The point you are missing though, is that ASIC's technology has been around since before bitcoin got going, and it has been know from the early days of mining that ASIC's would be the final word on efficiency (in both watts and $$ per GH/s). If an strong adversary, lets just call him 'Uncle Sam', had decided to attack the network previous to ASIC development and adoption, they would only need to invest in their own ASIC research and manufacture. A few thousands of chips later, they would own the network.
This doesn't change that. They just also buy half the ASICs on the commercial market. You might argue that it makes an attack cost more, and that is true, but if it raises the value of Bitcoins, it also increases the value of an attack. So again, it would cancel out.
In short, there was always a fear that the first person out with ASIC's could bring down the network easily. Now the only way is to make a whole shitload of chips, and the number (cost) required is growing by the day. There is not another foreseeable shortcut, except perhaps quantum computing, and that is why it is harder to attack now than pre-ASIC.
You're assuming the ASICs aren't being bought by the attacker. One could equally well argue that Bitcoins are more of a threat now than ever and an attacker might see the deployment of ASICs as closing their attack window and therefore, now would be the time for them to buy up ASICs to attack.
However, I suppose there is another argument I can't refute -- people might expect ASICs to raise the value of Bitcoins even though there may be no rational reason for this expectation. If that is so, it could be a self-fulfilling prophecy as people buy in the expectation that the price will go up, driving it up. (This could actually happen.)