Tesla also begins token "traffic limiting"

```

After aggressively promoting internal AI adoption, Tesla abruptly hit the brakes, setting a spending cap on employee use of AI tools, reflecting the increasingly prominent challenge of balancing AI investment and cost control for businesses.

According to a recent report by The Information, Tesla informed employees last month that starting July 6, the weekly spending cap for employee use of AI tools will be set at $200, with any excess requiring supervisor approval. In previous months, some software engineers’ AI token consumption reached several thousand dollars per week. Sources cited in the report note that test versions of xAI products are not included in this limit.

This shift happened as Tesla accelerated company-wide AI adoption, mirroring the trajectories of companies like Meta, Uber, and Walmart—all of which experienced a rapid transition from encouraging employees to embrace AI fully, to tightening related expenses.

For Tesla, this adjustment is particularly noteworthy. Elon Musk has repeatedly emphasized that Tesla’s future value depends on whether it can achieve large-scale AI implementation in the Robotaxi network and the Optimus humanoid robot, rather than relying solely on car sales. Against this backdrop, how to improve the efficiency of AI investments has become a key issue for management.

From Personal Accounts to Unified Control: Tesla Tightens AI Tool Access

According to four sources, Tesla launched an internal AI unified access platform last year called "Bottle Rocket", giving employees access to models like OpenAI, Anthropic, xAI, and Cursor, including some versions not yet publicly released. Previously, many employees used various AI tools mainly through personal accounts.

However, after this platform went online, company-level policy on AI use remained scattered, with guidelines primarily set by vice presidents or directors of each business unit, lacking standardized regulations.

It was not until this spring that Tesla began promoting unified company-wide management, including restricting employees from accessing AI models outside of Bottle Rocket using company computers and internal networks, and organizing internal communications to remind employees not to input company confidential information into unapproved AI systems.

Regarding personnel, former IT VP Raj Jegannathan once led the promotion of AI within Tesla, expanding AI applications from R&D to sales and service—including deploying AI customer service agents. However, months before his departure, some of his responsibilities were reassigned. After Jegannathan left in February this year, Tony Tran began reporting directly to Musk, overseeing IT, AI, and cloud infrastructure.

AI Tool Adoption Uneven, Limited Acceptance of Grok Internally

Tesla’s push for wider AI tool adoption has not been smooth sailing.

Earlier this year, some teams launched internal dashboards to track token use, encouraging engineers to use AI more and ranking employees by token consumption within each department. Meanwhile, management continually reminded staff to control usage costs and handle sensitive data carefully.

Elon Musk himself continues to promote his companies’ AI products among employees. In April, as xAI and Cursor began deep collaboration, Musk sent a company-wide email urging employees to try Cursor’s programming model Composer. In June, he stated that SpaceX and Tesla are testing the latest xAI model, Grok 4.5.

However, according to sources, Grok’s acceptance within Tesla is not high, with many employees still preferring Anthropic’s Claude for daily development tasks.

Systematic Promotion of Company-Wide AI Application, Nova Platform as Internal Unified Engine

Tesla’s AI deployment is not limited to software engineering teams.

Last year, the company launched the Nova AI platform based on internal data training, and continues to iterate and upgrade it. Nova aims to provide unified knowledge and process support throughout Tesla; employees can look up daily information such as vacation policies or use it to assist with more complex workflows like troubleshooting factory production line faults.

Tesla’s VP of Vehicle Engineering Lars Moravy recently said in an interview that the company is actively integrating AI into engineering development processes, including using AI agents to access engineering knowledge bases and using AI to detect quality issues in off-line vehicles.

Overall, Tesla is attempting to systematically advance company-wide AI application. But as AI adoption grows, balancing efficiency, controlling investment costs, and ensuring data security is becoming a management issue faced by more and more large enterprises.

Risk Warning and DisclaimerThe market carries risk, and investment should be approached cautiously. This article does not constitute personal investment advice and does not take into account individual users' unique investment goals, financial situation, or needs. Users should consider whether any views, opinions, or conclusions in this article suit their particular circumstances. Investments made based on this are at their own risk. ```