Tesla Caps Employee AI Spending at $200 Per Week

Tesla has imposed a $200-per-week spending limit on employee AI tool usage, effective July 6, according to an internal memo first reported by The Information. The cap represents a sharp reversal for a company that spent the past six months encouraging staff to adopt AI more aggressively.

Why it matters

Tesla's valuation hinges on its ability to deploy AI at scale for autonomous vehicles and humanoid robots. If the company struggles to manage weekly AI costs for its engineering teams, it raises questions about executing far more ambitious AI infrastructure plans. The policy also reveals how Musk is using internal spending controls to steer employees toward his own AI venture while they prefer competing tools.

From adoption push to spending brake

Over the past six months, Tesla leadership worked to consolidate employee AI usage onto a companywide platform with approved models and security policies. Some teams built internal dashboards that ranked employees by token consumption to encourage adoption. The strategy succeeded beyond expectations: software engineers were consuming thousands of dollars worth of tokens each week, according to people familiar with the usage patterns.

The new policy requires manager approval for spending above the $200 threshold. The cap mirrors similar moves across corporate America. Uber imposed a $1,500 monthly limit after exhausting its entire 2026 AI budget by April. Meta, Amazon, and Walmart have all introduced spending caps or pushed workers toward cheaper models as token-based billing exposes the true cost of AI adoption.

The xAI exception

The most significant detail in Tesla's policy is what it excludes. The $200 limit does not apply to beta versions of xAI products, effectively steering heavy users toward Musk's AI company rather than competitors.

Musk has spent months pushing Tesla staff toward tools connected to his other ventures. After his AI lab began working with Cursor in April, he emailed the entire company encouraging employees to try Composer, Cursor's coding model. SpaceX is now acquiring Cursor's parent company Anysphere for $60 billion in an all-stock deal expected to close this quarter. Tesla engineers have also served as early testers for unreleased versions of Grok and Composer.

Despite the internal promotion, Grok remains unpopular among Tesla staff, with many engineers preferring Anthropic's Claude instead, according to people familiar with the matter. Musk himself acknowledged xAI was "not built right" weeks after Tesla invested $2 billion in the company.

AI as the core thesis

Tesla has expanded AI usage beyond engineering. The company released Nova, an AI tool trained on internal data, to standardize practices from holiday lookups to factory troubleshooting. Vehicle engineering VP Lars Moravy said Tesla is integrating AI into engineering through agents with access to company expertise and using AI to detect vehicle defects on production lines.

The stakes are high because Tesla's valuation now depends on AI rather than car sales. Musk has stated the company's future value rests on deploying AI across its Robotaxi network and Optimus humanoid robot as revenue growth has stalled over the past two years.

Tesla has also tightened AI security. Starting in spring, the company restricted access to models outside its internal "Bottle Rocket" platform on company devices and held sessions warning staff against feeding confidential data into non-approved systems.

The Information first reported these details from internal company communications and interviews with current and former employees.