Gross margins are obviously an indicator of gross profits.
And that "obviously" isn't what you said in the first place. Gross margins don't have a direct relationship with net profits, as you were trying to imply in the first place.
I can recall dozens of situations of high gross margins companies having losses and low gross margins companies being extremely profitable, but let's drop that subject here, I can't stand see you tapdacing around basic financial metrics.
No you can't. There is absolutely no way that you or anyone else would be able to break down NVIDIA's GPU profits net of expenses by line of business. You would need extremely detailed information on revenues, cost of goods sold, operating expenses, R&D expenses by each line of GPU business.
Just because *you* can't doesn't mean others can't. That's why banks and consulting firms pay millions of dollars each quarter to have access to channel and insider information. See how much it costs to get access to Mercury Research standard data, then try to figure how much would cost a *custom* report of what you want.
Of course you shouldn't have spend much money to have a good idea of where a business sits. By reading the Q&As and the annual reports, and messing a bit with the numbers you can get a very good idea of each product line. That's what most stock brokers do.
Which is very illogical given that Tesla GPU R&D is shared with other lines of business, given that gross profit margins are higher than corporate average, and given that this is no longer a nascent business.
And do you think that the Tesla business is just about getting a Geforce, re-flash the firmware and sell it for 5 times more the MRSP price? No, it isn't. They have to develop specific CUDA libraries and functions, they have to field engineering teams at their biggest customers to help to tune the code for GPU execution, they must send engineers to help OEMs to design products with Tesla, and their customers will demand around the clock support if anything goes wrong. That costs *a lot* of money, it's OPEX, and it is *not* shared with the rest of the GPU lines.