DuPont Keeps Promising One Last Breakup. There Is No Last Breakup.

On July 1, 2015, DuPont handed its shareholders stock in a brand-new company called Chemours and called it a fresh start — a leaner, more focused DuPont, unburdened by its commodity chemicals.⁴ Five months later, that same focused DuPont agreed to merge with Dow into a $130-billion colossus.¹⁰ Then it broke the colossus into three. Then it sold a chunk to Celanese. Then it announced a three-way split, revised it to a one-way split, and put Kevlar up for sale. A decade of press releases, every one promising the same thing — focus — and not one of them the last word.

The official story is that DuPont is a master capital allocator, restlessly sharpening its portfolio to surface hidden value. The truer story is that DuPont has no portfolio it can hold still. Each 'New DuPont' is announced as the final, focused company. Each one promptly announces the next separation. There is no terminal state — only the next slide deck.

The merger that existed only to be torn apart

Here is the strangest fact in the whole saga, and it gives the game away. Dow and DuPont signed their merger on December 11, 2015, and closed it on August 31, 2017 — but the stated intent from the day of announcement was to split the combined company back into three.¹⁰ The board's plan was never to run DowDuPont as one company. It was to merge the two firms, reshuffle their businesses into cleaner buckets, and spin those buckets out: Materials Science as Dow, Agriculture as Corteva, and Specialty Products as the residual DuPont.² The Dow spin closed April 1, 2019; the Corteva spin followed on June 1, 2019 — sequentially, not simultaneously.²³ A two-year merger built for the sole purpose of being dismantled is not a strategy. It is an admission that the businesses never belonged together in the first place.

Notice the chronology, because the popular telling gets it backwards. Chemours was carved out on July 1, 2015 — months before the Dow merger was even announced.⁴ It was not post-merger cleanup. DuPont was already in the business of subtracting itself from itself before it ever decided to add Dow. The merge was an interruption in a longer pattern of splitting, not a departure from it.

The plan that couldn't survive eight months

If you want proof that the logic is improvised rather than industrial, look at 2024. On May 22, 2024, DuPont announced — with the full ceremony of a company that has finally figured itself out — a plan to separate into three publicly traded companies: Electronics, Water, and a slimmed New DuPont, all via tax-free spinoffs.⁶ Eight months later, on January 15, 2025, that plan was gone. DuPont would keep the water business after all and spin off only Electronics, as Qnity, targeting November 1, 2025.⁷ A company confident in its industrial logic does not reverse a flagship breakup before the ink is dry. The reversal tells you the breakup was never driven by an irreducible truth about which businesses belong together. It was driven by whatever the market and its loudest holders wanted to hear that quarter.

And the subtraction keeps going past the headline split. On August 29, 2025, DuPont agreed to sell its Aramids business — Kevlar and Nomex, two of the most iconic materials it ever invented, generating $1.3 billion in 2024 net sales — to Arclin for roughly $1.8 billion.⁸ These were supposed to be retained, core, the durable heart of a focused materials company. They turned out to be just the next thing on the block. At DuPont, 'core' is not a category. It is a temporary label applied to whatever hasn't been sold yet.

Who actually decides what DuPont keeps

Here is the mechanism, worked down. A conglomerate trades at a 'discount' — the sum of its parts, valued separately by specialist analysts, exceeds the value the market assigns the whole. Activists and bankers price that gap precisely, then arrive with a fix: split the company, and the discount converts into a one-time pop. The pop is real. But it is a financial event, not an operating one — nothing about the molecules changed, only which ticker holds them. Once the pop is captured, the residual company is itself a smaller conglomerate, and the same math reappears: there is always a sum-of-the-parts case for cutting one more time. This is why the cycle has no end. The discount can always be re-discovered in whatever remains. So the sequencing isn't dictated by which businesses share customers, plants, or chemistry — the M&M deal went to Celanese as a $11-billion cash sale precisely because shareholders got money, not the synergy of a combined company.⁵ The sequencing is dictated by where the next valuation gap can be argued. That is finance setting industrial strategy, and finance never runs out of gaps.

Isn't perpetual pruning just good capital discipline?

The fair objection is that this is exactly what a well-run modern industrial should do: own nothing it isn't the best owner of, sell to whoever values a business more, and return the gap to shareholders. Sentiment, not chemistry, should decide. And there is real evidence DuPont creates value with each move — Chemours, Dow, and Corteva are all viable companies that arguably run better unbundled.²⁴ The honest version of that case is strong. But it proves the point rather than refuting it. If every separation is correct, then the previous combination was wrong — including the $130-billion Dow merger DuPont itself orchestrated and closed barely two years before tearing apart.¹ A company that is right to split is, by the same logic, a company that was wrong to assemble. And the reversal of its own 2024 plan within eight months⁷ shows the deciding variable isn't a stable read of which businesses belong together — it's the moving target of what the market will reward this quarter. Disciplined capital allocation has a thesis. Perpetual churn has only a next move.

DuPont was founded in 1802⁹ and spent two centuries inventing things — nylon, Teflon, Kevlar, the materials that built the modern world. For the last decade it has mostly been inventing transactions. The Aramids sale⁸ means even Kevlar, one of its crown jewels, is now just inventory awaiting the right buyer. There is nothing wrong with selling a business worth more to someone else. The problem is that DuPont can no longer tell you what it is keeping — only what it is selling next. A company is supposed to be the thing that survives the deals. At DuPont, the deals have become the thing, and the company is whatever happens to be left over between them.