In This Issue

• Why AI is the first real stress test for economic moats in a generation
• Which moats just got downgraded by Morningstar in 2026 and why
• The one company with a literal monopoly no competitor has cracked
• Why Visa and Mastercard remain the cleanest network-effect story on earth
• What ROIC is telling you before the income statement does
• How to know if you own a real moat or just a story about one

Something interesting happened in early 2026 that most people filed under “AI selloff” and moved on. What actually happened was more significant. Morningstar quietly conducted one of the most sweeping reviews of economic moat ratings in its history, covering 132 companies globally. The result: more moat downgrades than any prior review they have ever done, including the 2008 financial crisis and the COVID crash.

Adobe. Salesforce. Workday. ServiceNow. Wide moats, downgraded to narrow. The reasoning was blunt: reduced confidence that these companies can sustain excess returns over the next decade as AI reshapes the software layer beneath them.

Here is where I think this gets interesting. The Morningstar team was careful to say AI is not a universal disruptor. It is a sorting mechanism. Some moats are being deepened by it. Others are cracking. The job for investors right now is figuring out which is which before the stock price tells you.

The Moats That Are Getting Stronger

Start with the clearest case in the market. ASML is the only company on earth that makes extreme ultraviolet lithography machines, the equipment required to manufacture chips at 7 nanometers and below. Every Nvidia GPU, every AMD AI chip, every Apple silicon design runs through ASML’s tools. Nikon and Canon walked away from EUV development years ago. There is no backup supplier. Each Low-NA system sells for roughly 180 million euros. The new High-NA generation clears 380 million each. Customers pay because there is no alternative below the node threshold.

That is not competitive advantage. That is a structural monopoly. And AI is making it stronger, not weaker. As chip designs grow more complex and demand scales from millions to tens of millions of units, ASML’s order book compounds with it. The company raised its full-year 2026 revenue guidance to a range of 36 to 40 billion euros. Q1 2026 posted 10.34 billion in revenue with 53% gross margins. The stock is up roughly 77% year-to-date as of mid-2026.

Slight tangent, but worth sitting with: this is what a true moat looks like when you strip away the label and look at the structure. One vendor. No substitute. Pricing power that compounds with every AI dollar spent. The moat is not wide. It is vertical.

Network Effects vs. Everything Else

The other category holding up well is network effects. And the cleanest example, still, is Visa and Mastercard. Together they process roughly 85% of global non-Chinese card payment volume. Visa’s quarterly operating margin has averaged 67% over the past five years. Mastercard’s has averaged 57%. These are not good margins. They are extraordinary margins for businesses of this scale.

The reason the moat is so hard to attack is structural. Any new payments system trying to displace them needs to simultaneously sign up merchants and consumers. Without merchants, consumers have no reason to switch. Without consumers, merchants have no reason to sign up. That chicken-and-egg dynamic took Visa 60 years to resolve. Unless a new system is genuinely ten times better across every dimension, the threat is theoretical, not practical.

Real-time payment rails like UPI in India and Pix in Brazil are worth watching. UPI already processes more monthly transactions than Visa and Mastercard combined globally. These are government-backed systems that bypass interchange entirely. Over a 5 to 10 year horizon, this is a real variable. For now, the networks look to be adapting rather than being displaced. But this is the data point I would watch more closely than any other.

Mastercard’s Q3 2025 revenue grew 17% year over year. Mastercard’s most recent quarterly results showed net revenue of $8.4 billion, up nearly 16%, with adjusted EPS of $4.60 beating estimates for the fourth consecutive quarter. The numbers are clean. The moat, for now, is intact.

Where Moats Are Cracking

Back to the Morningstar downgrades. The logic is worth understanding because it tells you something broader about where moats are fragile right now. The companies that got downgraded share a common characteristic: they monetize seat-based software licenses and human workflow automation. Adobe sells creative seats. Salesforce sells CRM seats. Workday sells HR and finance seats.

AI compresses the number of seats needed to do the same work. It also creates a genuine opening for new entrants to build competing products on top of large language models without needing decades of development time. That is the threat. Not that these companies disappear, but that their long-term excess returns become harder to sustain with certainty over a 10 to 20 year window.

The part that does not get said enough: Morningstar also upgraded several companies in the same review. Cloudflare was upgraded to wide moat. The reasoning was that AI actually increases demand for its services. Cybersecurity and financial infrastructure broadly came through the review in better shape than enterprise software did.

Microsoft kept its wide moat rating. The reasoning is useful: Microsoft’s moat is built across four of the five primary sources simultaneously. Switching costs, network effects, cost advantages, and intangible assets. No single AI disruption vector takes all four out at once. That multi-source structure is what separates the most durable moats from single-source advantages that look strong until one variable changes.

What You Are Actually Looking For

Here is the thing about moat analysis that most people get backwards. A moat is not a story you tell about a company. It is a financial measurement with a leading indicator. The leading indicator is ROIC.

When a moat starts eroding, ROIC compresses before revenue does. Margins narrow first. Growth decelerates second. Then the multiple contracts. By the time the income statement shows the damage, the stock has usually already repriced. Watching ROIC trends across five to ten year periods, comparing them against the company’s cost of capital, and tracking when that spread starts narrowing is the earliest signal available to a fundamental investor.

Wide moat companies almost always trade at premium valuations. That is part of the deal. The risk is not that they turn out to be bad businesses. The risk is that you pay so much for the moat that even perfect execution produces mediocre returns. Price matters alongside quality.

What 2026 has clarified is that the moat framework is not broken. It is being pressure-tested. Companies with advantages rooted in physical infrastructure, network scale, or technological monopoly are coming through that test well. Companies whose advantages lived primarily in software complexity are being asked harder questions. Some will answer them. Some will not.

The question worth sitting with is not whether moat investing works. It does. The question is whether the specific moats in your portfolio are the kind that get stronger under pressure, or the kind that looked strong until the pressure arrived.