I began making investments in junior mining companies in 2008 and 2009, while markets were still recovering from the financial crisis.
I was in my twenties, living in Vancouver and learning from geologists, financiers and company builders. Some early investments worked exceptionally well. Good timing played a role, as did the powerful recovery in copper, gold and silver.
That period taught me to separate a strong company from a strong capital cycle.
When money pours into one area of the market, weaker businesses gain access to capital they would never attract under normal conditions. Management teams begin to confuse fundraising with progress. Investors spend less time asking what the money will accomplish because they are afraid of missing the next deal.
I see that pattern forming around artificial intelligence.
U.S. startups raised more than $400 billion in the first half of 2026, according to the PitchBook-NVCA Venture Monitor. The figure surpassed every previous full-year total, but the market was unusually concentrated. Most of the capital went to AI companies and financings of at least $100 million.
AI deserves substantial investment. But that doesn’t mean every company positioned near it should be funded.
AI Is Becoming a Physical Economy Story
The market still speaks about AI as though it were primarily a software story. That description leaves out the source of its most important constraints.
More capable models require more computing power. Computing power requires electricity, semiconductors, cooling, transmission capacity, data centres and physical infrastructure.
The International Energy Agency expects global data-centre electricity consumption to more than double by 2030, reaching roughly 945 terawatt-hours. In the United States, data centres are projected to account for nearly half of electricity-demand growth through the end of the decade.
AI is 'digital'. But its expansion is governed by the physical world. This should change how you think about investing in this new technological miracle, because it doesn’t scale like ones in the past.
The crowded part of the market is the visible AI product. I am increasingly interested in the companies one or two layers beneath it, including grid technology, power management, cooling, cybersecurity, data protection, industrial automation and advanced materials.
Application companies still matter. The better question for investors is what every successful AI platform will eventually need, regardless of which model wins.
That question usually leads away from the fashionable part of the market and toward businesses whose products become more necessary as AI adoption grows.
A Financing Must Remove a Risk
I’ve watched management teams raise tens of millions of dollars without being able to explain which risk the financing was supposed to remove.
Capital should complete a technical milestone, prove customer demand, secure an asset, strengthen a critical part of the team or move the business toward commercial scale. The next investor should be able to see that something important became more certain because the previous round was spent well.
Before investing in a funding round, the question I ask is straightforward: what will this company know, own or prove after spending the money that it does not know, own or prove today?
A vague answer usually means the financing is filling a hole. A precise answer shows that management understands how capital can create value rather than merely extend the runway.
That distinction matters in any sector. It matters even more when so much money is chasing the same idea.
Remember, capital is time purchased from investors. The company must use that time to become more valuable, less risky or both.
The Labour Constraint
The physical limits beneath AI are only part of the investment case.
Fertility is below the replacement rate in nearly every OECD country, and the working-age population is expected to decline across much of the developed world. Businesses will face increasing pressure to operate with fewer available workers.
This is one reason I take industrial automation seriously. The strongest automation businesses will do more than lower costs. They will help customers continue operating when qualified workers are unavailable at a price the business can afford.
The investment case for automation is supported by two forces at once. The technology is improving, while the demographic need for it is becoming harder to ignore. For investors, necessity is usually a stronger foundation than excitement.
What I Am Looking For
I’m most interested in companies positioned where technological ambition runs into physical constraint. They may help generate, transmit or conserve electricity. They may reduce the labour required to operate a mine, factory, warehouse or data centre. They may protect proprietary data or solve a materials problem that software can't touch.
The AI boom is real, but its next stage will require an enormous buildout of power, infrastructure, automation and physical capacity.
I want to invest in the companies that make that buildout possible.
