Darkbloom Warns Idle-Mac AI Inference Earnings Have Fallen to $1.23 a Day

Darkbloom, a network that pays people to rent out their idle Macs for AI inference work, is warning participants that the economics behind the idea are deteriorating fast. According to Tom's Hardware, earnings on the network have fallen to roughly $1.23 a day per machine, driven by a sharp drop in the price of the tokens those machines produce.

The reported cause is a 77% decline in Anthropic token pricing. For a network like Darkbloom, token prices are effectively the revenue line: contributors supply local compute, the output is model inference priced in tokens, and when that price falls, per-device payouts fall with it.

The pitch behind such networks has been a staple of the AI boom — turn hardware you already own into a small stream of income by letting it take inference jobs while it would otherwise sit idle. Macs, which are numerous and often left powered on, have looked like a natural fit for that model.

A payout of about $1.23 a day suggests the arithmetic no longer works. If token prices fall faster than the cost of the hardware and electricity that produce them, the margin that makes idle-compute networks viable disappears.

What makes the report notable is who is delivering it. This is not an outside critic arguing that distributed inference is overhyped; it is the network itself flagging collapsing returns, which suggests operators are seeing a structural shift in pricing rather than a temporary dip.

The direction of travel is good news for developers building on AI models, since cheaper tokens lower the cost of running applications. It is much worse news for anyone whose business is reselling raw capacity, where the same price decline eats directly into the spread they depend on.

The next things to watch are how Darkbloom adjusts — whether it rewrites its payout rules, shifts to different workloads, or shrinks — and whether other idle-compute networks report similar compression. If they do, distributed inference may have to justify itself on something other than price.