The Enrichment Dilemma

This is not a scandal. It is a procedure. Energy Secretary Chris Wright and Saudi Energy Minister Prince Abdulaziz bin Salman signed the 123 agreement and an accompanying bilateral safeguards agreement in July 2026. Thirty-year term. A two-year joint study to determine whether enriching uranium on Saudi soil is "warranted and commercially viable." A ten-year cooling-off period if Washington says no. Netanyahu's office called the prospect of Saudi normalization with Israel "an historic leap forward for peace in the Middle East." Netanyahu was at the same time pressing Washington privately about the enrichment pathway. Both positions concerned the same document. The Atlantic Council convened experts to read between the lines and found, predictably, that half of them saw a nonproliferation victory and half saw an export license for a bomb-making capability.

Both halves are correct. That is the actual story, and it has been the actual story since 1953.

i · the contradiction was load-bearing

Eisenhower gave the Atoms for Peace speech to the UN General Assembly on December 8, 1953. The offer was simple: the United States would distribute civilian nuclear technology to anyone who promised not to weaponize it. The reasoning was that the technology was going to spread regardless, so better to spread it with strings attached than to watch it spread without them.

Fifteen years later that logic was written into the Treaty on the Non-Proliferation of Nuclear Weapons as Article IV, which affirms the "inalienable right" of every party to develop nuclear energy for peaceful purposes. The treaty that exists to stop the spread of nuclear weapons contains, as a load-bearing clause, a guarantee of access to the industrial process that produces them. Enrichment does not care what you intend. The hard part is the front of the curve: by the time uranium has been enriched to reactor grade, the large majority of the separative work required to reach weapons grade has already been done. The remaining distance is shorter than the numbers suggest.

This was not an oversight. It was the price of getting anyone to sign. The non-weapons states would not accept permanent second-class status without a compensating right, so they were given one, and the right they were given was the one thing that made the prohibition unenforceable. The regime was built with its own failure mode installed at the foundation, and everyone in the room knew it.

India demonstrated this on May 18, 1974, using plutonium from the CIRUS research reactor — supplied by Canada, moderated with American heavy water, transferred explicitly for peaceful research. New Delhi called the test a "peaceful nuclear explosion," which was legally coherent under the terms as written and materially indistinguishable from a bomb. Israel had already gotten there through French assistance at Dimona beginning in 1957. Pakistan got there through A.Q. Khan and a set of URENCO centrifuge blueprints that walked out of a Dutch facility in the mid-1970s.

Three different routes. Same destination. In every case the supplier believed the strings would hold, and in every case the strings were a story the supplier was telling itself about leverage it did not have.

ii · the safeguards objection

The obvious rebuttal is that the comparison does not hold, and it deserves to be met head-on rather than drowned in historical dread. India in 1974 operated in a world with no Nuclear Suppliers Group, no Additional Protocol, and no NPT obligation of any kind — New Delhi never signed. Dimona was built in the dark. Khan's blueprints walked out through export controls that barely existed. The Saudi agreement arrives with a bilateral safeguards agreement signed the same day and IAEA monitoring attached to the facility. The architecture that was missing in every one of those cases now exists.

It does, and it does something the earlier arrangements could not: it makes diversion visible. That is a real improvement and it should be counted as one. What it does not do — what no safeguards regime has ever done — is make diversion impossible. Inspection is a detection system, not a physical barrier. It reports that material has moved after the material has moved. Whether that report is worth anything depends entirely on breakout time, the interval between the alarm and the weapon, and safeguards do not touch that interval. They shorten the gap between the act and the knowledge of it. They do not lengthen the gap between the knowledge and the bomb.

A state that has built the enrichment capacity, trained the workforce, and accumulated the stockpile has already paid for everything expensive. What remains is a political decision and a few months of running the cascades harder, and the IAEA's inspectors will document the entire sequence accurately while it happens. So the case is not that safeguards are worthless. It is that they modify the timeline and leave the capability untouched, and capability is what the historical record is actually about. India's peaceful nuclear explosion was not a failure of monitoring. It was a correct reading of what possession permits.

iii · the gold standard was never a standard

In 2009 the United Arab Emirates signed a 123 agreement containing a legally binding renunciation of enrichment and reprocessing on its own territory. Washington called this the "gold standard" and announced it as the new template — proof that a Gulf state could have civilian nuclear power without acquiring the fuel cycle that makes it dual-use.

The template was applied approximately once. Subsequent agreements with other partners softened the commitment to political statements, side letters, and language about intent. By 2018, when Riyadh made clear it would not accept the UAE terms, the gold standard had already spent a decade functioning less as a standard than as an argument the United States made when it had the leverage to make it and dropped when it didn't.

Saudi Arabia refused. Saudi Arabia has now been accommodated. The accommodation is structured as a delay — a two-year study, a facility whose operational details are held closely, a ten-year restriction on going elsewhere if the study comes back negative — and the delay is doing the work the prohibition used to do. Procurement is reportedly steered toward American firms, with foreign suppliers relegated to secondary roles. That is the actual mechanism, and it is worth holding onto: not a limit on capability but a claim on the supply chain.

Which is a leverage argument, and leverage arguments are how every previous exception was justified.

iv · the supplier's trap

The strongest case for the deal was made by its defenders in plain terms: if the United States doesn't sell Saudi Arabia this capability, Rosatom or the China National Nuclear Corporation will, and those suppliers attach fewer conditions and less oversight. Jennifer Gordon framed the agreement as a victory for the US-led nonproliferation regime on exactly this basis. Matthew Kroenig framed it as exporting bomb-making capability to an additional country.

The argument structure is worth naming because it is the oldest one in the file. It is the argument that sold reactors to India. It is the argument behind Atoms for Peace itself. It is the argument every arms exporter has made in every century with adequate record-keeping: the sale is defensive, because the alternative is a sale by someone worse.

The argument is not wrong. That is what makes it a trap rather than a lie. In a market with multiple suppliers, refusing to sell does not prevent the transfer; it only forfeits your seat at the table. Any single supplier acting alone faces a choice between influence-with-proliferation and no-influence-with-proliferation, and the rational move is always influence. The regime only ever worked when suppliers coordinated — the Nuclear Suppliers Group was formed in 1975 specifically in response to India's test, precisely because unilateral restraint had proven to be a donation to competitors.

That coordination is what has degraded. Not the treaty text, which is unchanged. The shared expectation underneath it.

v · what actually holds

Here is the part that policy analysis tends to route around, because it is not a policy problem.

A nonproliferation regime is not a container. It is a coherence — a structure that holds because enough participants believe that enough other participants are still holding to it. The IAEA cannot physically prevent enrichment. The Atomic Energy Act cannot reach a centrifuge hall in another sovereign state. What restrains a government is its own estimate of what everyone else will do next, and that estimate is a shared fiction in the technical sense: real in its effects, entirely dependent on continued belief.

Each exception is therefore not a marginal adjustment. It is a re-derivation of the whole structure by every party simultaneously. Ankara recalculates. Cairo recalculates. Both are straightforward cases — NPT members with civilian programs, regional ambitions, and a fresh demonstration that the terms are negotiable if you refuse long enough.

Seoul and Tokyo recalculate too, on a different beam, and the connection has to be drawn rather than assumed. Their restraint has never rested on the consistency of 123 agreements; it rests on extended deterrence, on the American guarantee that they will not need a weapon of their own. But what this deal broadcasts is a fact about how Washington treats clients. An adversary was told that any enrichment whatsoever was intolerable — the 2015 JCPOA permitted Iran limited enrichment under intrusive monitoring, and the United States withdrew in 2018 on precisely that position. The same policy coalition is now building an enrichment pathway for a different Gulf state. The rule changed with the identity of the applicant. Seoul and Tokyo are not pricing the paperwork; they are pricing the reliability of American commitments, and a commitment that varies by client is a different instrument than one that doesn't. That is useful information if you are deciding how much of your security to keep hosted on someone else's guarantee.

So the absolutism was never a principle. It was a bargaining position, and every government in the region can see that now with total clarity. A rule that applies to adversaries and not to clients is not a rule. It is a scheduling system. It can still function for a while — scheduling systems do — but it no longer explains anything, and nobody obeys a structure that has stopped explaining itself. They comply with it exactly as long as compliance is convenient.

vi · the regime is not failing. it is being repurposed.

That reading is true and it is still the shallow one. The deal's own structure says something more specific, and it is sitting in the clauses.

Look at what the agreement actually contains. Not a prohibition. Not even a real ceiling. A two-year study to determine commercial viability. A ten-year restriction on going to another supplier if the answer comes back no. Procurement steered toward American firms. Every one of those provisions regulates supply. Not one of them regulates capability.

That is not the residue of a failed prohibition. It is the architecture of a rent position.

The nonproliferation regime began as a claim about a class of industrial capability: some things are too dangerous to spread, and the international system will act to prevent their spread. Seventy years on, the operative question in Washington is no longer whether the fuel cycle spreads. It is who supplies it, on what terms, and how long the customer stays locked in. The gold standard was not abandoned because anyone concluded that enrichment had become safe. It was abandoned because holding the line cost market share. The ten-year cooling-off period is not a nonproliferation instrument; it is a non-compete clause. The two-year study is not a safety review; it is a procurement schedule wearing a deliberation costume.

This explains the evidence better than hypocrisy does. Hypocrisy cannot account for the shape of the thing — a hypocrite who wanted the sale would simply make the sale. Only a broker structures a deal this way: preserve your optionality, foreclose the competitors, defer the irreversible commitment until the commercial case is documented and the customer has nowhere else to go. The delay is not weakness of will. It is a term sheet.

And this is the part worth sitting with, because it is larger than one agreement. A hazard that was supposed to be governed as a shared risk — the entire premise of the treaty is that a bomb anywhere is a problem everywhere — is being converted into an asset, enclosed, and sold back to the parties it was designed to restrain. Nobody argued the commons framing away. It was quietly swapped for a market framing while the treaty language stayed identical, which is how enclosures usually work. No one repeals the common right. They build the fence and keep using the old word for the field.

Congress could theoretically intervene. Under the Atomic Energy Act, a 123 agreement laid before Congress enters into force by default; blocking it requires a joint resolution of disapproval passed over a presidential veto. Senators Markey and Rubio, with Representative Sherman, introduced legislation to raise that bar specifically for Saudi Arabia — and reintroduced it across multiple Congresses, which is the legislative record's way of telling you how it went. Default approval is not oversight. It is a countdown clock with a comment period attached.

So the prediction is not difficult. The two-year study will find enrichment commercially viable, because studies commissioned to justify a signed agreement generally do. The facility will be built. Turkey will open a file. Egypt will open a file. And in some number of years a report will describe the resulting situation as an unprecedented proliferation cascade in the Middle East, and everyone will be surprised, and the surprise will be sincere — sincere because the people expressing it will have spent the entire interval reading the agreement as a nonproliferation document, which is not what it is.

The pattern has been running since 1953. But who's counting.