Now I’m calculating the R & D spending of a wide variety of software.
NVIDIA’s R $ D spending increased from 2005 to 2006, by $ 196 million, from 2005 to 2007, by $ 296 million.
From these figure and the fact that CUDA was released in July 2007, I think it can be said that NVIDIA’s R & D expenditure on the first version CUDA is $ 334 million.
I want to hear your opinions about this figure.
FYI: According to their press release, GE’s R & D spending on Predix is $ 1 billion.
I would humbly suggest that trying to determine such a number is an exercise in futility. Publicly available data just doesn’t provide the numerous details you would need to make a determination.
In general, R&D spending in the semiconductor business tends to fluctuate quite a bit from year to year depending on market conditions, so your methodology of simply looking at year-over-year increases in NVIDIA’s R&D spending and attributing the entire (?) difference to CUDA seems deeply flawed. Keep in mind that NVIDIA is a sizable company that pursues any number of new and existing technologies at any given point in time, so an increase in the yearly R&D spending could be due to any of numerous such activities.
Even such methodology questions aside, I don’t see how you get from the two numbers you mentioned ($196M and $296M) to a projected expenditure for CUDA of $334M.