You Can't Own A Computer Anymore...

Summary

Mutahar argues that rising computer-component prices and the growth of centralized AI services threaten people’s ability to own and control their computing hardware. He criticizes Sam Altman’s argument that local AI is inefficient because it leaves scarce DRAM sitting idle in personal computers much of the time, while cloud data centers could use it more intensively. Mutahar sees this view as serving AI companies’ business interests and favoring rented, company-controlled computing over local ownership.

Local AI and ownership

Mutahar contrasts cloud AI with running models locally. He says he uses local models—including Qwen and DeepSeek models, as named in the subtitles—on a computer with 128 GB of memory. He also describes using a smaller local model for web searches.

Local models may be less capable than leading services such as ChatGPT or Claude, and they require suitable hardware. However, Mutahar values their privacy, offline availability, and freedom from API fees, rate limits, and provider controls.

Memory and hardware costs

A major part of the video focuses on memory and hardware costs. Mutahar gives examples of steep prices for an RTX 5090 and DDR5 memory, arguing that RAM costs matter far beyond AI: memory is used in PCs, phones, consoles, cars, televisions, and other electronics.

He says memory manufacturing is difficult and concentrated among a small number of companies. The video also cites claims that OpenAI signed letters of intent with Samsung and SK Hynix for large DRAM supplies—described as 900,000 wafers per month and about 40% of global production. Mutahar criticizes the lack of regulatory scrutiny. These figures are presented in the video as reported claims, not independently verified in the subtitles.

Cloud services and digital ownership

Mutahar extends the ownership argument to cloud gaming and digital media. If games run only on remote servers, he warns, access or content—or even an entire game—could be withdrawn, making long-term preservation and offline play impossible. He connects this concern to self-hosting and Linux as alternatives that can give users more control.

Chatbots, safety, and reliability

Mutahar questions the value and safety of chatbots. He describes how language models generate responses through statistical prediction and argues that users can mistake fluent output for understanding or consciousness.

Citing a reported case involving a chatbot and a user’s delusions, along with other chatbot-related incidents, he criticizes companies for not doing enough to address risks. He also warns that AI-generated code can be unreliable unless users know how to review and test it, and objects to chatbots being promoted for sensitive areas such as parenting or personal information.

Reviews, guides, or tutorials

None. This is commentary and analysis rather than a product review or how-to guide.

Main speakers and sources

• Mutahar (SomeOrdinaryGamers): Main commentator.
• Sam Altman: Appears in an embedded clip discussing local inference and data-center efficiency.
• News articles and reports: Cited by Mutahar regarding DRAM supply commitments and chatbot-related incidents.