Sector Rotation: Are We an AI Portfolio?

Are We an AI Portfolio?

It is a fair question, and we will give a straight answer. Today, 12 of our 15 equal-weighted holdings are technology companies tied to the AI build-out. By that simple count, the portfolio is roughly 80% technology. So yes, right now we look a lot like an AI portfolio.

But that picture is only a few months old. In January and February, gold and diversified miners made up about 35% of the book. That was an equal share with technology, not a small position. The mining sleeve included NEM, RIO, AU, CDE, KGC, and IAG. The technology sleeve was smaller then, but it was not a different bet. Several of today's core holdings were already in the book. MU, WDC, and LITE have been held every month this year, and STX since February. What changed was the screen, not a decision in a meeting. Month by month, the gold names stopped clearing our quality and momentum filters, and more technology names started passing them. We did not choose to lean into the AI build-out. The data kept selecting it, name by name. The heavy technology concentration you see today is the output of that process, not our personal opinion about where the market is heading.

Where the Early-Year Alpha Came From

Through the end of February, the strategy was up roughly 21%. The S&P 500 over the same stretch was close to flat, near 1%. That early alpha was not a one-sector story. Gold and silver miners drove a large part of it. A cluster of memory and optical technology names drove much of the rest. The real difference from today is concentration. Back then technology was about a third of the book, sitting next to an equal-sized gold sleeve.

The pattern held when markets turned. In March the S&P 500 fell 5.2%, and we matched it at 5.2%. The main reason we held the line was a sharp cut in equity exposure, which moved a large part of the book into cash. Among the stocks we did hold, our hard-drive and memory names held up best, while the gold sleeve gave back some of its winter gains. Our worst drop since inception stayed at 8.8%, slightly shallower than the S&P 500 at 9.1%.

Then the data shifted. From April onward, the same process that held both miners and technology in winter began surfacing more technology names instead. The chart below shows the changeover.

The Rotation, Month by Month

Here is the sector mix at each monthly rebalance, measured as the share of holdings in each sector.

The sector percentages only tell half the story. The names behind them rotated too. Here is who made up the two largest sectors at each monthly rebalance. The remaining quarter to third of the book sat in financials, industrials, energy, and consumer names that also turned over month to month.

January 2026

Technology (35%): BE, MU, WDC, CIEN, LITE, CLS, COHR. Gold and miners (35%): AU, CDE, KGC, PAAS, EQX.

February 2026

Technology (35%): MU, WDC, STX, CIEN, LITE, GLW, BE. Gold and miners (35%): NEM, RIO, AU, CDE, KGC, IAG.

March 2026

Technology (32%): MU, WDC, STX, SNDK, LITE, TTMI. Gold and miners (26%): NEM, RIO, AU, CDE, IAG.

April 2026

Technology (40%): MU, WDC, STX, SNDK, LITE, AMAT, KEYS, NXT. Miners (10%): RIO, IAG.

May 2026

Technology (60%): MU, WDC, STX, SNDK, LITE, DOCN, TTMI, TSM, CLS. Miners (13%): CDE, RIO.

June 2026

Technology (80%): MU, WDC, STX, SNDK, LITE, DOCN, TTMI, CIEN, VIAV, FLEX, SANM, AMD. Balancers: PWR and STRL in infrastructure, BWA in auto parts.

No one made a top-down call to sell gold and buy chips. Each month, every name had to re-qualify on the same factors. We weigh profitability and growth, the quality of those earnings, the direction of earnings revisions, and price strength, which reflects real investor demand. Miners cleared the screen in winter. By spring, semiconductor, memory, and networking names were posting the strongest numbers. So the portfolio followed the data.

By June, the qualifiers were almost all tied to the AI build-out, spread across the technology stack:

• Memory and storage: MU, STX, SNDK, WDC
• Compute: AMD
• Optical and networking: CIEN, LITE, VIAV
• Electronics manufacturing: FLEX, SANM, TTMI
• Cloud infrastructure: DOCN

Which Holdings Drove Each Month

Concentration is one question. Contribution is another. The holdings that actually drove the gains line up with the rotation. In winter the leaders were a mix of gold miners and technology. By spring, technology names were doing almost all of the work.

These figures are the price moves of each holding over the month it was held. They are not the strategy's net return. Because the risk overlay often keeps part of the book in cash, the portfolio captures a share of these moves, not the full amount. The pattern is the point. Early on, gold miners and a memory and optical technology sleeve shared the work. By spring, technology names led on their own.

So, Are We an AI Portfolio Today?

Honestly, to a large degree, yes. Twelve of fifteen names sit somewhere on the AI build-out. We will not pretend the book is evenly spread across the economy right now. It is concentrated in one theme, because that is where the data is strongest today.

Two things keep that concentration in check. First, three holdings sit outside the chip-demand cycle. PWR and STRL build the electrical and physical infrastructure that data centers rely on, working off multi-year order books rather than chip orders. BWA earns its revenue from the auto-parts cycle. Their beta drivers differ from a pure semiconductor name.

Second, and more important, our risk overlay currently holds equity exposure near 49%. That means roughly half the portfolio is in cash. In our F1 terms, we are in Cruising mode, not full acceleration. Even a technology-heavy book is only half-deployed today, which softens the blow if the AI trade pauses.

“We did not decide to become an AI portfolio. The data decided, one monthly rebalance at a time. When the data changes, so will we.”

Why This Is Process, Not Coincidence

The deeper point is about how these decisions get made. The same three-pillar process produced the gold tilt in winter and the technology tilt in summer. Nothing about the method changed. Only the data did. This is what we mean by data-driven, not gut-driven.

• Stock Universe. A wide screen of large-cap US stocks, filtered on quality and momentum. Whatever sector posts the strongest fundamentals rises to the top. In February that was mining alongside memory and optical names. Today it is the AI supply chain.
• Optimizer. From the qualifiers, this engine selects which 15 to 30 names enter the book. It chooses the names, not the position sizes. Every holding is then set to the same weight, so no single AI thesis can quietly grow into 20% of the portfolio. We explain the reasoning in our piece on portfolio construction.
• Risk Overlay. A daily check of 23 named indicators decides how much of the portfolio we actually run. It does not care whether the holdings are miners or chipmakers. It scales market exposure up or down based on conditions. Know when to accelerate, know when to brake.

What Happens If AI Cools?

A reasonable worry follows from all this. If the portfolio is 80% technology, are we trapped in the trade when it turns?

No. The book is rebuilt every month, and every name must re-qualify. If earnings revisions fade or price momentum weakens, those names drop out. New names take their place, drawn from any sector. They are the candidates that rank highest on the same factors that month, the best opportunities the data surfaces. The rotation from materials to technology is the proof. It ran in the other direction not long ago, and it can run that way again. It can also move into a different sector entirely, if that is where the data points next. This is sector rotation at work, decided by the process rather than by us. Whichever sectors the data ranks strongest are the ones we own.

For a closer look at how the current AI exposure is built, and the deliberate limits we place on it, see how our portfolio captures the AI boom while staying balanced. For the latest positioning and performance, see our most recent monthly review.

“Our performance is not a lucky bet on the right theme. It is the output of a repeatable process that rotates into whatever the data confirms is worth owning.”

*Past performance is not an indication of future results. Your capital is at risk. Holdings are accurate as of the publication date and rotate at each monthly rebalance.*

Weekly investment newsletter, every Sunday — positioning, performance vs SPY/QQQ, and what we're watching next. No spam.