Over the past few years AMD has been slowly cutting back on the money they put towards R&D and not always by choice. As their profits have declined (and debt grown) they have tried to narrow the focus of the money they spend on new technology. This does not mean they are not working on new things, it simply means they are focusing on key areas where they feel they will get the most return on investment. In the last quarter AMD spent a paltry $238 Million on R&D. This is a drop in the bucket for a company that has as many products as AMD. However, AMD has also been working to build R&D partnerships like they did in the K6 days to help them overcome budget challenges.
By contrast Intel and NVIDIA tend to keep R&D in house so that the technology they develop is theirs to control. In the case of Intel they spend more money on R&D than NVIDIA and AMD combined. Of course Intel also has a lot more money in the bank than AMD (and NVIDIA) does so they can afford to spend a lot more (around $2 billion). By comparison NVIDIA spent around $348 Million with a much smaller product line up to worry about. AMD has a product line that is comparable to Intel, but they are spending drastically less.
Yup, that is right. AMD started rebuilding their partnerships back in late 2010 to allow them to cut back on the amount they need to bring to the table to create new technologies. In the 1990s AMD had big companies like, Samsung, IBM, Motorola, and Texas Instruments helping them to change the way they built CPUs. It was an advantage that allowed them to build the Athlon CPU with only a small R&D budget and engineering team. Now they have some of the same partners helping them in the CPU world (HSA etc.) while they dump the majority of their money into new memory and GPU tech. Even in the memory world (HBM) they are only a small partner with Hynix so that they reduce costs. It is a smart strategy, if you know that you can win big on the other end of the game.
This is the problem… AMD is betting big on HBM and also on integrating ARM processors inside their APU/CPUs. They are letting their R&D partners do a lot of the heavy lifting money-wise while they provide many of the engineering minds. If the bet fails, AMD is left with little to show while companies like Hynix, Samsung, Toshiba etc. walk away with new technology to sell to others.
So if we are reading the signs right, AMD is shifting their R&D money to focus on the GPU (and APU) side of the business while their partners help the offset the rest. As we said, it is a gamble and one that could leave AMD worse than they are now if it does not pay off. Sadly, for AMD and their fans, it is one that AMD must make to try and get back on their feet.