I'm kind of lazy so I'm not going to bother with the whole peak uranium nonsense.
Instead I'd like to comment on fast reactors, because there are some fundamental facts that seem to be ignored or left out of the debate in this thread so far.
The equilibrium distribution of transuranic elements in thermal and fast reactors is approximately as follows(warning, hideous formating ahead):
Isotope
Thermal
Fast
Np-237 ;
5.51
;
0.75
Pu-238 ;
4.17
;
0.89
Pu-239 ;
23.03
;
66.75
Pu-240 ;
10.49
;
24.48
Pu-241 ;
9.48
;
2.98
Pu-242 ;
3.89
;
1.86
Am-241 ;
0.54
;
0.97
Am-242m;
0.02
;
0.07
Am-243 ;
8.11
;
0.44
Cm-242 ;
0.18
;
0.40
Cm-243 ;
0.02
;
0.03
Cm-244 ;
17.85
;
0.28
Cm-245 ;
1.27
;
0.07
Cm-246 ;
11.71
;
0.03
Cm-247 ;
0.75
;
2.E-3
Cm-248 ;
2.77
;
6.E-4
Bk-249 ;
0.05
;
1 .E-5
Cf-249 ;
0.03
;
4.E-5
Cf-250 ;
0.03
;
7.E-6
Cf-251 ;
0.02
;
9.E-7
Cf-252 ;
0.08
;
4.E-8
Source:
http://www.osti.gov/bridge/purl.cov...44DC01B6BC7A?purl=/459313-d9NYz8/webviewable/
Thermal reactors cannot be breeders for the U-Pu fuel cycle for the reason that the ratio between capture cross-section and fission cross is too low. You waste far too many neutrons transmuting Pu-239 into all these higher isotopes. That's why multi-recycle of plutonium in LWRs degrades the plutonium into a form that's not even suitable for LWRs anymore.
In a fast reactor the Pu-239 still has a decent chance at capturing a neutron but Pu-240 tends to fission most of the time so you don't get that much other stuff.
Weapons grade plutonium is >93% Pu-239, with the rest being mostly Pu-240 and a little bit of Pu-241. When you subject U-238 to neutrons you first get U-239, which beta decays to Np-239, which beta decays to Pu-239. As the concentration of Pu-239 builds up the ratio of Pu-239 to U-238 atoms increases, increasing the odds that a neutron will either fission or transmute Pu-239 into undesirable isotopes of plutonium. If the U-238 is in the fuel elements rather than some breeding blanket the U-235 can also be transmuted into an expecially undesirable isotopes, Pu-238, the stuff they use as a heat source in space probes.
In both cases the quality of the plutonium is very far from weapons grade
when you approach equilibrium conditions. In a thermal reactor it is enough that the fuel spends a handful of months inside the reactor for the quality of the plutonium to be sufficiently degraded to not be desirable in weapons. Producing weapons grade plutonium in an LWR produces a very suspicious pattern of ~10 times higher fuel demands than normal(because you're taking it out, chopping it up and disolving it in nitric acid long before the fuel has been used up) and extremely frequent refueling outages visible to inspectors and spy satellites looking at smoke comming out of your cooling towers.
In an IFR, S-PRISM or BN-600 style fast-reactor you have a fuel region and a breeding blanket region. The blanket contains mostly U-238; its purpose is to make sure that neutrons that leak from the fuel region of the core get put to use producing more Pu-239 rather than going wasted. This is necessary if you want a breeding ratio significantly above 1.
The plutonium in the fuel region of the reactor is going to come from spent fuel from thermal reactors, which is already far below weapons grade and will move towards the equilibrium isotopic distribution over time.
The plutonium being bred in the blanket could be of weapons quality; that depends on how long the blanket stays in the reactor before being switched out and reprocessed.
In the integral fast reactor design the reprocessing would be integrated such that you don't need to move any fuel or blanket material off-site for reprocessing. The reactor is designed so that the breeding blanket can be replaced in whole or in parts with neutron-reflecting steel pins allowing the breeding ratio to be adjusted from less than one all the way up to ~1.25; that way you can choose to destroy transuranics with a breeding ratio of less than 1, produce more reactor grade plutonium for starting up more IFRs or produce only as much as you need to be self-sustaining as long as feed more U-238(so that no plutonium-bearing material ever leaves the site apart from small amounts that end up in the vitrified fission product waste). Even if the reactor is operated with frequent enough blanket replacement that weapons grade plutonium could be obtained from the blanket the integrated reprocessing regime and its safeguards are intended to make it difficult to divert material; and once the blanket and fuel materials are mixed up during reprocessing the plutonium quality will be uniformly poor as far as weapons are concerned.
The method of reprocessing in the also chosen to never separate out pure plutonium. This is done in order that even if someone managed to steal the reprocessed material it would be significantly radioactive and would require further reprocessing to even try to make a nuclear explosive(it is not unreasonable to expect a fizzle from an implosion weapon in the 100 tonne to 1 kT of TNT range. Low and unreliable yield and poor shelf-life is a deal-breaker for a nation-state but not for hypothetical terrorists sophisticated enough to build a working implosion device).
If a breeding ratio as low as ~1.05(theoretical value for the lead cooled BREST reactor) is acceptable it is possible to do away with the breeding blanket altoghether such that no weapons-grade plutonium is produced.
The problem with fast reactors is that the specific fissile inventory is too high for rapid deployment; at 12 tonnes of reactor grade plutonium per GW for the S-PRISM the ~50 000 tonnes of spent fuel for the entire US fleet of LWRs is enough to start just 40 GW. The doubling time is very roughly 30 years(this is my back of the envelope estimate given a breeding ratio of 1.25, a specific fissile inventory of 12 tonnes per GW and 190 MeV released per fission) so there's not much relief from that side either.
It is possible to start fast reactors on U-235(fresh fuel) instead of reactor-grade plutonium. Then you need ~2000-2400 tonnes natural uranium per GW of reactor started(DU tails contain 0.1-0.2% of U-235). If you went after the currently known ~4.5 million tonnes of uranium available at <$130 kg with war-like efficiency to start up S-PRISM reactors you'd get 1.9-2.3 TW of installed capacity. That's not enough, you'd still need a couple of doublings to get where you want to be.
Thermal breeder reactors can do much better. They can't attain the same high breeding ratios as fast-reactors, but still slightly above 1, and they they can't operate on the U-238/Pu-239 fuel cycle; but they have much lower specific inventory(~1000-2000 kg per GW depending on size and design) and rely instead on the Thorium-232/U-233 fuel cycle. In order to achieve breeding in the thermal spectrum you need to keep removing neutron-hogging fission products very quickly; the way to do this is to have the fuel remain liquid so that reprocessing is quick and easy. The molten salt reactor concept was first developed at ORNL, an offspin of tiny high temperature reactors intended for the Aircraft Nuclear Propulsion program. Because the fuel is a molten salt it can be operated at a high temperature(temperature is limited by corrosion concerns in the reactor walls).
It would be best to start them on U-233 obtained by some other means(e.g. see the Indian nuclear programme), but failing that they can be started on U-235 or plutonium, with plutonium being least preferable.
If you want to start them on highly enriched uranium or plutonium I believe you'd want to have a small dedicated fleet of molten salt reactors in a secure location in a nuclear weapons-state in order to avoid proliferation concern. They'd chew through U-235 and plutonium and spit out U-233 that can be used to start 'clean' molten salt reactors elsewhere.
Once operating on U-233 they are proliferation resistant; U-233 was never persued for nuclear weapons because neutron knock-off reactions(i.e. (n, 2n)-reactions) produce U-232 which is a strong gamma emitter that makes the production, storage and handling of bombs difficult and messes with the electronics of the device. Pu-239 is also a better fuel for a nuclear primary than either of U-235 and U-233; in a fast spectrum it produces ~0.5 more neutrons per fission which means the reaction take off faster and the inertial confinement does not need to be as good(small warheads are highly prized).
I'm note sure what combination you'd want for the fastest scaling up of nuclear energy but I think you'd want higher rates of uranium mining, continued expansion of light water reactors for a few more decades before tapering off, fast breeder reactors to soak up the entire inventory of reactor grade plutonium and minor actinides from spent LWR fuel, molten salt breeder reactors started from U-233 produced by surplus neutrons in fast reactors(I would prefer to scale MSRs faster at the expense of fast-breeder reactors, they
appear to be cheaper and provide better passive safety).
