The Reset: Committing to the Fall Line

I am writing this with slightly sore legs and a fresh perspective.

Many of you know I spend my winter weekends at Whistler, but you might not know that I have been a snowboarder since I was a teenager. I love the sport—truly. There is a specific, meditative joy in the feeling of "surfing" on fresh powder that remains one of my life’s great highlights.

However, I recently decided to take up skiing as well. I bought a pair of skis and took a lesson because I simply wanted to enjoy the mountain experience in the same way my wife and kids do. My kids are currently 8, 6, and 4, and my wife is still a novice skier, so it’s been about being on the same level and sharing that time together. I’m not giving up the board, but standing at the top of a run as a beginner again was profoundly humbling. I realized that while the mountain hadn’t changed, the tools had. The "equipment" that served me for decades was suddenly the wrong approach. I had to relearn how to move with a different center of gravity.

I share this because the financial markets are currently asking us to do the exact same thing. The "investment equipment" of the last decade—where software and digital processing were reliable staples—is the wrong gear for the terrain ahead. We are witnessing the transition of Artificial Intelligence from a "software trend" into a full-scale Industrial Revolution.

---

1. The "Zero Cost" Revolution: Why the Models are Changing

To help bring everyone up to speed, I wanted to take a moment to explain why this shift isn't just a "tech upgrade," but a fundamental change in the "physics" of how we invest. For years, we have maintained a diversified approach, but software-driven processing and intelligence services have always been a consistent part of our strategy. Today, AI is changing the rules of that game.

What is Agentic AI? (From "Chat" to "Do")

Most of us have seen AI that can write an email or summarize a meeting. We call that "Generative AI." But we are now entering the era of Agentic AI. Unlike traditional assistants, Agentic AI acts like a specialized worker that can independently plan, make decisions, and execute multi-step tasks to achieve a goal—like resolving a customer dispute or managing a supply chain—with minimal human supervision. To put it in perspective, McKinsey estimates the annual incremental value add of Agentic AI is upwards of $2.6 trillion, roughly equivalent to the GDP of Canada.

The "Zero Cost" Thesis: Intelligence as a Utility

The core idea driving our current strategy is that the marginal cost of intelligence is trending toward zero.

• The Shift: We are moving from "Software-as-a-Service" (tools humans use) to "Service-as-Software" (where the software is the worker).

• The Impact: In the past, scaling "intelligence"—whether coding, analysis, or processing—required human-led oversight. With AI agents, the cost to perform the "next" intelligent task drops to nearly the cost of the electricity and chips used to run it.

This creates a risk of counter-positioning and margin compression for legacy software and processing firms. When an AI agent can perform a task that software used to merely "facilitate," the value shifts toward the entity that owns the infrastructure. While many of these companies have been reliable performers for us, the pace of innovation creates a level of uncertainty that is now too high. As such, we have made a number of portfolio adjustments through the start of 2026 to reduce our exposure here to move into areas of greater certainty.

---

2. Beyond Tech: The Industrial Benefactors

It is a mistake to think this is only a "technology" story. This revolution is quickly moving beyond software, becoming a foundational layer for almost every major sector.

• Industrial & Manufacturing: We are seeing "Physical AI" transform factories into self-optimizing "Smart Factories". In logistics, autonomous routing is already driving significant cost reductions.

• Financial Services: Banks are using self-learning agents to cross-validate transactions in real-time, shifting from reactive damage control to proactive risk management.

• Consumer & Retail: AI agents now influence a growing percentage of purchasing decisions by redefining how consumers discover products and how retailers forecast demand.

The Breakthrough in Medicine

Perhaps the most profound impact is in healthcare. 2026 is the year AI has become essential in drug discovery. Identifying disease targets now relies on computational exploration, shortening feedback cycles from years to months. From precision oncology to genomics, pattern recognition is unlocking treatments that were previously hidden.

---

3. The Physical Infrastructure: Building the Global Nervous System

To understand the scale, we must look at the "factories" where this intelligence is manufactured: Data Centers.

• Growth: At the end of 2024, there were approximately 5,200 major data centers worldwide. By the end of 2025, that is expected to reach over 6,100.

• The 2030 Projection: Analysts anticipate that by 2030, global capacity will essentially double, adding nearly 100 Gigawatts (GW) of new supply—enough to power the entire United Kingdom and Germany combined.

The Copper Reality Check

A single one of these new "Hyperscale" campuses can require between 30,000 to 60,000 tonnes of copper. For context, a typical world-class mine produces roughly 150,000 to 450,000 tonnes per year. This means one large AI campus can consume up to one-third of the annual output of a major world mine. We are building these campuses at a rate that the physical world of mining—which takes 15 years to bring a new project online—will struggle to match.

---