But you'll notice the companies that successfully gain value through merging are typically competitors that are able to generate more value together instead of fighting for the same profits.

The same does not seem to be true for any of the Alphabet companies, except in the sense of being able to leverage shared user data for better targeted ads. Perhaps that does make it worth it to remain one company, but it's debatable, especially for something like Waymo which likely won't benefit from having access to search data.

There are tons of industrial conglomerates, from GE to Samsung to Siemens and there's not really any universal agreement that they're any worse than more focussed companies.

This goes all the way back to the most basic theories of organization - why do companies even exist? Why is everyone not effectively a contractor? Why do companies not outsource every non-core function? And there are good reasons why companies exist - see https://en.wikipedia.org/wiki/Theory_of_the_firm External transaction costs are real and breaking Google up could significantly increase those.

Someone from Siemens once told me "Siemens is an investment bank that happens to own all of the companies it has invested in." I'm not sure whether that's true or not, but it's a good description of conglomerates in general.

So there was this thinking in the 60s that the key skill of a firm is management, and so if one firm was a market leader in manufacturing pipes, they could buy up a struggling auto rental or accountancy firm and with their superior management skills they would make those better as well. This at least was the justification for a wave of conglomerates buying up smaller firms.

Others might say a better justification was the cheap corporate credit available to some (but not all) firms, and thus the competitive advantage of conglomerates was access to credit, rather than management. The smaller firms and individuals did not have the same access to credit.

But in the 80s, the pendulum began to swing the other way. The problem from the perspective of the credit markets -- whether shareholders or bondholders -- was the difficulty in obtaining detailed operational information from these large conglomerates. They became very opaque, as they could use the losses of one firm to subsidize another, and it was hard to drill down and figure out what was happening by looking at the financial statements. So then began a wave of slicing these companies up and selling off the pieces.

Or from a completely different perspective, it was the extension of new types of credit such as junk bonds that allowed insiders to do leveraged buyouts, which tipped the scales away from the conglomerates and led to a lot of asset sales that "unlocked value" while other operations were shutdown.

Not saying who is right, just offering some perspective that these arguments have been engaged with 60 years ago, and ended up with mixed results.

Yes, exactly. The notion that there's clear evidence that breaking up firms unlocks value is simply not true. What is true is that not all firms are well-run and when you change the corporate structure or management style you sometimes make a huge improvement. But it's certainly not universally true.

> "Siemens is an investment bank that happens to own

All conglomerates are like this, the clearest example is Berkshire Hathaway.

Exactly. Why not break up Berkshire?

I vaguely think people have written how having the insurance business attached to the others is a useful/complementary structure, compared to a regular mutual fund or something.

Maybe you could ask from the opposite perspective, why aren't all mutual funds organized like BRK?

Berkshire is for all intents broken up, everything operates independently.