"Conglomerate" is a latinate word for "put together", which is to say that it doesn't immediately refer to the idea of a parent company with controlling interests in several subsidiaries over different markets so you're bound to pick up different uses of the same word. For instance, "conglomerate" is also a technical term in geology. Just check out the hits from the 19th century.
Also, I'm wary of using n-grams to infer anything about the relevance of concept because there are such things as intellectual fads, and some expressions might suddenly hit public consciousness and later fade away without the idea it referred to ever becoming any less relevant. MacLuhan's media theory, for instance, became an intellectual fad in the 1970s and MacLuhan hit celebrity status. Later on, it faded away, but I don't think what he was describing became any less relevant. This kind of stuff is noticeable to anyone, though they might be hard pressed for examples: just try and notice how the buzzwords in your field changed over time.
The expression "multinational corporation", as well as the word "globalization", were also fads of the 1970s too. At the time, they were more like academic fads: researchers started taking notice of these things and suddenly they became "a thing" among academics as it was a frontier research topic. Later on they hit public massively, probably peaking around the late 1990s and early 2000s with the 1999 Seattle riots, the FTAA debates and just the overall shift towards massive multilateral agreements coupled with shrinking governments. Nowadays neither of the two is as "in" as it was back them either in political or academic debate, but the concepts themselves aren't any less relevant. Or at least that's my two cents on the whole issue.
If you look up Sorkin's earlier articles, he'd also posted recently on some failed / dismantled conglomerates -- the use here quite probably refers to that, which his regular readers might get.
There's a lot of invective in this HN thread that is directly addressed (Berkshire-Hathaway, etc.) within the article itself. This suggests things.
I'm well aware that ngrams aren't absolute proof and that language itself changes (quite particularly in business and finance). But it is a datapoint to be considered. Answering it from ignorance isn't a particularly strong counter.
Also, I'm wary of using n-grams to infer anything about the relevance of concept because there are such things as intellectual fads, and some expressions might suddenly hit public consciousness and later fade away without the idea it referred to ever becoming any less relevant. MacLuhan's media theory, for instance, became an intellectual fad in the 1970s and MacLuhan hit celebrity status. Later on, it faded away, but I don't think what he was describing became any less relevant. This kind of stuff is noticeable to anyone, though they might be hard pressed for examples: just try and notice how the buzzwords in your field changed over time.
The expression "multinational corporation", as well as the word "globalization", were also fads of the 1970s too. At the time, they were more like academic fads: researchers started taking notice of these things and suddenly they became "a thing" among academics as it was a frontier research topic. Later on they hit public massively, probably peaking around the late 1990s and early 2000s with the 1999 Seattle riots, the FTAA debates and just the overall shift towards massive multilateral agreements coupled with shrinking governments. Nowadays neither of the two is as "in" as it was back them either in political or academic debate, but the concepts themselves aren't any less relevant. Or at least that's my two cents on the whole issue.