Iran announced that it will begin assembly of 6,000 new centrifuges at its uranium enrichment facility at Natanz. This would be in addition to cascades already constructed, but it is uncertain if those cascades, which consist of obsolete P1 centrifuges, will be dismantled in favor of the new improved IR2 centrifuge. The February International Atomic Energy Agency (IAEA) report states that Iran did dismantle some of its early centrifuges to make room for the newer model at the facility at Natanz.  That report also indicated that the largest cascade that Iran had yet experimented with was 164 units.

It should be noted that, while 6,000 centrifuges would not be enough to make Iran self-sufficient in supplying the Bushehr reactor, it could produce enough fissile material for a crude weapon in less than a year. Of course, it will be some time before Iran gets these centrifuges up and running, and before they perfect full-scale operations. Furthermore, there is a very tricky question of how Iran would go about conducting a strategic breakout if its intentions are, indeed, nefarious. In other words, the Natanz facility is now under safeguards, and material diversion would presumably be difficult. If Iran announces it is leaving the NPT, expels the IAEA cameras and surveilling equipment, and bars inspectors from entry, there would probably be reasonably high support for counterproliferation strikes against Iranian fuel cycle facilities.  

Another path that Iran could attempt is to build a covert parallel program. This is of greater concern to me. It would be expensive, and the risk of discovery under such high scrutiny is not inconsequential. It has also been pointed out that the current plant was originally Iran’s covert plant (Iran did not declare its fuel cycle related efforts up front, but, rather, was caught), and this should be taken into account when considering a covert parallel hypothesis. However, if Iran can divert enough material for weapons while under safeguards (granted it is not under Additional Protocols), then we do indeed have a lot of which to be afraid.