From the picture of the 9mm by 9mm tile, it looks like it has about 30 electrodes, each something like 0.1 mm wide maybe? (The article says they are "hair-thin"). They are obviously large enough to be somewhat stiff and have a fixed position on the tile. Presumably they are implanted by pushing the tile onto the brain surface.
According to the paper Neuralink published last year[1], Neuralink's system has 3,072 electrodes per array distributed across 96 threads. The threads are flexible and 4 to 6 micrometer wide, and are inserted individually by a robot that avoids hitting blood vessels.
Yes, exactly. So, why didn't they just use a Neuralink instead? Now it feels like they unveiled the Amiga a few months after having unveiled some Alienware laptop.
Edit: no, the real reason why the Neuralink could not be used here, and the answer to the previous comment asking for the difference between the devices, is simply that the Neuralink is, for now at least, purely a sensor, whereas the aim with this vision-restoring device is to stimulate the cortex (so that would require different currents, electrodes, battery life, etc), which the Neuralink cannot do (in its current version).
In the neuralink paper which I linked above, they say at the end of the "Discussion" section that the system is also designed for stimulation on every channel, but that they have indeed not yet demonstrated that capability. Basically the difference is that Neuralink is a general BMI with some interesting innovations that is still in the research phase, while the Gennaris system is a specific application that is about to enter human clinical trials.
According to the paper Neuralink published last year[1], Neuralink's system has 3,072 electrodes per array distributed across 96 threads. The threads are flexible and 4 to 6 micrometer wide, and are inserted individually by a robot that avoids hitting blood vessels.
[1] https://www.biorxiv.org/content/10.1101/703801v3