Nearly each week one other AI firm declares an even bigger mannequin, an even bigger knowledge middle, and an even bigger electrical energy invoice — whereas pocketing billions and quietly strolling again its local weather pledges. It’s fairly disheartening background noise. Then alongside comes CrankGPT, a tiny chatbot that refuses to reply your query until you bodily generate the electrical energy your self by turning a hand crank. And actually, I sort of adore it.
What CrankGPT Truly Is
CrankGPT is a undertaking from Squeez Labs — a voice assistant that runs completely on a Raspberry Pi 5, powered by a 20W hand-crank generator. No Wi-Fi or cloud. No knowledge middle. Simply your arm, a crank, and an area language mannequin doing inference on-device.
The tech stack is impressively considerate for one thing this small. It runs llama.cpp with compact fashions like Liquid AI’s LFM2 350M and Gemma 3 1B, utilizing Moonshine for speech recognition and Piper for text-to-speech. The entire thing boots in about 30 seconds from a chilly begin.
Squeez Labs constructed a customized capacitor board to clean out the generator’s voltage spikes — as a result of when the LLM and speech synthesis fireplace collectively, the crank actually will get tougher to show. You’re feeling the inference load in your arm.
That final half is the entire level.
Why the Inconvenience Is the Characteristic
Squeez Labs isn’t pretending that is sensible. They’re making a degree: the price of AI is actual, you simply don’t see it.
Proper now it’s fully invisible. You sort a immediate, get a solution, and someplace a server farm attracts extra energy. World knowledge middle electrical energy consumption hit 460–490 TWh in 2025, up 17% 12 months over 12 months. AI infrastructure alone drove a 50% vitality surge — rising at 5 instances the speed of total world electrical energy demand. The IEA tasks that quantity will double by 2030.
In the meantime, tech corporations that after posted bold local weather pledges have quietly began constructing gas-burning energy vegetation to feed their GPU clusters. The 5 largest hyperscalers dedicated roughly $660–690 billion in capex in 2026, with about 75% tied on to AI infrastructure. That cash got here from someplace. Principally from you, paying for subscriptions and API calls at margins that make the underlying vitality value look trivial.
CrankGPT makes the price seen once more. You need a solution? Then crank, child.
The Broader Shift Towards Native AI
What Squeez Labs constructed as a provocation, lots of people are constructing severely. Enterprise AI inference has shifted dramatically, with 55% of enterprise AI inference now operating on-premises or on the edge — up from simply 12% in 2023. Privateness is the principle driver. Working a mannequin regionally means your knowledge by no means leaves your system, which issues for healthcare, authorized, and anybody who’d desire their most private questions keep that means.
The {hardware} has gotten ok to make this sensible. Squeez Labs notes that smaller fashions on modest {hardware} can deal with a stunning vary of duties — and that the business’s reflex to achieve for the most important, costliest mannequin is usually simply laziness dressed up as high quality.
The Uncomfortable Level
I’m not suggesting you swap ChatGPT for a hand crank—although that is perhaps enjoyable, for some time. However I believe CrankGPT is onto one thing actual. We’re so used to “free” AI that it’s simple to neglect each immediate consumes precise assets someplace — electrical energy, water, land, capital. Tech executives are gaining extraordinary income whereas these prices get externalized onto the grid, the local weather, and everybody who pays an electrical energy invoice.
A undertaking that makes you’re feeling that value in your bicep is a extra trustworthy interface than a clear white textual content field.
Squeez Labs isn’t promoting CrankGPT — they open-sourced the entire construct for underneath $300 in elements. That’s sort of the purpose too. The reply to resource-extractive AI isn’t extra resource-extractive AI. Typically the correct transfer is to make one thing smaller.
Lauren has been writing and modifying since 2008. She loves working with textual content and serving to writers discover their voice. When she’s not typing away at her laptop, she cooks and travels along with her husband and two youngsters.
