Qlik has a very specific data-structure so that your IT system would need to be capable to address it in the same way like the Qlik engine does to access the in-memory data. Therefore I doubt that there is a real way especially not only by considering the technically aspects else also the legal ones.
Therefore I think you will need at least one layer between Qlik and your IT system. For this you could use any file-exchange or any write-back approach to a database or maybe a connector like this one: tiq-solutions.de - tiq-java-service-connector. You could contact the head - Ralf Becher - behind the connector to find out if it might be possible / sensible in your usecase.
I think there are perhaps some products for getting data out of QV in-memory models but I have my doubts that works well.
What I would do instead is to run a Qlik Sense server beside where you binary load your QVW, or having an app with same load script.
Then you can use the engine API to access the in-memory model by defining a HyperCube. There is also an open source project q-risotto I started which provides REST API wrapper for this to make it easy:
I could imagine that you run into various problems with such a big table with or without NPrinting in regard to limitations/bugs within the object, the RAM consumption and the calculation/transfer-times. I wouldn't do a try in this direction else I would create this table within the script and store it as csv.
