Taking Gen AI through the inflection point of value creation, can this further disrupt market fundamentals?
Its common knowledge that disruption presents an opportunity to steal market share, particularly in what used to be the staid world of Australian and NZ banking. For a very long time the Big 4 Banks dominated market share in retail and small business, with fighting increasing amongst the global / international banks for large corporate accounts into Capital markets and M&A.
History of the market
In general, market share rarely changed. Most customers opened a bank account as a child and visited their local branch where the staff may know them and applied for a mortgage or business loan when necessary. Back in 2012, the retail mortgage market was >80% dominated by the Big 4 Banks, with CBA leading (27% mortgage market share), Westpac a close second (25%), NAB and ANZ both around 15-16%. Noting a large part of the differentiation was through price rises in Sydney relative to Melbourne. The remaining 20% was shared around regional Brands such as BOQ, internationals like HSBC, ING, new entrants like Macquarie and Non-Bank Lenders.
It was generally accepted that barriers to entry were strong and lasting, namely the depth of personal customer relationships to brands and local branch staff, plus the ability to run and maintain a bank cost effectively at scale.
Core to running costs is the cost of funding, which drives the Net Interest Margin (NIM) or the gap between average borrowing and lending costs. There are elements of credit rating (and implicit government guarantee “too big to fail” etc) but from a customer perspective this generally varies by customer type / competition dynamics for that customer. Retail deposits are stable and a relatively cheap form of funding when compared to that available in wholesale markets, so the bank with the most retail deposits can often have the best NIM, with front book lending in mortgages heavily contested.
Other running costs that are not scalable include meeting growing regulatory requirements and rapidly advancing technology, with some customers requiring new ways to interact with their bank, while others will want physical networks and cash / cheques.
These three factors all swung in the favour of scale, forming significant strategic “moats” around the Big 4 (including in NZ).
Shift 1 Occurred with improvements in technology – Sass (in particular banking as a service) and cloud services allowed new entrants to source out of the box core banking with overlays to manage clients and process lending. This allowed new entrants a choice of providers at different running costs and speed of establishment, plus over time, smaller banks to replatform and offer up to date aps and services, which are continuously updated via global providers.
Through this change we have seen Kiwi Bank in NZ grow at 2-3x system in retail deposits and lending, to reach a 5-6% market share behind the NZ subsidiaries of the Australian Big4. Kiwi bank was established in 2002 to establish a credible NZ owned rival with backing by the NZ government and superfund, operating through NZ post branches. They do operate at a lower target ROE of 8-10% (as do many mutuals/smaller banks), have reached significant and sustainable scale and have moved to grow in SME and Business.
We also saw Macquarie grow significantly above system to reach an approx. 6% share, with Bendigo and Adelaide finalising their core transformation and growing mortgage share to 3%. Front book retail mortgages often fall in terms of NIM to below the general average of retail divisions (it’s hard to compare apples to apples but most are around 1.7% currently).
At times only around 60% of new mortgages are written by the Big4, and I should also mention that the increase in customers use of mortgage brokers rather than walking into a bank branch influences market share (and profitability). By 2025 the Big 4 had shrunk to 74% of the market.
Price competition and margin compression in retail leads to more attention on higher margin areas
Institutional and large corporate lending usually runs low at around 1%, which can be lifted higher by additional banking to closer to 1.7-2% across the client, these relationships are competed by International Banks, with several increasing their Australian balance sheets over recent years (e.g. BOA and MUFG)
Small business lending is ripe ground for competition, with strategic moats similar to retail on local relationships, physical presence and higher historic NIMs (anywhere from 3-5%). Within the Big4, NAB is the historic leader at 22 to 28% share depending on size of business. Over the past few years Westpac, CBA, regional and new entrants are all increasing focus and potentially gaining share.
Spotlighting Judo, who entered the market in 2018/19 and have grown rapidly. Focusing on providing a personal banking service to SME customers who would generally be too costly to serve with an individual advisor. Technically, Judo started out with a selection of banking service vendors, that they are adjusting as they grow.
In general smaller competitors are also likely to have implemented relatively uncustomised versions of global providers packages, so could more rapidly benefit from rollouts of AI developments such as SalesForce’s rapid development of SalesAgent for Relationship Manager efficiency.
Potential further shifts in the market
Smaller competitors often struggle with both the cost and consistent availability of funding, which impacts both their NIM and the security of growth plans. Reliance on expensive term deposits and availability of securitisation has been a problem but as we see private capital firms such as Apollo (Atlas) and others bringing their continual appetite for long credit driven by the acquisition of Insurance balance sheets this could well change.
If we believe the US firms are here to stay, and will have a continual appetite for high quality long lending, which they can provide at a cost of capital almost the same as the Big4 – does this allow for sustained capital provision for smaller players with disadvantaged deposit bases?
Also addressing the regulatory overlay, on 7th August APRA announced the intention to introduce a lighter touch framework for very small banks and appears to be considering the balance between safety and competition / innovation, noting this is a great intention but execution could be lengthy. Also, the ongoing Productivity Commission appears to be supportive of the general idea that unconstrained regulation has dampened productivity, but it may be wishful thinking that much will change…
Coming on to AI – it seems after a very long time of discussing this, models may or may not reach a level of sentience, and most people conflate automation, robotics, machine learning and generative AI, however there is no doubt that whatever you call it, computers can now work very fast and replace at least some of the things a knowledge worker can do. Usually this will be alongside a human being (human in the loop or HIIL) with programmers enhanced not replaced, and lots of care needed in a heavily regulated banking environment, but either way, change is coming…
Does AI achieve lasting change? Last weeks Alphabet results would imply yes, investment and increasing sales in cloud infrastructure is interesting but significantly, Google has started to disrupt itself and the sacred cow of search with AI overviews. Google needed to retain search users and offer them a more up to date AI type experience with AI overviews, while there is much contention over the changing economics of consumer behaviour around number of links followed on seeing an AI overview instead of a straight search and the follow on impact of advertising revenues, at the moment Google has managed to retain and grow revenue 10% on search in the latest results. Significantly, consumers are liking the AI overviews and are changing their behaviour immediately.
Australia is certainly behind in technology, but our population is using and experiencing AI. All banks I have spoken to are running large and small pilots, through to scale implementation, from areas as simple as rolling out Co-Pilot, delivering broad market scans for M&A document preparation, using AI to produce market results videos, identifying potential payments fraud, automating the collection of documents for internal audit, through to improving complicated corporate client onboarding and lending approvals (where multiple steps and handoff’s in the process can be tracked and simplified).
In this case, the limitation factors are availability and location of data, complexity of existing core banking and CRM systems etc, plus difficulty to organise and scale from pilots. Noting that to achieve real results workflows and job designs will need to be adapted. My working hypothesis, still to be validated, is that this presents a huge opportunity to well-funded competitors, who also have the most available large data sets required to train agents. At the same time, a large risk exists that smaller and more nimble organisations may have a faster go to market to gain competitive advantage, either in service uplift or workflow design / cost of service, particularly if they can utilise solutions provided by global software providers.
Nearly everyone I have spoken to has highlighted the order of implementation. Encouraging staff to embrace AI, with its ability to free up their time, empower them to work smarter and improve customer experience is the right place to start. To deliver the Human In The Loop necessary to train agents and redesign workflows, you need your best people onboard and unafraid, however there is an undercurrent of future cost reduction that is inevitable.
Not all staff will be able to retrain; not all areas will need as many people once the organisation is fully transformed. Australia and NZ are limited markets, how many banking services can be sold into our economies? Added value and increased productivity will have to have an upside limit, unless we again have ambitions to expand into adjacent products such as wealth or geographic markets?
It remains unpredictable who will win this race. Will the large banks, particularly CBA who seems to have an established management and faster to market with technology, again be the market leader (reference their new cooperation with OpenAI)? Will smaller banks gain an advantage? Will new entrants such as MYOB or Xero achieve banking disruption with embedded payments? Maybe, but it costs a lot to train models and many of the new entrants should already have frictionless processes so little additional cost out to gain?
Also, what are the ongoing economics of AI? There seem to be no scale economies once the models are trained and electricity costs are only going up? But maybe chip improvements will address this issue and scale economies will emerge? Have the big US companies overinvested and will sky high share market valuations go through a crash as the US economy swallows tariffs, slower growth and potential stagflation?
For me there are too many unknowns, hence this is certainly a part 1 note – with a watch this space on the complexity of consumer behaviour, trade economics, share prices etc These are certainly interesting times… If you would like to take this conversation further and explore how your organisation can find the right people to be part of the change, I would welcome the opportunity to connect in person. At Blenheim Partners, our dedicated search work is focused on helping clients build leadership teams and boards capable of navigating precisely these challenges. Let’s arrange a time to meet and discuss how we can support you in shaping the future of your business.
Julia Patterson
Partner, Financial Services