IBM - the century company

IBM's importance to this catalogue is not that it was large. It is that twice, decades apart, it created a market on the same principle and kept only one of them.

The System/360, announced on 7 April 1964, cost an estimated five billion dollars and was known inside the company as the bet-the-business gamble. What it bought was compatibility: for the first time one architecture spanned the whole range, so software written for one machine ran on the others. Before it, buying a new computer meant discarding your programs and starting again. That is the platform business model, invented here, and the industry has been rebuilding it ever since in search, in commerce and in the cloud.

The consequence was permanent and is stated well by one historian: the effect was so complete that IBM could never again introduce a computer incompatible with its existing ones. Winning that way sets the terms of every later decision - compatibility stopped being a strategy and became an obligation.

Seventeen years later the Personal Computer applied the same openness to a different problem, and the outcome inverted. Built quickly from off-the-shelf components with a published architecture, the 5150 of August 1981 created an enormous market and handed its economics to the supplier of the processor and the supplier of the operating system. The 360 made IBM the platform; the PC made IBM a manufacturer on somebody else's platform. The difference was who owned the parts that could not be substituted.

The networking record follows the same shape. SNA defined how large estates communicated for a generation, and Token Ring was the better-engineered local network by most contemporary measures - deterministic under load, no collisions, predictable latency. It lost to Ethernet, which was cheaper and good enough, and which a company founded by its co-inventor made purchasable as a card. That is the catalogue's most repeated finding, and here it happened to the largest company in the industry.

What IBM did with the century is the part usually skipped: it survived. The services turnaround of the 1990s, the sale of the PC business, and the 2019 acquisition of Red Hat are one continuous decision to stop defending the thing being commoditised and go where the margin moved. Most companies in this section did not manage that once.