I think he's suggesting that there are two kinds of profit. The short-term, next-quarter-results kind which fuel wall street, and the longer play when money is invested now and achieves a payback later.
Unfortunate when the word "profit" appears in text, it is often unclear which one is being referred to.
The easiest way to maximize short-term profit is to simply fire all r&d staff. Reduce expendses, sell exiting products and profits for that quarter will go up by plenty. That'll work for 2 quarters, maybe even a year, before the whole thing implodes.
Obviously we don't do that - we forgoe some of the immediate profit to gain profit in the future. But there's no guarantee that what r&d are working on will ever pay off. There's risk involved.
There are also solid reasons for doing one product, which will never make profit, in order to complement another product which will. That complement might be direct, or indirect.
Taking Tim's example of making sure apple devices can be used by blind people, that may indeed be profitable if one considers both the blind people who buy your product, and also the people who _know_ blind people who made their decision based on your blind-friendliness. This is of course impossible to measure.
And therein lies the root problem when it comes to "only doing things to maximize profits". It's mpossible to measure the actual profitability, over both short and long term profits, of pretty much anything we do. There are slimply too many unknowns.
Had Tim answered the question another way, would that impact profitability? If the question had not been asked, would that improve profitability? Given that the question was asked, and answered the way it was, does Apple gain some goodwill? Does that mean that by soundly rejecting the question it accomplished what the question asked for?
Profit is impossible to optimize for. To make profit you encounter risk. And by definition some risky actions will ultimately be unprofitable. Thus the line about companies responsibility being just to make profits, and the promise the Question asked for, is not meaningful in any practical way.