’ versions, they form the basis of money making activities, whether software with advanced features, or service and support activities. In the period preceding this much of the development was taken over by suppliers.
• Vendors have grown their revenues through acquisition, whether to change the nature of their business (Citrix as a Information Access company, EMC as an Information company, etc.) or as a reaction to the market downturn.
• Through the acquisition of Sun, Oracle is now a hardware vendor for the first time; The addition of 3Com to HP’s business makes it even wider in its product coverage; Cisco’s introduction of its Unified Computing System moves it from the network and storage systems layers to compete in the server layer and Apple crossed (highly successfully) from IT to C with its introduction of the iPhone just three years ago. Exit strategies for smaller IT suppliers are now based on how their specialisation could enhance a major vendor’s integration aims, rather than on the value of its offerings to customers.
• I believe that IBM’s services approach to the market – begun some years ago – has led it to a different approach from the extremes of dis-aggregation of the 1990s. I remember well attending a two-day IBM software conference in which the company failed to mention a single IBM hardware platform. Especially with the addition of PWC to its Global Business Service group IBM has the ability to sell solutions without the need to unbundle the component parts.
• Google aspires to be everything to everybody and has a business based on advertising revenues. In a sense it is a natural extension of the Open Source movement in offering ‘free at the point of use’ services and software’. However it is constantly in the news for the ways in which it seeks to exploit the information it gathers as a result.

The beginning of the change came at the start of the last recession in 2001 with the growth of data centre (and other) outsourcing. For many users the high and unpredictable costs of IT when owned and run internally led them to hand their resources lock stock and barrel to vendors, who obviously run vertically integrated services as a result.
I’ve decided to name the new type of vendor make-up as ‘Matrix Integration’ for a number of reasons. In particular:

• The underlying standardisation of components and infrastructure remains largely the same. Component suppliers such as Intel and AMD may ‘go back under the covers’ as a result, but nobody is proposing to design and manufacture their own variants on widely accepted devices, techniques and software.
• The Internet will remain a massive influence on the IT market in the future and allow smaller companies the ability to compete with larger ones as a result. Smaller companies will continue to operate in disaggregated corners of the industry at first.
• To call the new stage ‘vertical integration’ would be too negative in the connotations it has with proprietary systems, high cost solutions and the difficulty to change from one supplier to another. As an optimist I hope that vendors will ‘do the right thing’ to avoid such mistakes and, if not, that government legislators will continue to be vigilant

I’ve summarised my thoughts on the pros and cons of the three different types of vendor integration in Figure 3 below.

Some Conclusions – The Pros And Cons Of Matrix Integration

I make no apologies for introducing a discussion about vendor strategies and the changes in level of integration. I know I’ve been talking about a period of 60 years, but there are important issues to think about, whether it’s the impossibility of playing an Apple download on any mobile device other than one made by Apple, or the fate of MySQL in the hands of Oracle, or Google and privacy issues. I have no doubts that the change to Matrix Integration will have a deep effect on channels, user choice and the industry as a whole. Last time the change to horizontal industry proved difficult – if not impossible – to resist and those companies that held out (as perhaps IBM did with the PS/2) got their fingers burnt. I believe that Sun’s demise was at least in part due to its over dependence on the horizontal market – Jonathan’s Schwartz insistence that you could make money by giving software away and getting customers to pay you for service and support was badly executed by his company, but is highly unlikely to be a strategy deployed now its products are owned by Oracle.
One result of Matrix Integration is that it is less likely that start-up companies will make rapid progress independently of major suppliers – they are more likely to be bought or stopped by competitive actions; another is that customers’ choice of vendors and solutions will become more restricted over time. I certainly hope that we won’t reach the levels of consternation and conflict I remember from the 1980s. Finally we can’t complain too much about Apple – its products have always been proprietary, which was a great asset to many users as it also managed to be more technically advanced than the PC market. The challenge of the new market is for vendors to ‘innovate within a standard’, but to offer users genuinely different experiences than those of their competitors.
Are you interested in trying to assess the changes in vendor strategies? Do you disagree with my analysis? As always please let me know by commenting on this article.