Don't want to mess up your economics exam so read this later, this is for you to think about afterwards, because this will not apply to your exam unless you're doing senior university or grad work, and have already focused exclusively on oligopolies-monopolies or technology economics. So just a heads'up if you're studying I don't want to confuse you.
The whole thing burns down to this: to make optimal use of their situation (highest profits), oligopolistic formations produce less goods at a given price; in pure theory, a monopoly produces 1/2, a Cournot-type duopoly 2/3 and a Stackelberg-type 3/4 of a perfectly competitive market.
While this is not exactly so in real life, as there are many other specific factors to consider, but they do produce less.
Keep your two cents for funding your economics studies, or at least your grammar correction
First of all you're missing his point, these are not standardized widgets with a long market lifespan and no development cost, nor even easily interchangeable (otherwise who would care A vs B) in which case your examples would apply, the economic modeling you're describing is very simplistic and is better suited to easy recovery raw materials or very basic goods and services and great ease of entry/exit from the market. The models you use exclude exogenous changes, whereas this market is all about incorporating exogenous change as a strategy, and heavily influencing endogenous factors (hence the relationships with developers, M$, etc). It just doesn't translate well from strict static model at all, especially since your argument is more concerned with resource allocation than innovation. You're not wrong, just applying it incorrectly to try and refute his statements.
His comment is regarding the development of the features and technology not reaching a lower price equilibrium (which sounds like a good goal, but who want a $50 Riva/Rage (down from $100) versus a $50 GF7300/X1300 or $100 GF7600GX/X1600P). In theory what you're saying is correct if the market was driven by little/no differentiation and used price alone as a determinant.
For the graphics industry features are as important as price at at the two extremes almost the only factor, and there is alot of differentiation not only between companies but between products within the companies, and even their application/use.
Companies in a perfectly competitive market involving multiple players all reaching E with the market reaching E as well, this means limited profitability (ie no 'excess profits'), but for something like technology companies this also means two things, a) not enough accumulated wealth to fund extremely expensive R&D that does not usually enter the equation until 2+years into the future; and b) the inability to reach economies of scale where the factors of production (TSMC gives discount on bulk and also moves you to the front of the line if you're a big customer), transportation, distribution are reached to lower fixed and variable cost. The later speaks to your focus, the former to his.
Technology is a strange fish economically, and there are lots and LOTS of factors that make the usual models break down without applying them in stages or in completely revolutionary ways. The problems with simply applying the basic models to such a complex situation is the snapshot static nature of the model versus the realistic dynamic nature of the market. Texas instruments almost single handed developed the forward pricing strategy for chips, and this was something that speak well to your model because it involves a standardized widget (basic calculator chip), what would be considered dumping in just ab out any other market was just in time pricing, because the market is so different. However in the case MatT is trying to express (I got it) the benefits of a duopoly give us those advancements, but protect us from the ills of a monopoly, and I would tend to agree with him that a duopoly or small oligopoly works well in this scenario. I see room for the entrance of one or maybe two more major players, but not much more than that if we wish to keep the pace of technology high. In reality there are already 3 big players, but the biggest producer plays the truely 'widget' sales games, and it's maintenance of that role in graphics is a completely unrelated factor which wouldn't fit into any simple model, and would imply that no matter what factors in the graphics industry you include you'll only be able to describe and model partial equilibrium because theinfluence of Intel is huge, and their main reason for success involves their MoBo and CPUs sales, whose influence would be impossible to model correctly, you could only describe the influence at the most basic levels. We can guess that with the more 'effective' GMA965/3000 coming out that Intel will likely only increase their market share from what they currently have, but alot of that increase is bound to be due to the return to Intel from AMD due to Conroe, not because of the graphics industry itself.
Anywhoo, needless to says it's kinda complex and not just selling widgets. And while I understand where you're coming from from a theory standpoint, the market is much more complex than that and really it benefits from less competition than more. It still requires some, because without at least a duopoly there will be little/no motivation to innovate, and even with what many perceived to be a highly competitive duopoly in the discrete market, there was a statement 2 years ago by both of calming the pace of technology and not pushing each other to launch new cards as often, which would amount to collusion and really negate some of the benefits of the limited competition. In the end is all about balance, but not just equilibrium. :mrgreen: