Sam Altman has publicly accused Elon Musk of promoting unproven space-based data centers to public market investors, reigniting tensions between the two technology leaders over competing visions for artificial intelligence infrastructure. The OpenAI CEO's remarks on X targeted SpaceX's AI1 satellite program, which aims to build orbital computing nodes capable of delivering up to 150 kilowatts of peak power. Altman has previously dismissed the concept as impractical for OpenAI's near-term needs.

The public spat reflects deeper disagreements between Altman and Musk about how to scale AI computing. Altman told a podcast in February that space-based compute would not provide OpenAI with meaningful capacity within two, five, or even ten years, making the technology irrelevant to current business planning. His latest criticism came as SpaceX stock traded significantly below its post-IPO peak, closing at $145.30 on Friday, roughly 36 percent below its mid-June high of $225.64.

What's Behind the Altman-Musk Feud?

The escalating dispute emerged after Apple sued OpenAI, alleging the company stole trade secrets through two former Apple employees who joined OpenAI and allegedly took proprietary data with them. Musk seized on the lawsuit to renew attacks on Altman, accusing him of "stealing an open source AI charity" and misusing Apple's phone technology. Altman's response targeted SpaceX's orbital computing plans, suggesting Musk was using speculative technology to attract retail investors during a period of stock weakness.

Musk has outlined plans for a large-scale satellite network capable of supporting AI workloads, including up to one million compute satellites. The company announced that satellites would begin flying next year, according to posts on X. However, Altman's skepticism reflects a broader industry view that space-based computing remains years away from practical deployment at scale.

How Are AI Companies Responding to Cost Pressures?

Beyond the Altman-Musk feud, the AI industry is experiencing a significant shift toward cost efficiency. Three major AI developers have introduced new models designed to reduce operational expenses for enterprise customers:

• OpenAI's GPT-5.6: Designed to complete more work while consuming significantly fewer tokens, making the model more economical for enterprise customers.
• Grok 4.5 from xAI: Offers twice the token efficiency of comparable models from competitors, according to Elon Musk's AI company.
• Meta's Muse Spark 1.1: Positioned with pricing that CEO Mark Zuckerberg described as attractive, leveraging Meta's profitable advertising business to compete aggressively.

The emphasis on efficiency reflects growing concern among business customers about unexpectedly high AI bills. Some companies previously encouraged employees to use AI extensively through a practice known as tokenmaxxing, but tighter spending limits have since emerged after developers like Anthropic moved toward usage-based pricing rather than flat subscription fees.

"Enterprises are now examining both their spending and the value they receive from AI," said Sam Altman.

Sam Altman, CEO at OpenAI

One Paris-based AI startup CEO reported that several executives had accumulated substantial charges from using models supplied by OpenAI and Anthropic, with one business's monthly invoice reaching millions of dollars. In response, OpenAI has introduced credit-usage analytics and updated spending controls to help customers manage costs more effectively.

Which Companies Face the Most Pressure?

The changing pricing environment could place additional pressure on higher-cost developers while rewarding providers that deliver capable models more efficiently. Anthropic may face particular scrutiny because its Opus and Fable models rank among the most expensive on a cost-per-task basis, according to benchmarking service Artificial Analysis. Meanwhile, Chinese technology companies like DeepSeek have expanded the supply of lower-cost open AI models that, while trailing the most advanced U.S. systems, remain capable of handling many routine tasks.

OpenRouter, a model-routing service that allows users to select among hundreds of AI models for different assignments, raised more than $100 million in May as demand increased, signaling investor confidence in cost-conscious AI infrastructure. The shift toward efficiency-focused models suggests that affordability is becoming a more important competitive factor as companies examine whether their rising AI spending is delivering sufficient value.

The tension between Altman and Musk over space-based computing infrastructure reflects a broader industry debate about where AI computing will be deployed in the future. While Musk believes orbital infrastructure will eventually support AI workloads, Altman's skepticism suggests that terrestrial data centers and more efficient software will remain the dominant approach for years to come. For now, the immediate competitive pressure centers on reducing the cost of existing AI models rather than building speculative new infrastructure.