This is not news (yet). See either:
Hellman, M. E., "DES will be totally insecure within ten years", IEEE
Spectrum, Vol 16, 1979
or
Diffie, W. and Hellman, M. E., "Exhaustive cryptanalysis of the NBS
Data Encryption Standard", Computer 10(6), 1977.
When DES was first deployed only the NSA could afford the cracking
engine, but it was always clear that a 40% annual improvement in
silicon would remove the security of DES by now. The manufacturing
cost of a practical DES-breaking engine (one that could break a key in
less than a week) is now down below $100K. Further advances in silicon
are almost uninteresting until it becomes practical to break DES in
software on general purpose CPUs (which is still about 10 years out).
What matters now is the design cost for such an engine. It's not cheap
to engineer a complex chip like this one and the boards to use them in
parallel. The best estimates I've gotten are about $400K. A
university with access to cheap labor could probably do it for a lot
less. So yes, the machines could be sold for $100K in bulk but there
is an initial hurdle. And that cost is *not* dropping 40% per year.
It's a fair guess that a dozen or so DES breaking machines have been
built. It's a safe guess that NSA has at least one. It is not news
that we know how to build them. It will be news when someone admits
publicly that they have one and demonstrates it and it will be news
again when one appears for sale in the Sharper Image catalog. Until
then, DES is what it has always been... better than nothing.
--Charlie
(kaufman(_at_)zk3(_dot_)dec(_dot_)com)