Arm is set to release its own chip before the end of this year, having reportedly secured Meta as its first customer for the project.

According to a report in the FT, the chip is expected to be a data center CPU, built on a customizable base. Its production will be outsourced to manufacturers such as TSMC.

The move marks a departure from Arm’s traditional business model of designing and then licensing out its chip IP to companies, including Apple, Google, Microsoft, Nvidia, and Qualcomm, the latter of which Arm is already involved in a legal battle with regarding an alleged breach of licensing terms.

By launching its own silicon, Arm would be in direct competition with some of its customers, a position chip companies are unlikely to welcome.

Reports that UK-based Arm was developing its own chips first surfaced in May 2024, when it was alleged that the company had created a chip division with the aim of producing a prototype product by Spring 2025, with mass production expected to follow in the Fall of that same year.

SoftBank, the Japanese conglomerate that bought Arm in 2016 for $32bn and owns a 90 percent stake in the company, was reportedly already negotiating with TSMC and other unnamed companies in order to secure production capacity.

According to the FT, SoftBank’s rumored Ampere takeover considerations are “central” to Arm’s chipmaking project. Citing people familiar with the plans, the deal could see Ampere valued at $6.5bn. In addition to that alleged acquisition, Masayoshi Son's SoftBank has recently pledged to invest $19bn in OpenAI's $500bn Stargate AI data center project.

Arm declined DCD's request for comment, while SoftBank did not immediately respond.

Earlier this week, it was reported that Meta was in talks to acquire South Korean AI chip startup FuriosaAI to support its own internal chip-making efforts.