On the many roads from monazite sand to megawatt-hours — and where a fusion-driven neutron source fits in.
On 6 April 2026, the Prototype Fast Breeder Reactor at Kalpakkam attained first criticality. For those of us who have spent our working lives around India’s nuclear and plasma programmes, it was a moment to sit with quietly. Construction began in 2004. First operation was once planned for 2010. The reactor is now running, and India has formally entered the second stage of the programme Homi Bhabha sketched in 1954.
I want to use that milestone to ask a question we do not ask often enough: when we say “thorium,” what exactly do we mean?
Because thorium is not a technology. It is an element — and there are at least six distinct engineering routes by which India could turn it into electricity. Some are decades old and nearly mature. Some exist only as design studies. They are not competing religions, and they do not all answer the same question. Treating them as one undifferentiated national aspiration has, I think, cost us clarity.
What thorium is, and what it is not
Natural thorium is essentially all Th-232, and Th-232 is fertile, not fissile. You cannot build a reactor that runs on thorium the way one runs on enriched uranium. Th-232 must first absorb a neutron, becoming Th-233, which beta-decays to Pa-233, which beta-decays with a 27-day half-life to U-233. Only then do you have fuel.
That 27-day step is not a footnote. Protactinium-233 sitting in a neutron flux absorbs neutrons that would otherwise breed more U-233, which is why serious thorium fuel-cycle designs care a great deal about whether and how protactinium is held outside the core while it decays. It is one of the reasons molten-salt designs are attractive on paper: you can chemically remove the protactinium.
Once you have U-233, however, you have something genuinely special. U-233 is the only fissile nuclide with a neutron yield per absorption comfortably above two across the thermal spectrum. That single fact is why thermal breeding is physically possible in the Th–U-233 cycle and essentially impossible in the U–Pu cycle. Bhabha’s insight was not merely that India had thorium and little uranium. It was that thorium’s daughter is the one fissile material that permits a self-sustaining cycle in the reactor types a developing country could actually build and operate.
Thorium also brings mundane engineering virtues that deserve more attention than they get. Thorium dioxide has higher thermal conductivity and a substantially higher melting point than uranium dioxide, retains fission products better, and is chemically stable and largely insoluble after irradiation — which makes it a good waste form as well as a good fuel. And the U-233 you breed always arrives contaminated with U-232, whose decay chain includes a hard 2.6 MeV gamma from thallium-208. This complicates fuel handling considerably. It is also, precisely, what makes the material unattractive to divert. The inconvenience and the safeguard are the same physical fact.
The bottleneck was never thorium
Government figures put India’s in-situ monazite at roughly 11.93 million tonnes, containing on the order of a million tonnes of thorium. It is worth being careful with that number rather than repeating it triumphantly. A parliamentary answer in 2012 clarified that of the many identified heavy-mineral deposits, about 4 million tonnes of monazite were then assessed as exploitable — corresponding to something closer to 225,000 tonnes of recoverable thorium metal. That is still an extraordinary endowment, sufficient on BARC’s own analysis for hundreds of thousands of gigawatt-electric-years. But “largest reserves in the world” and “largest economically extractable reserves, characterised and permitted” are different statements, and we do our credibility no favours by eliding them.
Even granting the full endowment, though, the constraint has never been the thorium. It is the seed.
Every thorium pathway needs an initial inventory of fissile material to get started, and needs a neutron economy generous enough to breed more U-233 than it consumes. India’s own Department of Atomic Energy has been admirably honest about the consequence: extensive thorium deployment is projected for beyond 2070. That is not pessimism or institutional inertia. It is arithmetic. Building a fissile inventory through fast breeders is a slow compounding process, and compounding is unforgiving of a low starting balance.
So the real question for anyone serious about Indian thorium is not “which reactor burns thorium best?” It is: how do we get more U-233, sooner?
Every road below is, in the end, an answer to that.
Six roads
Road 1 — Fast breeder blankets. This is the sanctioned route, and it is now real. PFBR uses uranium-plutonium MOX fuel with a fertile blanket; the design intent is that thorium can be placed in that blanket, where fast neutrons transmute Th-232 into U-233. India has already closed a version of this loop at laboratory-to-semi-industrial scale: thoria blanket assemblies irradiated in the Fast Breeder Test Reactor have been reprocessed at IGCAR, and the recovered U-233 used for further irradiation experiments. KAMINI at Kalpakkam remains, so far as I know, the only reactor anywhere fuelled by U-233. Strength: it works, and it is ours. Limitation: breeding rates are modest and each new breeder must be seeded from the last. This is compounding, and compounding takes decades.
Road 2 — Thorium in heavy water reactors. BARC’s Advanced Heavy Water Reactor is a 300 MWe pressure-tube design intended to draw the majority of its power from thorium, using Th-Pu and Th-U-233 MOX, with passive safety features throughout. Design of the nuclear systems is complete, AERB has accorded pre-licensing approval, and there is in-principle approval for a site at Tarapur. Separately, thoria bundles have been irradiated in operating PHWRs for flux flattening, with the Power Reactor Thorium Reprocessing Facility built to handle them. Strength: deep indigenous competence, and a genuine technology demonstrator for the whole thorium fuel cycle. Limitation: as of this year, no AHWR has begun construction. A design that has been ready for a decade and not built is telling us something about our project-execution pipeline rather than about our physics.
Road 3 — Thorium in light water reactors. Since roughly ninety per cent of the world’s operating reactors are light water reactors, there is obvious commercial logic in a thorium fuel that drops into existing hardware — typically thorium-plutonium MOX pellets, with reactor-grade plutonium as the fissile driver. Several ventures internationally are pursuing exactly this. It deserves a fair hearing, and it also deserves scepticism on two counts. First, this route consumes U-233 as fast as it breeds it; it is thorium utilisation, not cycle closure, and it does not build the national fissile inventory. Second, the India fit is weaker than it first appears: our fleet is PHWR-dominated, and our light water units are essentially Kudankulam and Tarapur. A BARC team published an assessment in Current Science in March this year concluding that a proposed HALEU-thorium blend could not be dropped into our 220 MWe PHWRs without significant design change, and raising strategic questions about fitting foreign fuel concepts into a domestic fuel strategy. That paper is worth reading in full. Whether or not one agrees with every conclusion, the instinct behind it — check the fit before adopting the frame — is correct.
Road 4 — Molten salt. The Indian Molten Salt Breeder Reactor programme at BARC targets liquid fluoride fuel and coolant at around 650 °C, coupled to a supercritical CO₂ Brayton cycle, with a 5 MWth demonstrator and a dedicated development facility being designed. The attraction is real: online chemical processing lets you extract protactinium and fission products, which is the cleanest answer to the neutron-economy problem in Road 1. The difficulty is equally real — lithium-7 enrichment, salt chemistry, corrosion-resistant structural alloys, nuclear-grade graphite. China has demonstrated a small liquid-fuelled thorium molten salt reactor and reported converting thorium into usable fuel in it. We should treat that as a serious signal rather than a slight.
Road 5 — Accelerator-driven systems. Drive a sub-critical assembly with spallation neutrons from a high-power proton accelerator, and you decouple the neutron supply from the criticality of the fuel. DAE has worked on ADS concepts, including in collaboration with Fermilab. Strength: an external neutron source is not bound by the neutron economy of the fuel itself. Limitation: you need a very high-power, very high-availability accelerator, and spallation targets are demanding components.
Road 6 — Fusion-driven sub-critical hybrids. This is the road my colleagues and I are building, and it shares Road 5’s central logic with a different neutron source.
Why an external neutron source changes the arithmetic
The reason Roads 5 and 6 matter is a single point about neutron accounting.
In a critical thermal reactor, every neutron is spoken for. The chain reaction must sustain itself, so the neutrons available for breeding are only whatever the fuel’s neutron economy leaves over — and in the thermal Th–U-233 cycle that margin is thin. You breed roughly what you burn. Net surplus is hard.
An external source changes the budget. A deuterium-tritium fusion reaction produces a 14.1 MeV neutron — far more energetic than anything from fission. At that energy, neutron multiplication in a suitably designed blanket becomes substantial, and the assembly can be operated sub-critically, at an effective multiplication factor deliberately held below one. The fuel is not required to sustain itself, so its neutrons are not all committed to doing so. What remains is a genuine surplus that can be spent on breeding U-233, breeding the tritium the driver needs, and transmuting long-lived actinides — concurrently.
There is a safety consequence that I think matters as much as the physics. A sub-critical assembly at k-effective in the 0.90 to 0.95 range has no prompt-critical pathway. Switch off the neutron source and the fission power falls away. That is not a claim about operating procedures or engineered safeguards; it is a property of the configuration. In a country rightly cautious about nuclear risk, and now opening parts of the sector to private operators, configurations whose safety is structural rather than administrative deserve close attention.
What we are building at ASPL Fusion
Agnira Sanlayan Private Limited — ASPL Fusion — was incorporated in August 2025, and the SHANTI Act, 2025, which received Presidential assent that December, is what makes a company like ours legally possible. Our work traces back to Project Sanlayan, which I began mentoring in December 2020; the name carries that origin.
Our architecture is an axisymmetric linear magnetic mirror using REBCO high-temperature superconducting magnets. Mirrors are an old idea that has become a good one again, for an unglamorous reason: HTS tape lets you build the very high field strengths that mirror confinement rewards, in a machine that is geometrically simple and comparatively cheap to iterate. We are not alone in that judgement — comparable programmes are being pursued in the United States on similar physics grounds, which I regard as useful independent corroboration of the architecture rather than as competition.
We are building in four phases, each intended to be commercially self-sufficient and to fund the next. Phase 1 is a boron neutron capture therapy clinic for cancer treatment together with accelerator-based neutron services for industrial inspection, at Gandhinagar. Phase 2 is a linear-mirror neutron platform for medical isotope production and fast-spectrum irradiation. Phase 3, which is the one relevant here, is PRABHA-Hybrid: a deuterium-tritium fusion-driven sub-critical assembly, with thorium in the blanket, designed to breed U-233 alongside waste transmutation and electricity generation. Phase 4 is grid-scale fusion power.
I want to be precise about the status of these statements, because the field has a long history of imprecision. Phases 1 and 2 are in execution. Phase 3 exists as a design study benchmarked against established sub-critical hybrid work; our current schedule targets construction in the 2030s with grid connection towards the end of that decade. Those are projections from a design model, not measured results from an operating device, and I will label them that way every time I am asked. Our incubation arrangement with the Institute for Plasma Research is in progress, not concluded. We are one of two private fusion companies in India, not the only one.
One structural point deserves emphasis. Under the SHANTI Act, reprocessing and the management of spent fuel beyond on-site storage remain reserved to Central Government entities, as does enrichment above a threshold and heavy water production. This is the correct architecture, and we have designed around it deliberately rather than lobbying against it. Our position is upstream: we irradiate thorium and supply the irradiated material. Separation, custody and downstream fuel-cycle operations are DAE’s, as they should be. A private company’s role in the Indian nuclear fuel cycle is to add capacity to the national programme, not to duplicate or contest its sovereign functions. I have spent enough of my career inside that programme — at SINP, on ADITYA, on SST-1 — to hold that view sincerely rather than diplomatically.
What I would like to see
Three things, offered as an argument rather than a demand.
Fund several roads, not one. The pathways above have uncorrelated failure modes. Molten-salt corrosion risk, accelerator availability risk, and fusion-driver risk are independent of one another. A portfolio across them costs more than a single bet and is far more likely to produce a working thorium economy this century. India has historically been excellent at sustaining one long programme with great patience; the harder discipline is running several in parallel and letting evidence, not seniority, allocate the next tranche.
Treat U-233 as national infrastructure. If several routes may need seed fissile material, the rules for producing, holding, allocating and pricing U-233 need to be legible in advance. Right now they are not — which means anyone designing a system that produces or consumes U-233 is designing against an unknown. Clarity here is cheap and would unlock a great deal of private engineering effort.
Characterise the monazite value chain properly — including the rare earths. Monazite is not only a thorium mineral; it is a rare-earth mineral with thorium in it. India has spent decades treating the thorium as the prize and the rare earths as an adjacent industry, at a moment when rare-earth supply security has become one of the defining industrial questions of the decade. A serious, publicly available assessment of the coastal placer deposits as a joint thorium-and-rare-earth resource would improve decisions in both sectors. I suspect it would also change the economics of thorium extraction considerably.
A closing thought on patience
Bhabha’s three-stage plan was an act of remarkable strategic imagination, and it has been criticised mainly for taking a long time. I would put it differently. It has taken a long time because it was designed to be executed with a fissile inventory that had to be built one reactor at a time, under sanctions, by a country that in 1954 had almost no industrial base for any of it. That it works at all is the achievement.
What has changed is not the goal. It is that we now have candidate technologies — molten salts, spallation sources, fusion neutron sources — that were laboratory curiosities when the plan was written and are engineering programmes today. Each of them attacks the same bottleneck from a different direction. None of them is a substitute for the patient work DAE has done; all of them are only possible because of it.
The reactor that went critical at Kalpakkam in April is a Stage 2 machine, and it took twenty-two years to build. If we want Stage 3 in the working lifetimes of the engineers now entering the field, rather than after 2070, we will need to widen the front. That is the case for many roads. It is also, in the end, the case for taking the third stage seriously as an engineering problem to be solved rather than an inheritance to be awaited.
Technical figures relating to ASPL’s Phase 3 and Phase 4 systems are design-study projections and are identified as such.