Fruit Fly Neurons Enter Bitcoin Mining’s Efficiency Race

Bitcoin mining is confronting the physical limits of silicon semiconductors alongside escalating energy demands, prompting researchers and hardware developers to explore unconventional computing architectures. One of the more unusual approaches involves simulating biological neural networks to execute or manage cryptographic calculations.

FutureBit, the consumer-focused hardware manufacturer known for localized mining devices such as the Apollo series, created HashFly as a browser-based experimental demonstration. The project uses a digital reconstruction of a fruit fly (*Drosophila melanogaster*) nervous system to perform simplified hashing operations on conventional computer processors.

HashFly maps 2,914 neural pathways from MaleCNS v1.0, a detailed digital wiring diagram, or connectome, of an adult male fruit fly’s brain and central nerve cord. Collaborative neuroscientific efforts to map insect biology produced these datasets, which are increasingly being used as structural blueprints for alternative computational models.

The present HashFly simulation runs far below the computational power needed to secure the actual Bitcoin network. Users can manually adjust its target difficulties, but the browser-based software operates at only a fraction of Bitcoin’s current mining difficulty. FutureBit said on the social media platform X that it plans to expand the simulation to include all neurons in the MaleCNS dataset, apply the SHA-256 algorithm, and publish its empirical findings.

A separate project, FlyMiner, uses biological simulation as a control system rather than mapping mathematical operations directly onto neural pathways. Its digital connectome contains 139,255 simulated fruit fly neurons and approximately 16.8 million synaptic connections. The system monitors activity in the insect’s movement-control neurons, and a predefined activity threshold triggers a standard external hashing engine.

That linked engine can run at up to 700,000 hashes per second and submits its attempts to CKPool, a prominent solo mining pool. In solo mining, operators independently compete to solve cryptographic blocks. A successful miner receives the entire block reward, currently 3.125 Bitcoins, instead of sharing smaller incremental payouts with a collective pool.

The efficiency theory behind biological computing comes from the low energy consumption of organic neural networks. FutureBit estimates that translating the simulated pathways onto physical, organic neurons could theoretically allow calculations at approximately one watt per terahash.

Traditional silicon-based mining measures efficiency in Joules per Terahash (J/T), a figure that directly corresponds to watts consumed for every trillion hashes performed per second. Modern Application-Specific Integrated Circuit (ASIC) miners built with state-of-the-art 3-nanometer semiconductor nodes operate between 11.5 and 16 J/T.

→

A hypothetical biological computing system operating at one watt per terahash would be ten times more energy-efficient than the industry’s most advanced silicon hardware. The model assumes that every neuron in a biological system remains continuously dedicated to solving mining puzzles while consuming only the metabolic energy required by the organism.

HashFly remains a conceptual simulation, while FutureBit’s commercial five-inch Apollo III ASIC achieves a physical hash rate of 18 terahashes per second (TH/s) with specialized silicon chips. The comparison highlights the gap between the experimental neural model and dedicated mining hardware without changing the simulator’s ability to run on ordinary computer processors.

The neural-network experiments form part of a broader history of unconventional and home-brew Bitcoin mining projects that avoid standard commercial data centers. Over the past decade, developers and businesses have reused obsolete hardware, redirected thermal waste, and pursued alternative energy sources.

In March 2021, IT security researcher Sebastian Staacks, who is known online as stacksmashing, modified a classic 1989 Nintendo Game Boy to mine Bitcoin. Because the handheld had no native internet connection, Staacks connected a Raspberry Pi Pico microcontroller through the Game Boy’s Link Port so a computer could transmit block data.

The modified console used its original Sharp LR35902 processor, which ran at 4.19 megahertz, to perform SHA-256 hashes. Its rate was just 0.8 hashes per second, making the estimated time to mine a single block several quadrillion years.

Thermal integration has also been used at a commercial scale. In June 2023, Bathhouse, a boutique spa in Williamsburg, Brooklyn, incorporated Bitcoin mining into its heating infrastructure. Heat exchangers routed the intense thermal byproduct from mining hardware into the spa’s pools and saunas, offsetting traditional utility heating costs.

Alternative power projects have focused on environmental remediation. In August 2023, Utah-based energy infrastructure company Nodal Power raised $13 million in a seed funding round to expand its landfill gas-to-energy facilities.

The plants capture fugitive methane emissions, a potent greenhouse gas, from municipal landfills and combust the gas to generate electricity. That power is either returned to the local utility grid or used on-site to operate modular, containerized Bitcoin mining facilities.