(I decided not to wait longer. A couple of hours is long enough.)
You can use Tellegen's. And it doesn't matter how you decide to estimate the power in each branch shown. You can decide that a current moving up is positive, or negative, and you can decide that a voltage with the + on top is positive, or negative. Doesn't matter. Always get the same answer.
Let's start by assuming that the power is equal to the top-to-bottom voltage difference times the current, where a current moving from top-to-bottom is also positive. All the branches, including the Box 1 branch, must have \$+25\:\text{V}\$ across them. So in this case, you have:
$$+25\:\text{V}\cdot\left(1\:\text{A}-5\:\text{A}-I_x+2\:\text{A}\right)=0\:\text{W}$$
(I've collected the common voltage to the left, above. But feel free to distribute it across the currents so that you cast it directly in terms of Tellegen's. \$V_i=V\$, as \$V\$ is common to all of the branches. So collecting it to the left is fine in order to simplify \$\sum V_i\cdot I_i = V\sum I_i=0\:\text{W}\$.)
That solves as \$I_x=-2\:\text{A}\$. If you assume that a positive current is one that goes from bottom-to-top then:
$$+25\:\text{V}\cdot\left(-1\:\text{A}+5\:\text{A}+I_x-2\:\text{A}\right)=0\:\text{W}$$
That solves as \$I_x=-2\:\text{A}\$.
And I don't suppose I need to tell you that if you decided that the voltage across each branch was instead negative, that you would still get the same result: \$I_x=-2\:\text{A}\$.
Of course, KCL would also tell you this. If \$ V\sum I_i=0\:\text{W}\$ and \$V\ne 0\:\text{V}\$ then \$\sum I_i=0\:\text{A}\$, which is KCL. But if all you have is Tellegen's and the \$P=V\cdot I\$ formula, then you are still fine.