The moon is free-falling towards the earth but fortunately, the rate at which it falls to the earth is nearly equal to the rate at which the earth curves. In this answer, I will derive the formula for orbital velocity at the surface using pure geometry and later smartly generalize it. This will clearly establish the fact that the object is falling towards the earth but the earth is curving away sufficiently that the object does not hit the surface.
Building an intuition
If you drop a ball from a height near the surface, it will fall 5m towards the earth. If you throw a ball horizontally, as it moves in the horizontal direction, it will fall 5m towards to the earth too. If you give a small velocity, the ball will fall back to the ground.

In the second case, the ball travels sufficient distance horizontally that by the time it drops vertically, the earth's surface would've curved enough that it won't hit the surface of the earth. There is no magic with centripetal acceleration, it is simple geometry.
Deriving a formula for orbital velocity using pure geometry
For the sake of simplicity, let us consider throwing a ball off the surface of the earth as shown in the diagram instead of considering a moon orbiting the earth. However, note that, the surface can be any circle around the earth. We will use this idea to generalize the formula later.

In the above figure, a ball was thrown at a velocity $v$ from the surface horizontally. The ball travels distance $x$ in the horizontal direction and falls a distance $y$ in the vertical direction in one second (why one second? we can treat $x$ and $y$ as components of the velocity of the ball). On the surface of the earth, the ball would fall 5m in the first second.
Some quick observations before we begin:
- $BC = x$
- $\angle ACD =$ right angle (AD is diameter)
- $\angle ACB = \angle CDB$
- $\triangle ABC$ and $\triangle CBD$ are right triangles
- $tan(\angle ACB) = \frac{y}{x}$
- $tan(\angle ACB) = \frac{x}{2R - y}$
- $2R$ is significantly larger than $y$
$$tan(\angle ACB) = tan(\angle ACDB)$$
$$\frac{y}{x} = \frac{x}{2R - y}$$
I could have made use of calculus to derive but it isn't really needed as it would unnecessarily complicate the answer.
$$2R - y \approx 2R$$
To be honest, in the limit where $y$ tends to zero, this is not an approximation.
$$\frac{x}{y} = \frac{2R}{x}$$
Solving for $x$ in terms of the other variables, you get,
$$x = \sqrt{2Ry} - (1)$$
We had considered $y$ as the distance the object would travel in one second. We can write the same quantity in terms of $g$ on the surface.
$$y = \frac{g}{2} - (2)$$
Substituting $(2)$ in $(1)$, we get,
$$x = \sqrt{Rg}$$
We had earlier assumed that $x$ is the distance traveled by the ball in one second. Therefore, $x$ is the horizontal velocity.
$$v = \sqrt{Rg} - (3)$$
If you have studied gravitation, you'd immediately notice that the above formula gives the orbital velocity of objects orbiting the earth near the surface.
Smartly generalizing the above result
I had avoided deriving a formula for a general case where the object is a distance $r$ from the surface of the earth to avoid lengthening the answer. However, the above formula can be generalized easily.
We used $R$ as the radius of the earth. Well, the surface of the earth needn't really be the surface of the earth. Have we used the fact that $R$ is the radius of the surface anywhere other than for using $g$? All we said was that the ball shouldn't hit the surface. The surface could have been a virtual one, i.e: any circular orbit. We can rephrase the previous sentence as the ball should hit the orbit of radius $R$. We can generalize $R$ as $r$. We used $g$ as the gravitational acceleration at the surface of the earth. We can generalize it for any radius $r$.
$$g = G\frac{M}{R^2}$$
can be generalized to
$$a = G\frac{M}{r^2} - (4)$$
Substituting the $(4)$ in $(3)$, we get,
$$v = \sqrt{ra}$$
$$v = \sqrt{Gr\frac{M}{r^2}}$$
$$v = \sqrt{\frac{GM}{r}}$$
The above formula gives the orbital velocity for an orbit at a distance $r$ from the center of the earth.