Startup Pitches Old Macs as Paid Nodes for AI Inference Network

A new distributed computing scheme called Darkbloom wants Mac owners to rent out their machines' idle processing power to run artificial intelligence inference workloads, turning dust-gathering desktops and laptops into paid nodes on a shared network, The Register reports.

The pitch mirrors a wave of projects that treat spare compute like a commodity: rather than buying dedicated GPU clusters, an operator recruits a swarm of consumer devices and stitches them together so AI models can be queried across the pool. For Apple hardware in particular, the appeal is that even older Macs carry unified memory and capable chips that sit unused for most of the day, making them a plausible, if modest, source of inference capacity.

The Register's own framing captures the pitch's tone: hook up underutilized Apple kit to the network and let it work as part of what the outlet calls an AI inference "Borg," collectively assembling scattered machines into a larger computing organism.

Schemes like this typically hinge on trust and economics that are hard to verify from the outside — how much a contributed Mac actually earns, how workloads get distributed fairly, and what happens to a device's performance and power bill while it's donating cycles to someone else's AI queries. Those are the questions any owner weighing whether to plug an old Mac into Darkbloom's network will want answered before signing up.