Pearl Blockchain Mining Network Found to be Doing 'Zero Useful AI Computation'

A recent study claims that Pearl, a Layer-1 blockchain, is not using its mining network for AI computation as it claims. The researchers estimate that the network uses around 320,000 RTX 3090-class GPUs, drawing an estimated 112 megawatts of power, but produces 'zero useful AI computation.'

The study found that instead of performing real inference or training, the network grinds through random matrix multiplications that resemble AI math. The researchers argue that the protocol verifies that miners performed matrix multiplication correctly, but does not verify whether the work came from a real AI workload.

The study also analyzed 8,012 workers in a single pool and found that all of them ran hardware capable of AI inference. However, the dominant mining binary contained no identifiable code for any machine-learning framework. The researchers estimate that around $600,000 per year in additional rental costs are borne by independent researchers competing for cheap hardware due to Pearl's mining activity.