NVIDIA's $40 billion purchase of Arm collapses

NVIDIA and SoftBank announced the termination of NVIDIA’s proposed $40 billion acquisition of the chip-design firm Arm, first agreed in September 2020, citing “significant regulatory challenges” that neither side could overcome despite what the companies called good-faith efforts. Regulators in the United States, the United Kingdom and the European Union had all raised objections; the US Federal Trade Commission had sued in December 2021 to block the deal outright, arguing that giving NVIDIA control of the architecture licensed to virtually every major chipmaker, including NVIDIA’s own rivals, would let it unfairly undermine competitors reliant on Arm’s technology.

Under the deal’s terms, SoftBank kept a $1.25 billion prepayment NVIDIA had made at signing, which it recorded as profit. NVIDIA separately disclosed in subsequent financial filings that it would take a $1.36 billion charge reflecting the write-off of that prepayment. NVIDIA retained a 20-year architectural licence to Arm’s technology, and both companies’ chief executives — Jensen Huang and Masayoshi Son — issued statements describing the relationship as continuing on other terms.

SoftBank said it would instead prepare Arm for a public listing within its fiscal year, and Arm went on to list on Nasdaq the following year.

Had it closed, the deal would have been the largest acquisition in semiconductor history and would have placed the architecture underlying most of the world’s mobile processors, and a growing share of AI accelerators, inside the company that already dominated AI training hardware. Its collapse left Arm independent through the buildout of AI datacentre capacity that followed, and marked one of the clearest instances of antitrust enforcement reshaping the AI hardware supply chain during the period.