One of the complaints about some earlier digital cameras with high megapixel counts was that they didn't produce better photos than those of ostensibly lower specification. The issues include increased 'noise' resulting from cramming more imaging pixels into the same amount of space without otherwise improving the technology used.
Benefits of larger sensors include the collection of more light, reduced noise, and more accurate internal focussing.
And so Canon has developed what it claims to be the world's largest CMOS image sensor. With dimensions of 202 by 205mm, this isn't a component that's going to turn up in a consumer or even prosumer camera any time soon.
Where it is likely to turn up is in video cameras for recording the night sky or nocturnal animal behaviour. This is because it can capture an image in one-hundredth of the amount of light required by a 35mm full-frame CMOS sensor, which means it could be used to capture 60fps video with just 0.3 lux of illumination (approximately half the brightness of a moonlit night)
The size of the sensor presented various technical problems - see page 2.
Canon used the "cleanest of cleanroom environments" for the production process to minimise image imperfections and dust, company officials claimed. This would presumably be particularly important in order to achieve an acceptable yield as only one sensor is produced per wafer. Multiple copies of more conventionally sized chips are produced from each wafer, so even if one or two are rejected at the testing stage the entire wafer hasn't been wasted.
Another technical obstacle that had to be overcome was the design of a circuit that would allow sufficiently fast readout. Normally, enlarging a sensor leads to slower I/O performance.