Qubic, a project that has been flying under the radar in AI-focused communities, is making waves with its unique approach to distributed compute and AI training. The company’s Useful Proof of Work concept utilizes hardware to solve real-world problems, such as training neural networks, instead of wasting energy on random hash puzzles.

The Qubic network has been independently verified to process an impressive 15.52 million transactions per second by CertiK on live mainnet, surpassing Visa’s theoretical peak throughput. This is made possible by the architecture’s ability to run on bare metal hardware without a virtual machine layer, enabling unprecedented throughput.

Qubic is also set to launch a DOGE mining integration, which will allow their infrastructure to run Dogecoin mining simultaneously with other workloads. This will be made possible by the parallel operation of ASIC hardware for DOGE Scrypt mining and CPU/GPU hardware for other workloads.

What sets Qubic apart from other projects, such as Bittensor, is its focus on using distributed compute to train models from scratch with raw hardware power. This approach has the potential to revolutionize the field of AI research, as it could enable the development of more sophisticated and powerful AI models.

The concept of mining-funded distributed AI training is an interesting one, and it will be exciting to see how it is received by the AI research community. As big companies continue to pour billions into building massive data centers and training larger language models, Qubic’s innovative approach may just be the key to unlocking true AGI.

Photo by Tara Winstead on Pexels
Photos provided by Pexels