Is this Intel hubris, or do they know something we don't?
Everyone gets this wrong.
I blame crappy financial journalism for the widespread belief that stock buybacks are supposed to increase the price of the company.
According to financial theory, a stock buyback destroys cash on hand and outstanding stock at the same time. This reduces the value of a company by EXACTLY AS MUCH as it reduces the outstanding stock. Therefore the stock price should remain unchanged to first order effects.
Therefore a stock buyback in theory is just a way of returning money to investors with different tax consequences than a dividend.
So all that you should read from "stock buyback" is, "excess cash is available to distribute to the owners".
In this case excess cash is being raised by taking on debt. But the Modigliani–Miller theorem also says that the structure of financing of a company has no correlation with its performance, so that shift should not matter much.
"Miller and Modigliani (1963) and Miller (1977) addressed the issue more specifically, showing that under some conditions, the optimal capital structure can be complete debt finance due to the preferential treatment of debt relative to equity in a tax code. For example, in the U.S. interest payments on debt are excluded from corporate taxes. As a consequence, substituting debt for equity generates a surplus by reducing firm tax payments to the government."
http://www.econ.uiuc.edu/~avillami/course-files/PalgraveRev_...
Long: To say that a firm's equity is "underpriced" is to say that the firm believes it is expensive (risk-adjusted), compared to it's debt.
This implies a two things:
* Asymmetric information - that the firm knows something the market doesn't.
* That the firm is at a sub-optimal weighted average cost of capital, and can reduce the cost of its capital by restructuring.
As found by MR Leary at Duke and cited by 600 others[1], firms resolve this situation by rebalancing their capital structure. One way to do this is to issue debt at a low interest rate to buy back the firm's "expensive" stock.
The Modigliani–Miller theorem that you reference does indeed argue that capital structures don't matter and that rebalancing shouldn't need to occur. However the theory doesn't apply in practice since it assumes the absence of asymmetric information, taxes, and the presence of an efficient market [2].
1. http://scholar.google.com/scholar?cluster=179408952930709446...
2. http://en.wikipedia.org/wiki/Modigliani%E2%80%93Miller_theor...
In any case there isn't a lot of ARM assembly out there these days, and everything else can retarget x86 with little more than a recompile.
It seems to me that Intel has traditionally benefited from the fact that CPU architecture has mattered (in x86's favor). The mere fact that x86 is no longer the only sensible choice, while obviously not a doom prophecy for Intel, isn't really an argument for Intel's continued relevance.