⚠ Paying for the Next InphiModerate threat

Buying Inphi in 2021 for roughly $10 billion was widely judged aggressive at the time and looks like one of the better semiconductor acquisitions of the decade today. The problem with a famous success is that everyone has now read the playbook.

Two things follow. First, the price of the next Inphi has gone up: any asset that plausibly holds a leading position in AI interconnect is being bid by acquirers with far larger balance sheets, and the era of finding one before the market noticed is over. Second, the accounting bill for the last one is still arriving. The amortisation of Cavium and Inphi intangibles is the reason Marvell posted GAAP losses in six of the last nine fiscal years¹, and a further large acquisition would restart that clock rather than end it.

There is a temptation here that is worth naming plainly: a company whose shares have re-rated sharply has an expensive-looking currency that feels cheap to spend, and that is precisely when acquirers overpay.

The falsifier is a large deal done at a full multiple in a hot category. Watch what Marvell buys next and what it pays. Discipline here is worth more than any single technology it could acquire.