Both N-channel JFET and diode modulators are possible. In this digital age, you might note that the carrier signal (in this case 100 kHz) is a square wave...it might be generated from a logic gate or some other digital source.
Be aware that this is a very non-linear modulator, whose output resistance varies. As in Andy's circuit, a linear buffer stage should follow (high input impedance, low output impedance). Buffering and filtering are not shown here.
These are both shunt modulators, where the JFET or diode short the analog signal to ground for half-a-cycle of the digital carrier signal. Many harmonics are generated, which must be attenuated with a selective (inductor/capacitor) resonator.
The diode modulator has superior performance, with far less distortion. The difficult-to-suppress distortion products at 98 kHz and 102 kHz are well-attenuated in the diode modulator; the JFET modulator attenuates these distortion products less-so.
I have shown a 100kHz modulator rather than AM radio signals at 1MHz because LTspice simulations run faster at lower carrier frequencies. Since circuit impedances for the diode modulator are lower than JFET modulator, circuit performance at high carrier frequencies should be superior for the diode modulator.
Note that this kind of digital modulator could substitute an open-collector or open-drain transistor for V5 + series D1...something like 74HC05. A microcontroller with GPIO set for open-drain should work well, however R3 might be made somewhat larger-value.