The castle walls are melting. Warren Buffett’s favorite investment metaphor is failing in real time. For decades, the concept of an economic moat protected the portfolios of the cautious. It provided a buffer against competition and a guarantee of long term cash flows. That era ended this week. A new study from Westwood investment firm, led by Adrian Helfert, confirms the worst fears of the value investing community. Four out of five classic moat pillars now possess almost no predictive power for stock performance in an environment dominated by generative artificial intelligence.
The Collapse of Traditional Barriers
Capital is no longer a gatekeeper. Intelligence is becoming a commodity. The premium previously paid for ‘quality’ companies is evaporating as AI agents bypass the friction that once defined market dominance. According to the latest Morningstar analysis, the structural advantages that allowed firms to charge high prices for mediocre services are being liquidated by automated efficiency. The data suggests that the speed of disruption has outpaced the ability of traditional moats to provide protection.
The five pillars of a moat are intangible assets, switching costs, network effects, cost advantages, and efficient scale. Helfert’s research indicates that only one of these remains a reliable predictor of alpha. The others have been neutralized. When an AI can replicate a proprietary software feature in an afternoon, the ‘intangible asset’ of a patent or a brand loses its teeth. When an AI agent can migrate an entire enterprise database to a competitor with a single prompt, ‘switching costs’ become a historical curiosity.
Quantifying the Predictive Decay
The following table illustrates the shift in how these pillars influence market valuation as of February 2026. The decay is most pronounced in sectors previously thought to be ‘un-disruptable’ like legacy banking and enterprise resource planning.
| Moat Pillar | 2021 Predictive Weight | 2026 Predictive Weight | Primary Disruptor |
|---|---|---|---|
| Intangible Assets | 28% | 6% | Open Source LLMs |
| Switching Costs | 22% | 4% | Agentic Data Migration |
| Network Effect | 30% | 12% | Synthetic User Bases |
| Cost Advantage | 15% | 65% | Compute Efficiency |
| Efficient Scale | 5% | 13% | Physical Infrastructure |
Cost advantage is the last man standing. In a world where software is free, the only thing that matters is who can run the hardware the cheapest. This explains the relentless bid for energy-efficient data centers and proprietary silicon. The market is no longer pricing ‘brand loyalty.’ It is pricing ‘joules per token.’
Visualizing the Moat Efficacy Index
The chart below tracks the collective predictive power of the five moat pillars over the last five years. Note the sharp divergence starting in late 2023, accelerating through the 2025 fiscal year.
Moat Efficacy Index: 2021 – 2026
The Death of the Brand
Brands used to be shortcuts for trust. Now, they are liabilities. Consumers and procurement officers alike are using AI comparison engines that strip away marketing gloss to reveal raw utility and price. Per recent Reuters reporting on retail trends, private label goods powered by AI supply chains are gaining market share at the fastest pace in thirty years. The ‘Intangible Asset’ pillar is failing because the consumer’s decision-making process has been outsourced to a machine that does not feel emotion or brand affinity.
This is a technical crisis for the S&P 500. A significant portion of the index’s ‘book value’ is comprised of goodwill and trademarks. If these assets no longer predict future earnings, we are looking at a massive, silent write-down of the American corporate balance sheet. The Westwood study suggests that the market has not yet fully corrected for this reality. We are seeing a ‘zombie moat’ phenomenon where companies trade on past reputation while their actual competitive defense has already been breached.
The Network Effect Fallacy
Network effects were supposed to be the ultimate winner-take-all mechanism. But AI has introduced the concept of ‘synthetic density.’ A social network or a marketplace no longer needs millions of humans to be useful. Synthetic agents can provide liquidity, content, and interaction, lowering the barrier for new entrants to challenge incumbents like Meta or Amazon. The ‘moat’ provided by a large user base is being bypassed by platforms that can simulate that density from day one.
Investors are pivoting. The focus has shifted from ‘who has the users’ to ‘who has the compute.’ This is reflected in the S&P 500’s recent volatility as capital rotates out of legacy software and into infrastructure. The winners of 2026 are not those with the best logos, but those with the lowest marginal cost of intelligence. The Westwood data is a warning. If your investment thesis relies on a company’s ‘strong brand’ or ‘high switching costs,’ you are likely holding a relic of a pre-agentic world.
The next data point to watch is the Q1 2026 earnings cycle for legacy enterprise SaaS providers. If churn rates continue to decouple from historical norms despite high ‘satisfaction’ scores, it will confirm that the AI-driven migration has reached the core of the economy. Watch the 10-K filings for mentions of ‘agentic interoperability’ as a risk factor. That is where the next leg of the moat erosion will be visible.
