IBM is old enough to understand the pitfalls of that approach.

In year zero it sounds great. Integration! Efficiency! And that works for a bit.

Then internal politics causes you to pass on opportunities that competitors take because taking them yourself would cannibalize sales in one of your other lines of business.

Then somebody else comes up with a better microprocessor than you, or better software, or lower cost manufacturing, and it's going to take your internal team five years to catch up. In the meantime all your systems are integrated with your in-house solution, so you can't switch, but the deficient in-house alternative makes you less competitive, so you lose all the customers who can switch easily.

Then you have to cut R&D because of the lost revenue and fall even further behind the competition, which leaves you in a death spiral where the most profitable strategy is to lock-in your existing customers and milk them as hard as possible as you circle the drain.

Or you can concentrate on doing one thing and doing it well, so that you don't risk sinking your entire operation as soon as you fail to execute flawlessly in any one of the six different major fields you've created an internal dependency on.