There's also serious coordination problems internally. How easy is it to get access? Did you make buddies with the little emperors in charge of the historical archives? Is the archive in your company large enough? Are the chances of getting access to another competing company's dataset high enough to not require multiple decimal digits to express?

(This I think is part of why Chinese human genomics has been such a disappointment: yes, there's collectively a lot of data, but there ain't nothing like UKBB pulling it all together. Just thousands of fiefdoms.)