DuPont Didn't Collapse. It Was Taken Apart on Purpose to Unlock Value.

Kevlar took twenty years to get from the red to the black. One DuPont chemist used exactly that phrase to describe the company's most famous invention — two decades of losses before the body armor and the brake pads and the bridge cables turned it into a fortune.⁸ That sentence is the whole autopsy in one line. It describes a kind of company that no longer existed by 2017, killed not by a competitor or a bad bet, but by people who looked at a 215-year-old institution and saw a portfolio to be liquidated for parts.

The official story is that DuPont collapsed. A great American industrial company failed, lost the market, and broke apart under its own weight. What actually happened is stranger and more deliberate: DuPont was taken apart on purpose, by its own board and management, to release value the stock market said the whole was hiding. Nobody beat DuPont. DuPont was disassembled.

The activist lost the vote and won the war anyway

In early 2015, Nelson Peltz's Trian Fund Management filed proxy materials arguing that DuPont's sprawling conglomerate structure was destroying value, and put Peltz himself forward for a board seat.²³ The pitch was the classic activist diagnosis: a great collection of businesses trapped inside one slow corporate shell, worth more split than whole. The board fought back, and on May 13, 2015, shareholders sided with management — all twelve DuPont nominees were elected, and Trian was beaten.¹ CEO Ellen Kullman had won.

Here is where the popular memory goes wrong. The story everyone repeats is that Peltz then forced Kullman out. He did not. She resigned in October 2015 — five months after the vote — with no ultimatum from Peltz preceding it. She told Fortune that fending off the activist had been the hardest thing of her career, won the fight, and then left over an earnings shortfall.⁵ The victory was pyrrhic. She'd spent her ammunition defending a structure that her own successor would dismantle within two years, doing precisely what the activist had asked for and getting credit for none of it.

The split was the plan, not the panic

Kullman's successor, Ed Breen, did not improvise a rescue. He engineered a demolition. On December 11, 2015 — just seven months after Trian lost — DuPont agreed to merge with Dow Chemical. The deal closed effective August 31, 2017, creating a colossus called DowDuPont. But the merger was never meant to last. It was structured from inception with three divisions — Agriculture, Materials Science, and Specialty Products — and separations were expected within eighteen months.⁴ You do not merge two of the largest chemical companies on earth in order to keep them merged. The point of the union was the subsequent division: combine, sort the pieces by industry, then cut them loose as cleaner, more 'pure-play' stocks the market would reward.

This is the mechanism the word 'collapse' obscures. A conglomerate discount is real — investors pay less for a tangle of unrelated businesses than they'd pay for each one standing alone, because they can't price the tangle. Trian was right about that. But the cure Breen administered was not to fix the tangle. It was to delete the company that contained it. The merge-then-split design treated DuPont not as an institution to be run but as a balance sheet to be rearranged, and the rearrangement required the original entity to die.

The toxin that traveled with the pieces

There was a reason the disassembly happened in a specific order. Before the Dow merger was even signed, DuPont had already spun off Chemours — its Performance Chemicals business — on July 1, 2015.⁹ That business carried the legacy of C8, the PFAS chemistry behind Teflon, and the cancer and contamination litigation that came with it. In 2021, DuPont, Chemours, and Corteva agreed to a binding memorandum of understanding totaling up to $4 billion to resolve PFAS liabilities, split 50-50, including a $1 billion escrow for future PFAS claims funded over eight years.¹⁰ A liability spread across decades of manufacturing is exactly the kind of risk a single, whole company can absorb and a set of freshly split pieces cannot easily price. Breaking up didn't make the liability vanish. It just decided who got stuck with it.

What got cut wasn't fat. It was the lab.

The deepest cost doesn't show up on any balance sheet, because it was never on one. DuPont's Experimental Station — the central research lab founded in 1903 and formally organized as Central Research & Development in 1957 — invented nylon, Teflon, Kevlar, and the PET bottle, and trained Nobel laureates along the way. Under cost pressure it was shrunk, reorganized, and renamed Science & Innovation.⁸ That is the quiet part of the autopsy. A lab that took twenty years to make Kevlar pay runs on a clock Wall Street stopped tolerating. Pure-play companies optimized for quarterly clarity do not fund two-decade bets, because a two-decade bet is exactly the kind of opacity an activist sets out to eliminate. The conglomerate discount was the market's price for DuPont's patience. Remove the discount and you remove the patience too.

But wasn't the activist simply right?

The honest objection is that this is too romantic. Maybe the conglomerate discount was real, the patient lab was also a place where good money went to die slowly, and splitting up genuinely served the people who actually owned the company — its shareholders. There's evidence for it. Trian's own filing argued DuPont had trailed its peers by roughly 1,100 basis points going into the recession.³ The diagnosis of trapped value wasn't invented.

But the same record cuts the other way, and which way depends entirely on when you start the clock. Yale's Jeffrey Sonnenfeld documented a 266% shareholder return in Kullman's first six years against 165% for the S&P 500, and argued her exit reflected a dysfunctional board rather than leadership failure.⁶ Trian countered that the baseline was cherry-picked from a recession trough to flatter her. Both are right, which is the point: the 'failure' and the 'success' are framing, not fact. What is not framing is that a company can outperform the market and still be dismantled — because outperformance and being worth more in pieces are different claims, and an activist only needs the second one.

DuPont's headstone should not read 'failed.' It should read 'released for parts.' The company that took twenty years to make Kevlar pay was worth more, on a single afternoon, broken into pieces that would never again be allowed to spend twenty years on anything. That trade may have been the right one for the people who owned the stock that quarter. But an autopsy is not a verdict on the deal. It is a record of what was inside the body — and what DuPont was carrying, when it was opened up, was two centuries of the kind of patience the market had finally decided it could no longer afford to price.