The first phase of the AI boom has been easy to identify. The most visible winners have been the companies building the machinery: chips, cloud capacity, data centers, software platforms, foundation models, and the infrastructure required to make artificial intelligence work at scale. This is how technological revolutions usually begin. The first money is made selling the tools.

The more useful question is what happens next. The first winners are rarely the only winners, and often not the most important over the full life of a technological shift.

History gives us good reason to believe this. Electricity changed more than power generation. The internet changed more than early internet companies. Cloud computing allowed thousands of businesses to scale faster and operate more efficiently. Smartphones changed payments, maps, travel, photography, entertainment, commerce, and daily communication. Powerful technologies begin as products and eventually become infrastructure. The tool disappears into the work.

AI is likely to follow the same path. The next phase is likely to move beyond the companies selling AI and into the companies using AI to improve ordinary business performance: lower costs, faster service, fewer errors, smarter pricing, better forecasting, stronger customer retention, and higher output per employee. These are not abstract technology stories. These are business results.

That is why the next AI winners may not look like AI companies at all. They may be manufacturers reducing downtime, healthcare companies improving scheduling and diagnostics, banks and insurers improving underwriting and claims processing, logistics firms routing assets more precisely, retailers managing inventory more intelligently, or professional-service businesses producing more output per employee. AI’s next economic winners may be the companies that make the technology useful, not merely the companies that made the technology possible.

The investment challenge is that financial markets usually reward the obvious story first. By the time a theme is widely recognized and admired, meaningful future success may already be reflected in prices. Many early AI winners may remain exceptional businesses, but investors buy businesses at prices that already contain expectations. When expectations become enormous, even strong future results may not produce equally strong investment results from that starting point.

This is the practical meaning of “The Great Rotation.” It does not require a collapse in today’s most visible AI beneficiaries. It does not require abandoning technology. It does not require a tactical bet on one sector over another. It means the economic benefit of AI is likely to broaden. Investor attention and capital may shift from the companies that built the infrastructure to the companies using that infrastructure to improve their own economics. The rotation may be less about AI leaving the investment conversation and more about AI becoming embedded across the economy.

Why Diversification Comes First

This is why broad diversification—by country, company size, and fundamental style—continues to be of paramount importance to our long-term philosophy. Many investors think owning an index equates to owning a diversified and balanced set of businesses. It might, but it all depends on which index.

These days, many indices are skewed by size, so the largest companies can dominate the allocation. When a handful of companies become very large, a size-weighted index becomes increasingly dependent on those companies. The issue is not that investors own those companies. In many cases, they should. The issue is that a portfolio can gradually become more dependent on one prevailing story than the investor realizes.

If today’s largest companies continue to produce exceptional returns, that exposure can be beneficial. If future returns broaden into other parts of the economy, an overly narrow allocation may miss part of the next chapter.

Our answer is not to guess which current winner should be reduced or which future winner should be added. Our answer is to build a portfolio with enough breadth to participate in multiple possible versions of the future. That is the point of our investment approach. We have built the plan in a way that gives us the opportunity to participate without needing to guess. If today’s AI infrastructure leaders continue to dominate, diversified investors should participate. If the next wave of economic benefit comes from other sectors, countries, company sizes, or fundamental styles using AI well, diversified investors should participate there too. The plan is built around owning enough productive enterprise to benefit when innovation spreads.

When Conviction Becomes Concentration

Investors get into trouble when they confuse conviction with concentration. They see a powerful trend and conclude that more exposure to the most obvious beneficiaries must be the intelligent response. Sometimes that works spectacularly for a period of time. But the longer it works, the more dangerous the behavioral trap becomes. Success turns into confidence, then overconfidence, then dependence. Eventually, the portfolio is no longer participating in a trend. It is relying on that trend.

Rebalancing is one of the ways we prevent participation from becoming dependence. It is not a forecast around whether today’s winners continue to prevail or begin to revert to the average. It is a discipline that keeps no single winner, sector, style, country, or theme from becoming larger than its intended role. Left alone, strong performance can gradually convert a disciplined allocation into an accidental bet. Rebalancing restores the portfolio to its intended purpose before enthusiasm rewrites the risk profile.

For retired and retiring clients, reserves are part of the same discipline. Even if AI creates enormous long-term value, the path will not be smooth. There will be hype cycles, earnings disappointments, valuation resets, and periods when investors change their minds quickly. Long-term optimism requires short-term liquidity. The reserve side of the plan exists because progress arrives unevenly, and retirement income should never depend on financial markets being calm at the exact moment money is needed.

Own Progress Without Predicting Its Path

This is the actual planning lesson. AI may improve productivity, reshape labor markets, accelerate research, improve medicine, change client service, strengthen logistics, and alter the economics of businesses that do not currently appear to be technology leaders. We own productive enterprise broadly. We avoid overconcentration. We maintain reserves. We rebalance when success distorts the portfolio. We respect valuation. We build the plan so client outcomes are not dependent on predicting the next dominant company, sector, or theme.

We want to own progress without having to predict its exact path. If the next AI winners are the companies currently building the machinery, we expect to participate. If the next winners are companies that use the machinery better than others, we expect to participate there too.

A good investment plan should participate in growth, absorb uncertainty, and avoid the most expensive mistake investors make during every major innovation cycle: confusing the most exciting story with the whole story.