The AI Infrastructure Pivot: Why Your Business Strategy is Already Obsolete

The core idea here is that AI has moved beyond a software trend and into a fundamental challenge regarding infrastructure and power delivery. The implication is that the SaaS apocalypse is not a failure of software, but a shift toward a world where data moats are being bypassed by agentic models. For leaders, the competitive advantage no longer lies in software features, but in owning the underlying compute, power, and data layers. Those who treat AI as a search engine will be disrupted; those who treat it as an agentic infrastructure play will capture the next decade of value.

The Infrastructure Paradox: Why Software is Now a Commodity

While the market focuses on the SaaS apocalypse--the fear that AI agents will make traditional software companies obsolete--the real story is happening deeper in the stack. Aaron Kennon notes that the current disruption is maturing into a story about power, infrastructure, and productivity. The immediate benefit of AI agents is consumer convenience, but the hidden cost is the potential erosion of traditional software moats.

If an AI agent can scour the market to find the best price or manage complex workflows, the software that previously held the customer relationship becomes a commodity. We are seeing a divergence: companies like Shopify are integrating to capture the tailwind of AI traffic, while others, like Amazon, are erecting walls to prevent models from extracting consumer insights.

> "I think Amazon is very concerned that perhaps Meta is extracting insights into consumer activity in creating their own models that could disintermediate some important sort of moats that have existed at Amazon over the past several years."
>
> -- Aaron Kennon

The Geopolitical Feedback Loop in Energy

Systems thinking reveals that energy policy is rarely just about energy; it is a critical lever for the broader economy. Amos Hochstein highlights the danger of viewing energy exports through a binary ban or no-ban lens. A ban, while politically attractive in the short term, creates a negative signal that forces production decreases, eventually compounding into higher prices and economic instability.

The system responds to these interventions in ways that policymakers often underestimate. By restricting exports, you do not just lower domestic prices; you break trust in the U.S. economy for global consumers, who will then build systems that rely less on American infrastructure. The downstream effect is a loss of strategic leverage that lasts far longer than the temporary relief of a price drop.

> "The bad signal on an export ban is that it goes very quickly to production decreases, right? Not just at refineries but at the oil side. If I think I have nothing to do with this thing and the price is going to crash there. And then you have this cycle... of prices go down and eventually they will go off."
>
> -- Amos Hochstein

Regulatory Clarity as a Competitive Moat

The debate over AI regulation often assumes that rules stifle innovation. However, Juan Londoño suggests the opposite: pro-market regulation provides the clarity necessary for adoption. When consumers fear that an AI agent might execute a trade without approval or leak sensitive financial data, they stop using the technology.

The system responds to uncertainty with withdrawal. By defining the boundaries of liability--what is the responsibility of the model developer versus the user--regulators can accelerate innovation. In this view, pro-innovation regulation is not about picking winners, but about lowering the barrier to trust. Without this, the technology remains a toy for the few, rather than an engine for the many.

Key Action Items

• Audit your Data Moat (Immediate): Assess whether your current software platform is a primary source of customer insight or if it is becoming a commodity that AI agents can easily bypass.
• Evaluate Infrastructure Exposure (Next Quarter): Shift focus from software feature sets to the underlying compute and power requirements of your business. If your strategy relies on cheap, abundant compute, stress-test your supply chain against current infrastructure bottlenecks.
• Adopt Agent-Ready Governance (Next 6-12 Months): Instead of waiting for federal mandates, implement internal clarity on AI liability. Define exactly what your AI agents are authorized to do and where human oversight is mandatory to build the trust necessary for long-term adoption.
• Shift from Binary Policy to Inventory-Linked Planning (12-18 Months): For leaders in capital-intensive sectors, move away from reactive, binary policy responses. Develop frameworks that tie output or export levels to inventory thresholds, creating a predictable, durable system that avoids the boom-bust cycles of reactive governance.
• Prioritize Trust as a Product Feature (Ongoing): Treat regulatory and safety concerns not as a hurdle to be cleared, but as a competitive advantage. In a market flooded with agents, the platform that users trust with their financial life will win the loyalty of the next cohort of users.