Intel plans to include security features in upcoming chipsets, including copy protection, certificate management, and -- get this -- random number generation. This will be implemented in firmware, and, given that it is supposed to perform copy protection functions, there is a significant possibility that the code would be inaccessible, maybe (heaven forbid) within a tamper-resistant shell.
To clarify exactly why this is a problem: although it obviously won't affect the security of existing apps, it's likely that many newer apps will be dependent on this possibly-secret firmware for security, and we all know the evils of depending on a design not subject to intense public scrutiny. Worse yet, it seems possible that a certificate-based system could be used to ensure that neither non-Intel chips nor software fill-ins would be accepted by applications using the firmware. And, of course, except for what was mentioned in the TechWeb article itself, this is all conjecture based almost entirely on Murphy's Law.