Westpac's AI halved the cost of moving an old data warehouse, the kind of job nobody brags about at a barbecue
By Brett Gavaghan, Founder of Adaptd ·
Westpac says AI tools cut its St.George data warehouse migration from a $70 million budget to about $35 million, and finished it more than two years early.
The short version
- Westpac says AI tools sped up code conversion, testing and documentation on its St.George data warehouse migration.
- The project was budgeted at $70 million, will cost about $35 million and finished more than two years early.
- Smaller businesses can borrow the idea by testing AI on one repetitive job with a result they can check.
Data warehouse migrations are the IT version of moving house: everyone agrees it has to happen, and nobody wants to pack the garage. Westpac has shown what happens when AI does some of the packing.
What did Westpac use AI for?
Westpac used AI tools to speed up code conversion, testing and documentation while moving its legacy St.George data warehouse to a cloud data platform. CEO Anthony Miller said this helped finish the migration more than two years early at half the cost, iTnews reported on 2 November 2026.
The migration was budgeted at $70 million and will now come in at around $35 million. Westpac expects about $15 million a year in direct benefits, and plans to switch off the old warehouse in FY27.
Why does a bank's IT project matter to a small business?
Westpac's migration matters because it shows where AI tends to pay off first: repetitive, technical work with a right answer that people can check. Converting code, testing it and writing documentation all fit that description.
Westpac also says the tools and methods can be reused on other migrations. Once a team works out how to do a dull job with AI, it gets to do it faster the next time too.
What should your business do about it?
Look for your own version of the data warehouse: a repetitive job with a result you can check, such as moving customer records into a new CRM, cleaning up a product list or writing down how your systems connect.
My take: you don't need a bank's budget to borrow the idea. You won't save $35 million, but a week spent copying records between systems is still a week. Run AI on a small sample first, check it against the original, then do the rest.
It won't make the evening news, but whoever usually does that job might buy you a coffee.
Sources
Wondering what this means for your business?
That's exactly what the discovery call is for.