Your reasoning is only partly correct.

If the data is truly linear, then each additional bit gives about 1 stop (1 EV) of encoded range. So an ideal 8-bit linear recording can represent about 8 EV from black to saturation.

But that mainly applies to linear data. JPEGs are usually 8-bit with a gamma/tone curve applied, so they can visually represent more than 8 stops—roughly around 12 EV was noted in the answers—because the code values are distributed nonlinearly.

For raw files, cameras typically store linear sensor data at 12-bit or 14-bit, not 8-bit. So raw capture is more typically on the order of 12 or 14 EV of recordable range in this simplified bit-depth sense.

Also, bit depth is not the same thing as the camera’s full usable dynamic range in practice. The min-to-max recordable range is one thing; how finely tones are distinguished within that range is another. Real limits also depend on black point, white point, noise, ADC behavior, and processing.

So: 8-bit linear data ≈ 8 EV is a reasonable simplification, but it does not mean all images are limited to 8 EV, and it does not describe typical raw files, which are usually 12- or 14-bit linear data.