1. The size of the phased array is directly related to the distance over which you want to maintain focus. In order to focus a beam across a room, you need a hilariously large array.

2. Ultrasound transducers are very inefficient, so you’ll lose a lot of power.

3. Ultrasound transmission will be blocked by a person, or even by a sheet of paper in the way.

4. Even if you do achieve a tight beam, then you either have to actively steer it around, or it’s only good for transmitting power to something sitting still in a specific spot.

5. We don’t really know much about the effects of having watts of ultrasound present in a space with humans 24/7/365.

(And these are just the easy ones)

Decent efficiency seems possible. These people report 53% efficiency at 1 meter:

Roes, M.G.L.; Hendrix, M.A.M.; Duarte, J.L., "Contactless energy transfer through air by means of ultrasound," IECON 2011 - 37th Annual Conference on IEEE Industrial Electronics Society , vol., no., pp.1238,1243, 7-10 Nov. 2011

Abstract: An alternative approach to the wireless transfer of energy is proposed, employing acoustic waves in air. Unlike conventional methods, acoustic energy transfer is able to achieve energy transfer at high efficiencies over distances that are large in comparison to the dimensions of the transmitter and the receiver. This paper gives an overview of the principle and explains the different loss mechanisms that come into play. A theoretically limit on the achievable efficiency is calculated. It exceeds that of a comparable inductively coupled system by an order of magnitude. First preliminary measurements indicate that AET is feasible, although the measured efficiency is lower than the predicted theoretical limit.

They were using small low-power transceivers as their focus (no pun intended) was on large distances compared to transceiver size, and so were only dealing with a few dozen uW power.

Looking at a few of the things that cite them, it seems that low power is where most of the work in ultrasonic power transmission is, aiming for environments where you need to power something small in a location that makes the usual methods problematic, such as implanted biomedical devices. Example: "MEMS Based Broadband Piezoelectric Ultrasonic Energy Harvester (PUEH) for Enabling Self-Powered Implantable Biomedical Devices", https://www.nature.com/articles/srep24946

Repeat after me: directional radiation is still subject to inverse square law.

Agreed in the general case, but what about a laser?

Only in the far field.

This doesn’t invalidate the important claim here. In your example you are focusing sound not by concentrating it along a beam but by phasing out the signal that is not on the path.