Yeah, but you gotta destroy a lot of them...

Numbers win wars. And if the containerized payloads successfully conceal what the ship is carrying, the enemy won't specifically be able to target the weapons carriers. To guarantee that, they have to wipe out the whole force. But, the concept calls for as many forces like this as we can flood the theater with. So, they have to wipe out multiples of this.

The obvious "weak point" is the manned "heart" of it all. But, it will be defended by its escorts, as an aircraft carrier is by its. And, if the manned ship has released the firing command to the USVs, they are autonomous. They then are dangerous without the manned ship. The only real reason we want a "man in the loop" in the first place is to govern the decision to fire. Nobody should want AI deciding when to shoot. Once the shooting starts, though, the armed USVs will be dangerous (possibly even to our own side...!) I can see a situation where the manned ship releases the firing command, then exits the picture completely. The "programming" then runs the mission from there. The "minimal manning" part of the concept has a small crew aboard one of the USVs itself. They would be there to do things like release the firing command. Then, they could leave on a RHIB to be picked up far away. If this were used, the enemy would then also have to find the manned ship among the rest of them.

It's about volume. The more of these we have, the better. The sum is vastly greater than the parts.

I agree. Payload development is another area where we need to "get to work."