Earlier this month, DARPA released a call for proposals addressing a key challenge in the brain-machine interfaces (BMI) – the reliability of cortical electrode-tissue interface. As seen in the chart above (taken from this DARPA presentation by Prof. Jack Judy), existing intracortical electrode arrays (such as BrainGate) can extract a lot of information from the cortex (1500 events/s and more) but their performance drops off by 50% in about one year after their implantation. By 3-5 years after implantation, performance further deteriorates to a point where their informational flow is no longer above that of peripheral nerve electrodes. The DARPA initiative aims to explore novel revolutionary approaches to improve the long-term reliability of neural recordings to sustain high information flow needed for controlling an artificial hand or arm. The ultimate goal of the initiative is to develop the intracortical array that can provide the life-long information flow of 2000+ events/s to control the 22-degrees-of-freedom artificial arm recently developed by DARPA. Achieving this very ambitious goal will likely require a concerted effort of multiple research groups working together on different aspects of the problem, ranging from the design of novel biocompatible and neurotropic/ immuno-suppressive electrode materials to development of robust non-linear state-dependent decoding algorithms and advanced techniques for device packaging and wireless telemetry.