That said, the threat vectors for APTs aren't always solely network-based. IIRC, the behavior of the APT that struck RSA was fairly targeted and changed behavior once inside the network. It isn't unreasonable to think that the NSA would want to talk about putting sensors inside a private network where they could detect "successful" penetrations.
Again, we have every right to be wary of letting the (government) camel nose in the tent. I'm just saying that if you don't kinda sorta trust your government to help you fight a serious foreign power with nearly unlimited resources, you might need to reconsider your position. The threat from China is real and ongoing versus the theoretical threat from the US Government.
For example, the NYTimes has a bureau in Shanghai, and if they use a private contractor to help strengthen their networks, then I don't think that the Chinese government would be cancelling visas. If the NYTimes brought in the NSA, then I give the Chinese government five minutes before all of the visa get cancelled and the journalists get expelled.
Simply put, defense is hard. That's the main reason you've seen all of the national intelligence organizations playing hot potato with the issue. Hopefully this will be changing a bit for the better based on the executive order Obama signed recently that directs DHS, NSA and the FBI to expand their programs that share active threat information with private companies.