Sources: X employees are increasingly being asked to spend time on xAI initiatives; xAI is using a data center once leased by Twitter, but now leased by Tesla
Tesla is fast running out of early adopters, but its cars are still too expensive for most buyers Threads: Parker Thompson / @parkert : Elon has always had a disregard for shareholders, which worked out when everything was going up, but feels like a much bigger liability when Twitter is already collapsing and Tesla is more than due. Adam Cook / @motorcityadam : Honestly, all things considered, I much rather Tesla employees work on xAI than on, say, FSD Beta - if that is what is happening here. Tesla is not organizationally competent in safety-critical systems. Full stop. And while I do believe that these “chatbot AI” systems DO have safety components associated with them - it seems less hazardous for now. … Casey Newton / @crumbler : From people to servers, Elon Musk is borrowing from across his empire to build Grok. @reporterzoe and I report: https://www.platformer.news/ ... Zoe Schiffer / @reporterzoe : NEW from me and @crumbler: X employees are increasingly being asked to spend time on xAI initiatives, and xAI code is sitting on Tesla servers that operate in a data center that (stay with me here) was previously leased by Twitter: https://substack.com/...
Context & Ripple Effects
xAI had only recently been established as a Musk-led lab drawing on people with experience across Tesla and other major technology companies. This report shows its early operating model taking shape through resources already inside Musk’s corporate network, rather than as a fully separate buildout.
That arrangement foreshadowed later reporting that xAI drew on affiliated companies’ talent, data, and GPUs, an approach that became a focal point for Tesla-investor concerns in the reported sharing of xAI resources across Musk businesses.
First-order effects
- X staff face competing priorities as xAI work is added to their responsibilities, while xAI gains access to operational knowledge and labor without building every function internally.
- Using a Tesla-leased facility for xAI code places Tesla infrastructure in xAI’s development path, making allocation and governance of compute more consequential for both companies.
Second-order effects
- Tesla investors and management have a clearer reason to scrutinize whether company personnel and infrastructure are supporting a separate Musk venture, especially as later reporting tied such resource sharing to investor disputes.
- The arrangement reduces xAI’s near-term dependence on standalone hiring and data-center capacity, but it also makes delivery at X and Tesla more exposed to shared-resource bottlenecks.
Third-order effects
- If cross-company resource sharing persists, Musk’s businesses may operate more like an interconnected AI stack than independent companies, raising durable questions about cost allocation, fiduciary oversight, and execution accountability.
- The case is an early example of compute execution risk: access to facilities and staff can speed an AI lab’s launch, but unclear control over those inputs can create governance friction as the lab scales.
The trend: AI development is increasingly being organized around control of compute, talent, and data across corporate networks rather than within neatly separated business units.