Apple has reportedly reduced its daily AI spending limit for some engineers from $300 to $200, according to reporting cited by Stork AI Daily. The internal budget cut reflects what the source describes as a fundamental challenge: Apple’s AI expenditures are threatening to exceed allocated budgets despite the company’s substantial financial resources.
The reduction underscores a broader economic reality facing large technology companies. According to the source, uncapped access to advanced AI models creates an exponential cost curve that can damage profit margins even for highly profitable companies. This is not primarily a liquidity issue but rather a structural problem with the economics of generative AI deployment at enterprise scale.
The implications extend across the industry. The source argues that companies relying on “brute-force inference without strict, automated cost controls” are building unsustainable business models. As AI becomes more expensive to operate, competitive advantage will increasingly depend on efficiency rather than raw capability—through intelligent caching, routing to smaller models, and prompt optimization.
In related developments, researchers at the University of Michigan found significant disparities in AI resource consumption. Non-English prompts consume up to 179 times more energy than equivalent English prompts, according to their findings, often while producing lower accuracy. This disparity stems from tokenization systems optimized for English, creating unpredictable costs for companies serving global user bases.
Elsewhere in the AI sector, Palo Alto Networks is reportedly in talks to acquire AI security startup Clutch for approximately $500 million. According to the source, the acquisition aims to enhance control over AI agents and manage their access to internal systems—reflecting enterprise concerns about autonomous agents accessing sensitive data.
OpenAI has also launched GPT-6 Cyber, a restricted model designed to identify software vulnerabilities and generate exploits. Access is currently limited to approved security testers, keeping the offensive capabilities out of public availability for now.
Key facts
- Apple has reportedly cut daily AI spending caps for some engineers from $300 to $200
- University of Michigan researchers found non-English prompts consume up to 179 times more energy than English prompts
- Palo Alto Networks is reportedly in talks to acquire Clutch for approximately $500 million to enhance AI agent control
- OpenAI launched GPT-6 Cyber, restricted to approved security testers, for vulnerability identification and exploit generation