This article by J. E. Dyer warns that while Iran’s nuclear program could still be stopped by bombing, the damage would be much greater now than if it had been done two years ago:

Two years ago, military planners would have emphasized attacking the uranium-processing facilities at Esfahan and Natanz, particularly in an air strike of limited scope and duration (in other words, what Israel is capable of mounting). These facilities are “critical nodes” if they perform unique functions. But if they don’t — if Iran can process uranium at undeclared facilities elsewhere — then optimizing a limited strike requires identifying a bottleneck at another step in the process. The only real bottleneck left is the process of weaponization itself: developing a warhead that will detonate and mating it to a delivery platform. Interdicting the research and development for that is a task for which kinetic strike is less suited and would entail a higher political cost, in part because the Iranians have their weaponization laboratories in heavily populated areas of Tehran.

An American-scale air strike could still destroy Iran’s current facilities sufficiently to set the program back by a factor of years. But the time has passed when we could achieve something useful — say, setting the program back for 18-24 months [and sending a strong message – JB] — with a “surgical strike” against the declared uranium-processing facilities. If we wanted to be sure of taking out the uranium now, we would probably enlarge any existing strike concept to use Massive Ordnance Penetrators (MOPs) against multiple underground facilities. In combination with attacks on R&D facilities in Tehran, this would mean more destruction and loss of Iranian life than achieving the same effect would have required two years ago.

The political cost of a military attack on Iran’s nuclear program was always going to be high. But we have almost certainly reached the point at which there is no useful effect to be achieved with a limited, “surgical” strike. A massive, comprehensive attack, on the other hand, would impose such political cost that its objective might as well be regime change anyway. Even Israel still has some viable attack options, but the prospective effects are not what they would have been two years ago. We’re down to the stark alternatives we were always going to face in the end: a regime-changed Iran or a nuclear-armed one.

And what are the chances of regime change?

It looks ever more probable that a bellicose Iran will soon have nuclear bombs, and will use them.