The Roivant alliance: taking control of Sumitovant and 11% of Roivant (2019)

What ten years of buying probability came to

When the company paid $2.6bn for Sepracor, what it bought was an American sales network. When it paid about $3bn for the Roivant deal, what it bought was a mechanism for selecting and advancing drug candidates, and the twenty-five-plus programmes hanging from it. The shift over ten years from product to probability was driven, it seems, by both the experience of having hit once with a single agent in Latuda and the fixed date — February 2023 — on which that agent would disappear. It was a decision to buy with money what cannot be bought with time.

Buying probability, however, does not guarantee a batting average. All three acquired products reached the market, yet the business plans were missed, and ¥133.5bn of Myfembree patent rights were converted into an impairment. Given that Orgovyx landed at 136% of the initial forecast in the year to March 2025, what was wrong may have been less the strength of the products than the way the revenue plans built on them were set. When Kimura Toru chose his words carefully — that whether the alliance succeeded “is something we will verify” — it showed that even the people inside it cannot yet draw that line.