Image source: Getty Images
Is this the end of moats?
Okay, that's a deliberately provocative question… but perhaps not as provocative as it first seems.
But first, let's define our terms.
Warren Buffett made the idea of an economic 'moat' famous: a sustainable competitive advantage that protects a business from competitors and allows it to earn attractive returns over a long period of time.
It might be a strong brand. Scale. High switching costs. Network effects. Intellectual property. Regulation. Or simply being able to do something more cheaply or effectively than everyone else.
Find a company with a wide moat, buy its shares at a reasonable price and – assuming the moat remains intact – time can do much of the hard work for you.
It's a simple idea. And a very powerful one.
But AI is forcing us to reconsider which moats will continue to be sustainable now that we're a very different technical world.
Is AI going to destroy them all?
No.
Will it leave them all untouched?
Also no.
—
YouTube LIVE: Tomorrow at 12pm AEST
Well, earnings season has just finished, the economy is growing slowly (but also too quickly!), and rate rises are on the horizon.
There is a lot going on in the world at the moment. And it's affecting our economy and investments.
I'll be hosting a LIVE one-hour market update and Q&A on Friday, September 4, 2026 at 12pm AEST to update viewers with my thoughts on all of that and more.
Plus, taking your questions, LIVE, on YouTube in the process.
And… it's free! Join me, using this link or by clicking on the image below (and if you do it now, you can set a reminder).
See you there!
In reality? Some moats will probably remain largely unaffected. Some will narrow. And some might disappear altogether.
Which ones? And when?
I have no idea. Not with any certainty, anyway.
And nor does anyone else, despite the very confident predictions currently being made about what AI will and won't do.
But we can (and should!) think in probabilities. We can consider where the risks are highest and ask whether the assumptions we've made about individual companies still hold.
I'd start with businesses whose advantage is mostly based on being able to do something others can't.
Writing software. Producing advertising. Analysing documents. Creating images. Answering customer questions. Turning large amounts of information into something useful.
Until recently, those things required scarce skills, large teams, huge scale, or years of accumulated expertise.
Now? AI is making many of them cheaper, faster and more widely available.
That doesn't mean software companies, creative businesses or consulting firms suddenly become irrelevant..
But it does mean that some of the capabilities that helped distinguish them may become easier for competitors – and customers – to replicate.
If your moat is essentially "we know how to do a thing", what happens when the thing becomes much less difficult?
Then there are switching costs.
Some companies retain customers because leaving is genuinely difficult. Data has to be moved. Systems need to be rebuilt. Staff must be retrained. New software needs to be connected to everything else.
It's expensive. It's disruptive. And it can go wrong.
So customers stay put, even if they're not particularly happy.
Those switching costs won't disappear overnight. But AI can already help write code, translate data, build integrations and teach people how to use unfamiliar systems.
The moat may remain. It just might not be as wide as it used to be.
Brands could also come under pressure.
A trusted brand helps us decide what to buy. We recognise the name, know roughly what it stands for and feel reasonably confident we'll get what we expect.
But what happens when an AI assistant makes the decision for us?
If I ask an AI agent to compare every insurance policy, mobile phone plan or retailer and choose the one that best meets my needs, familiarity might count for less.
Worse for the company, the primary customer relationship might belong to the AI platform that makes the recommendation, rather than the business that provides the product.
And I'd be wary of cost advantages that come largely from processing routine work more efficiently than competitors. If similar AI tools are available to everyone, today's low-cost operator might find its rivals catching up.
Which brings me to something I've said before: simply using AI probably won't be a competitive advantage.
It'll be the ticket to the dance.
Oh sure, early adopters might enjoy a temporary boost to productivity and profit. But if competitors have access to much the same technology, those benefits will probably be competed away through lower prices, better products or both.
Good for customers. Good for society.
But not necessarily a wider moat.
Still, some competitive advantages look much less exposed than others.
AI can create a property website. It can't recreate REA Group Ltd (ASX: REA)'s listings and audience.
It can assist medical research. It can't quickly replicate CSL Ltd (ASX: CSL)'s plasma collection network, manufacturing capability, regulatory approvals and accumulated know-how.
And while AI can improve banking technology, it can't simply hand a new entrant Commonwealth Bank of Australia (ASX: CBA)'s licence, deposit base, customer relationships and public trust.
Nor can it manufacture scarce mineral deposits, prime locations, physical distribution networks or genuine economies of scale.
That doesn't make those moats invulnerable, by the way.
A company might retain its network but lose control of the customer interface. A trusted incumbent might keep its customers while finding its products easier to compare and its margins harder to defend.
So, what should investors do?
Don't predict. Prepare.
Ask what the moat is actually made of.
Does the company own something genuinely scarce, or does it merely possess a capability that AI could commoditise?
Are its switching costs structural, or is changing providers just difficult and annoying?
Does it own the customer relationship, or could an AI assistant insert itself between the business and its customers?
And if AI makes the whole industry more productive, who keeps the benefit?
The company?
Maybe.
But it could just as easily be its customers, suppliers or competitors.
And, as always, valuation matters. A wonderful company can be a lousy investment if its share price assumes the moat will last forever.
If the future of that moat has become less certain, investors should demand a larger margin of safety.
The other thing? You don't have to become a futurist, today. You don't have to know, with any certainty, what things will look like in five or ten years. But the preparation I talked about earlier means understanding potential risks and weaknesses, so you'll be more likely to notice if and when they start to impact a company's profits or prospects.
No, AI probably isn't the end of moats.
But it might be the end of taking their permanence for granted.
Fool on!
