Elon Musk has revealed a new artificial intelligence initiative called “Digital Optimus,” also referred to as “Macrohard,” describing it as a joint project between xAI and Tesla that aims to create a computer-controlling AI agent powered by the Grok system.

Musk shared details of the project in posts on X, saying the initiative forms part of Tesla’s $2 billion investment agreement with xAI.

According to Musk, Digital Optimus is designed as a two-layer AI architecture. Grok will act as the “master conductor” with a deep understanding of context and the wider world, while a Tesla-developed AI layer will analyse and respond to the most recent few seconds of computer screen activity along with keyboard and mouse inputs.

Musk compared the design to the dual-process theory described by psychologist Daniel Kahneman. In this analogy, Tesla’s system performs rapid, instinctive responses similar to “System 1” thinking, while Grok handles more deliberate reasoning comparable to “System 2”.

He also said the system would run primarily on Tesla’s AI4 chip, which he estimated costs around $650, combined with high-performance cloud infrastructure based on chips from Nvidia used by xAI.

Musk described the technology as potentially enabling real-time intelligent control of computer interfaces and suggested that, in principle, such systems could eventually emulate workflows carried out by entire companies.

The announcement marks a shift from Musk’s earlier comments in September 2024, when he said Tesla had “no need to license anything from xAI.” It also comes amid ongoing legal disputes in which some Tesla shareholders have accused Musk of breaching fiduciary duties related to the creation of xAI.

Neither Tesla nor xAI has yet released technical documentation or a development timeline for the Digital Optimus project.