RETROCADE RETROCADE
0
RETROCADE arcade machines, virtual pinball, racing simulator and parts banner

Computer hardware is becoming more expensive, from games consoles to smartphones, as demand for memory grows. The pressure comes largely from AI companies building data centres to support increasingly demanding large language models. While many consumers see soaring RAM prices as a problem, OpenAI chief executive Sam Altman has argued that keeping more memory in the cloud may be the more efficient option.

In a recent appearance on the Cheeky Pint podcast, Altman challenged the assumption that open-source AI will automatically move workloads onto personal computers.

Altman said many people want local AI inference and believe open-source models will let them run those systems on their own computers, reducing the need for large data centres. He argued that this assumption does not solve the underlying problem.

The argument sits within a wider dispute over how AI will be delivered. One possible future involves smaller open-source models operating on phones and laptops. The other concentrates powerful systems in large cloud facilities operated by major technology companies. Altman’s comments place him firmly in the second camp, at a time when those companies are also competing for limited supplies of memory.

Why Altman prefers centralised computing

Altman’s reasoning is based on utilisation. He suggested that having the world’s limited DRAM supply distributed across personal computers is inefficient when those machines may use the memory for only a small portion of the day. In his view, cloud infrastructure can keep that hardware working more consistently, which matters while memory remains difficult to obtain.

That position has uncomfortable implications for anyone who values local ownership of software and hardware. If more services move into the cloud, users could end up renting access to films, office tools and AI models rather than running them directly on their own devices.

The discussion was delivered in a relaxed podcast setting, including drinks and the sort of awkward beer-smashing moments that often accompany tech conversations about major changes to everyday life. The broader issue, however, is serious: a cloud-first AI economy could make access to computing dependent on ongoing service fees and the priorities of a small number of companies.

What this could mean for PC gaming

Under this vision, even demanding games could increasingly be streamed instead of installed locally. The example raised in the discussion is Grand Theft Auto 6, which could eventually arrive on PC and, in this hypothetical future, be played through cloud infrastructure rather than powered entirely by a player’s own machine.

That idea contrasts sharply with the traditional PC approach, where players choose their own components and upgrade them over time. Memory prices already matter to anyone planning a gaming build, while the growing demand from AI data centres adds another source of pressure. Players comparing hardware can still use RETROCADE’s Arcade Machine Buying Guide when considering a dedicated alternative to a conventional PC setup.

The same question applies to arcade enthusiasts. Local hardware remains central to the appeal of an upright cabinet, bartop system or racing setup, rather than relying on a remote service to deliver every game. RETROCADE’s arcade machines and arcade parts reflect that more hands-on model of ownership.

Whether cloud computing ultimately becomes the default for AI and games is unresolved. Altman’s view is clear: in a supply-constrained world, concentrating memory in heavily used data centres may be more efficient than leaving large amounts of RAM sitting idle in millions of personal computers.


Leave a Reply

Your email address will not be published. Required fields are marked *