4

Days

15

Hours

59

Mins

9

Secs

Reports from the Financial Times indicated a major reshuffle within Elon Musk’s AI efforts, stating that Ashok Elluswamy left Tesla to join SpaceX/xAI.

That account is incomplete. Elluswamy remains Tesla’s Vice President of AI Software, but his mandate has grown substantially. He is now leading MACROHARD (also referred to as the Digital Optimus project), serving as lead engineer for a large-scale joint initiative that tightly integrates engineering teams at Tesla and xAI.

This week on X, Elluswamy began recruiting top-tier engineers to join either organization to help advance the work.

MACROHARD

Announced earlier this week, the effort aims to build an enterprise-grade digital worker. The internal codename, MACROHARD, is a direct and humorous jab at Microsoft.

While Microsoft supplies software that helps businesses operate, Elon Musk envisions MACROHARD as an AI that can mirror an entire company’s operations. It is intended to autonomously sign in to a computer and perform the office work typically handled by a human employee.

The Brains and the Hands

To achieve this, MACROHARD employs a large two-part architecture that draws on the strengths of both of Musk’s AI organizations.

xAI’s Grok functions as the central decision-maker—the “brains”—using broad, macro-level reasoning and world knowledge to specify the tasks to be completed. Tesla’s physical AI agents, whether a Model Y or an Optimus, perceive their environment, process data, and physically manipulate real objects.

Blurring the Corporate Lines

Elluswamy’s parallel recruiting across Tesla and xAI underscores how quickly the boundaries between Musk’s companies are dissolving.

What began as a software collaboration is becoming a deeply integrated technical and financial ecosystem. Tesla recently invested roughly $2 billion into xAI, formally cementing the relationship.

By steering MACROHARD across Tesla AI and xAI, Elluswamy is building shared digital infrastructure to power Tesla’s physical Optimus robots, the upcoming Robotaxi fleet, and xAI’s large enterprise software ambitions.

Musk also said the software will run workloads concurrently on Tesla’s in-house AI4 chips—making use of idle vehicles and dedicated Supercharger nodes—alongside xAI’s large Nvidia-based server clusters.

Under Elluswamy’s leadership, Tesla and xAI are merging hardware and software capabilities into a unified, distributed supercomputer.