That is a fair point. Intel's ROI on graphics certainly took a hit with the Larrabee fiasco (although they are successfully pivoting that into Xeon Phi).
However I would frame your statement, one that I agree with by the way, a little differently.
My perspective on this topic is that AMD overextended themselves in ways that ultimately led to undermining their strategic vision and efforts. Ultimately they had no business attempting to make the giant leap they made, it wasn't financially sound or prudent for them to take such a large risk versus pursuing risky strategic plans that they could afford in the event it didn't pan out.
That is what Intel does. Yes the absolute dollar volume on their risky projects (McAfee, Itanium, Larrabee) might be larger than AMD's, but at the same time the risk was contained and small enough as to never put the entire livelihood of the company (and its shareholders) at risk of collapsing the way AMD's risk appetite did.
Intel takes risks, sometimes large risks, but doesn't over-reach or over-extend themselves in taking those risks. AMD did the opposite and their shareholders have paid the price ever since.
Very true. Here is AMD's annual R&D expenditures, having peaked in 2008 at $1.85B and since steadily declined to $1.1B in 2014.
^ AMD is now operating at the same R&D budget level they had in 2004, 10 years ago, when all they had to focus those R&D dollars on was the K8 and K10.
Now compare AMD's R&D to Intel's R&D budget:
Intel isn't contracting, which bodes well for their shareholders and their customers.
Intel is large, so we expect that, but perhaps what is even more striking is to look at Nvidia's R&D budget versus AMD's:
Not only has AMD's R&D contracted, but they investing in future products to a lesser extent than even Nvidia