Should the Mac Pro pay for itself?
Yes. "Apple should take a substantive hit on profits, just to make a highly narrow few happy. " is a narcissistic, wet dream that won't die on these forums.
The 10K Apple watches did little to pull that product line into volume profitable status. Expensive for expensive stake does alot of nothing to Apple's business model.
Couple of Apple analysts recently said there must have been a right mess up with the Mac Pro - for the millions they invested in it they surely haven't seen that back in sales.
What "analysts" are these. It is relatively easy that Apple made at least $10M on the Mac Pros.
Apple's general profit margin is around 25-30% so let's take 15% as a conservative marker ( other stuff to pay for in general macOS 'overhead'. ). Take an average selling price of $3,200. ( so major dominate sales are only those of the lowest end mode. Again a conservative estimate. ).
Apple sells __________ break even covered.
35K _______ $16.8M
30K _______ $14.4M
If Apple sold just 45K per year for two years that is easily a $29-33M breakeven they could more than cleared. IHMO, Apple's low-end floor on run rate is probably in the 50-60K/year range ( if don't project selling more than that; don't bother. ). It isn't going to help them stay in the 5-7% of the classic PC market range that is the overall strategic target.
There was a backlog for the Mac Pro for 4-5 months. The notion that Apple didn't sell in the 10's of thousands is relatively weak. Sales have probably significantly dropped now in year 3+ but the notion that they couldn't have gotten there in 1-2 years is suspect. [ Apple may be taking a bit of a hit at the moment because of minimal contract pricing on run rates through the contract factory. That could have been limited by just going outside of norm and putting a estimated need into long term inventory.... and then just stopping contract. The notions that Apple is blowing tons on money building product that is going no where is deeply unmotivated. The company doesn't work that way.... at all. I saw a reference in these forums to that being a root cause and it is quite suspect. Additionally, more than kind of lame to blame that cost on the current product because that fault primarily lies in not doing the next iteration on time. ]
Additionally, There is little indicators that the Mac Pro sold in the approximately 100K/yr rate range but it didn't need to. The 2008-2010 run rate was probably not in that range either. So both designs flirting with the "floor" threshold of high enough run rates.
It would have been a drag on overall Mac product line profit margins, but didn't make their money back at all? That is a stretch.
Is it likely they'd plunge as much money in a new Mac Pro vs rejigging the current design?
Apple burned a more than several $100M hole in the ground with that failed sapphire screen plant in AZ. It isn't like they don't have money to survive a "learning exercise". The 'car' thing also probably has another $100+M hole in the ground aspect to it.
IMHO, Apple primary problem right now is not the current design, it is the product management. Hiding in a hole for 3 (or more) years at a time isn't going to cut it if going to adhere to Apple's standard disclosure policies. (not just Mac Pro ... Mac Mini is in the "land of the lost" too). The design needs some tweaks ( a bit more diameter and height to get a better "safety buffer" on thermal envelope. some incremental improvements so that can roll out incremental custom GPU updates without having to much depot/jig adjustments. ). Getting the Mac product line their own industrial design team wouldn't necessarily be a real increase in costs; more so a pull back of a subsidy to iOS devices that they really do not need.