The Minotaur Global Opportunities Fund fell 7.8% in March, underperforming the MSCI ACWI (Net, AUD) which fell 3.4%. This was disappointing with March compressing a lot of previously earned alpha into one month, but it didn’t erase the broader case for the strategy. This was the first underperforming month since November 2025 although it was the worst month of relative performance since inception and the gap to the benchmark was wide. Many of our larger exposures – memory, precious metals, European defence, and growth-oriented technology – fell simultaneously in a way they hadn’t before.

Whilst that obviously hurt, our response has been to lean in, not pull back. Some of the positions that were hit hardest are ones where the 3-to-5-year thesis has actually improved. SK hynix and Micron are cheaper today on better fundamentals. The same is true across parts of software and our European defence holdings. When prices fall and the business hasn’t changed, the expected return goes up. That’s the maths, and we’re acting on it.

We’re conscious that every fund manager says this after a bad month. The difference, we hope, is specificity. We can point to Micron trading at 3.5x next year’s earnings with supply locked up for years. We can point to GitLab at $22 building the governance layer that every enterprise will need as AI coding agents scale. We can point to European defence stocks correcting 15-25% while NATO spending grew 20% in real terms last year. These aren’t vague assertions that things will work out, they’re concrete positions where the gap between price and value widened this month.

We’ve also been cutting what isn’t working. IperionX, Wizz Air, First Solar, and several other positions were exited during March. We don’t hold names out of stubbornness. But where the thesis is intact and the price has moved against us for macro reasons rather than fundamental ones, we think the right response is to add and that’s what we’ve done.

We explore the sectors that didn’t work and our response in more detail below.

Memories… of the way we were…

Memory was the largest single detractor this month, with SK hynix and Micron both falling sharply despite reporting some of the best fundamentals either company has ever seen. Several things hit at once. Micron’s result was excellent but higher capex guidance revived investor memories of past cycle blow-offs. Google’s TurboQuant paper then landed at exactly the wrong moment, and the market leapt to the conclusion that AI models might need structurally less memory. Add in a sell-the-news move after enormous share price runs, plus a broad Iran-driven risk-off tape, and the sector went from leadership to liquidation in a matter of days.

Our view is that the market has once again mistaken inference efficiency for demand destruction. The headline TurboQuant numbers compare against baselines frontier labs largely no longer use, the technique itself is not new, and software-side compression does nothing to change the physical HBM shipped inside GPUs and TPUs. More importantly, lower inference cost has so far increased total usage, not reduced it: longer context windows = more agents = more deployment. Supply remains sold out through 2026, with new capacity not arriving until late 2027, Micron has signed its first five-year customer agreement, and both Micron and SK hynix now trade at valuations that look cheaper relative to their fundamentals than when we first bought them. We have since added to our memory position.

Not being precious

Precious metals were the other notable frustration. We put the basket on as hard-asset insurance, and March delivered exactly the sort of geopolitical shock that should, in theory, have helped. Instead, gold fell and the miners fell harder. The transmission mechanism was our downfall. The Iran-driven oil spike pushed inflation expectations and real yields higher, strengthened the dollar, and overwhelmed the usual safe-haven bid. In other words, the hedge did not fail because the world became less dangerous; it failed because the market responded first through rates.

That is an uncomfortable outcome but also a useful reminder that macro relationships are not static. Gold, growth, and European defence all sold off together in March, a combination we would not have expected going in. Our takeaway is not that hard assets have no role, but that some exposures are better inflation hedges than crisis hedges, and some miners are more cyclical than defensive when real yields rise quickly. We are reviewing that part of the book with a colder eye as a result.

No shelter in the arsenal

European defence was a headwind this month because the sector started to trade like expensive growth in a rising-yield environment. This was despite the fact that the rearmament story hasn’t changed. Rheinmetall’s guidance seems to have broken sentiment across the group at a point where share prices already reflected a lot of good news. Then came the odd “sell the war” dynamic: as the Iran conflict escalated, investors unwound positions that had already become crowded, and defence stocks fell alongside gold rather than offsetting it.

We think that was a valuation reset and not a thesis break. European NATO spending grew strongly last year, Germany remains on a path to materially higher defence budgets over the rest of the decade, and order backlogs across the major primes remain at or near record levels. A multi-year rearmament cycle does not end because a few expensive stocks corrected in one month. If anything, the sell-off gives us better entry points into a theme we still think is early rather than late.

The Agent Provocateur

Software hurt us in March though the more important point is that this is not one clean call. Our view has shifted over recent months: constructive in December, net short in January, covering into weakness in February, and then rotating selectively back into names where we think the market is overestimating AI disruption. In hindsight, we were early. GitLab sold off on weak guidance, MongoDB fell sharply, and the market is still applying a blunt “AI loser” label to a wide range of businesses whose economics are actually being affected in very different ways.

That bluntness is creating opportunity. In GitLab’s case, AI does not reduce the need for the product; it increases it. AI coding agents are generating more code, but enterprise review backlogs, security issues and pipeline failures are rising with that complexity. More code without governance is a liability, and GitLab is building the governance layer that lets enterprises safely ship what their agents produce. Intuit is a different version of the same theme. Its moat sits in the accountant-SMB relationship, and AI makes that network more useful to both sides rather than disintermediating it. We may have leaned back into parts of software too early this month but we remain steadfast that some of the best opportunities are in names the market has prematurely written off.

What worked

Not everything moved against us. Cybersecurity was the strongest contributor, which made sense in a month when geopolitical stress translated directly into higher perceived cyber risk, especially given Iran’s history. Cloudflare rallied strongly, Palo Alto contributed positively as we added to the position, and we initiated CrowdStrike.

The short book also did its job, with Mitsubishi Motors the standout contributor as traditional automakers wore both the oil shock and the accelerating EV transition narrative. We also moved quickly into three Latin American oil producers that benefited from higher crude prices without carrying direct Middle East exposure. Along with positive contributions from Unity and Intuit, these positions did not offset the broader drawdown but they did show that responsiveness still matters when the tape turns chaotic.

What stands in the way becomes the way

March was painful and there is no value in pretending otherwise. But bad months are also where process is tested most honestly. We exited positions where conviction had weakened, added where the gap between price and value widened, and finished the month with a portfolio that we think is stronger, not weaker, than the one we started with. One difficult month does not settle the questions that matter over the next three to five years. Our job from here is the same as always: stay specific, stay humble, and keep allocating capital toward the dislocations where expected returns have improved the most.

Given we see this monthly performance as a departure from our usual, we thought we’d end with a Roman philosopher instead of a Greek one. March was challenging but as Marcus Aurelius tells us “The impediment to action advances action. What stands in the way becomes the way.”