June 2026

One year ago we launched the Laser Beam Fund with a simple ambition: compound capital at attractive rates and protect the downside. Year one was an encouraging start.

The fund returned +16.8% net of fees for the twelve months. We got there while running an average net exposure of 46% and holding close to half the book in cash. This letter covers what worked, what we learned and where we see opportunity ahead.

Our worst drawdown in year one was shallower than the market, 8.4% against 10.7%, and we were back at new highs by early May, around six weeks before the benchmark. We were positive in 7 of 12 months.

We’re aware: One year does not make a track record. And protecting capital has to be earned in every market selloff.

Every position starts with a theme that identifies where capital is flowing, followed by deep fundamental analysis, position sizing based on risk and reward, and usually a stop loss set on the day we buy and adjusted for volatility. If we are wrong, we exit quickly and accept a small loss.

Due to this stop loss strategy we trade a high number of positions, around 130 positions passed through the portfolio this year. The average winner was approximately 1.6x the size of the average loser, which is how the strategy is designed to work.

You can now explore the themes and individual holdings at https://www.laserbeamcapital.com/thematic. The full year performance by theme is below.

AI & Neoclouds
The year's biggest contributor. AMD was the single biggest winner: +6.6% on the shares, +4.1% net of option hedging. The April selloff put it on our re-entry list, we sized up when the setup confirmed, and trailing stops walked us out near the May highs. Nebius added +3.6%, Micron +2.4% and ARM +1.3%.

Critical Minerals & Gold
Nearly as big as AI. A sleeve of strategic minerals positions and gold producers added +6.7% between them. Currency debasement and strategic supply at work. When Software and Agents were smashed in February, this portion held up. The sleeve finished February and March up around 2% against software down close to 14%.

Macro trades
Added +5.7%, representing about one third of the year's gross return. This includes our actively managed currency positioning. We moved cash from Australian dollars into US dollars when opportunities arose, including ahead of the June selloff. Those decisions added meaningfully to returns. Currency is a position whether you manage it or not, and we choose to manage it.

Software & Agents
The year's biggest detractor. Five workflow software positions cost us about 9.0% between them, partly offset by Oracle (+1.7%) and DigitalOcean (+1.6%). What looked like an attractive entry point after the initial software selloff proved to be a value trap as the market continued to reassess the long term impact of AI agents on seat based software. Valuations compressed further and the sector suffered another sharp leg down. We exited the positions, accepted the loss, and rebuilt the sleeve around infrastructure software that agents run on rather than workflow software that agents replace.

Consumer & Internet, Power & Energy
Both detracted. Consumer internet names Vipshop, Sea and Alibaba cost roughly 2.7%. A data centre cooling position cost 1.9% before our stops took us out, offsetting wins in Exxon, Talen and crude oil trades. We still believe the underlying themes are right. Position selection within them has to be better.

We hedge at four levels: index futures and options, single stock options, commodities and currency. We enter every hedge expecting to pay a small cost for downside protection. It is an insurance policy.

This year the index hedge book made money. The Nasdaq futures shorts added +4.4% and the index put options cost 1.4% net. It is encouraging when we get paid to lower the volatility and reduce the downside of the book.

Our hedging process kept evolving through the year as we refined both the discretionary and systematic sides. The core index hedge is now rule based. It arms when the market breaks below trend, sizes to offset the portfolio's market exposure, and is closed when the trend turns up again. Alongside it, we still add discretionary hedges when we think conditions call for more protection.

The systematic overlay came out of our own testing. Backtested across Nasdaq data, the setting we run cut the maximum drawdown in the test period from 24% to 11%, more than half, while giving up roughly 1.5% a year in return. Return per unit of drawdown nearly doubled. Combined with our discretionary hedging, it helped produce a shallower drawdown and a positive contribution from the hedge book this year.

One area to improve: stock specific options. We hedged several positions into earnings using collars, selling calls to fund the puts. The results were blowouts, the stocks ripped higher, and the sold calls capped our share of the move. That layer cost about 3.8% this year, most of it upside given away rather than protection paid for. In future we will buy the puts outright and pay for the protection, rather than fund them by selling away the upside.

Right: de-risking early. We cut exposure into the October and November shakeouts and again before the June semiconductor selloff, and made money or held flat through all three. Buying the April low, which became our best month at +9.3%. Riding the AI beneficiaries on the way up and selling them toward the top over recent months as the trade grew crowded. Exiting Alphabet up 60%. Keeping cash high when setups were thin. Letting the winners run.

Wrong: the market consolidated through the new year and then sold off properly in February and March, and we were too slow to cut workflow software into it. It cost us. A couple of names where we bought the oversold value opportunity instead of the sector leader, and the leader kept winning. A handful of consumer internet positions that never earned their place. The hedges did their job in the pullback. The new positions we added did not. We put them on too early and got stopped out as the market kept falling. Every loss stayed manageable because sizing and stops held.

AI has been part of how I run this fund since day one, and I put significant time into refining it this year. It’s a tool, not a decision maker. The Engine runs fourteen specialist agents across research, risk, portfolio review and operations. They watch the book through the US, European and Asian sessions, grade every call once the outcome is known, and reconcile daily against our administrator. All the fund’s data is stored in a secure database for AI agents to access.

The Engine can suggest a trade. It never places one. Every portfolio decision is mine.

I made over 2,400 code commits to our AI server this year, and will continue to refine it. There is more detail on how it works in the Team section of our home page and at laserbeamcapital.com/engine.

One example is our live company research. Our fundamental analysis agent constantly scans filings, news and anything else affecting each holding, updates the investment thesis, and publishes the result automatically for every position at laserbeamcapital.com/thematic. I encourage you to read some of it in detail.

The hedging backtest in the section above is an example of the same work: tested on the Engine, stored in the database, now running live. All of it is proprietary. Every analysis, signal, grade and backtest is retained, and the database grows every day, giving us more evidence about what works, what doesn't, and where the process can improve.

I built Laser Beam around these tools from day one. Three years ago this level of analysis would have required a substantial team. Today I do it with software built in house, and our process sits on top of all of it.

We enter year two about 50% net, with 26% cash and hedges on. Concentrated in our best ideas, plenty of dry powder to add.

Consumer, Internet, Software and Agents remain our largest exposures, but the portfolio has evolved. Early in the year, semiconductor companies generated exceptional returns. As valuations became more demanding, we gradually recycled capital into businesses benefiting from the next phase of AI adoption: compute hardware, power generation and enterprise software.

The transfer of cash from the spenders to the receivers is unlike anything we have seen before. The five largest hyperscalers are expected to spend around US$725 billion on infrastructure in 2026, the largest corporate capital cycle on record.

We want exposure to the layers capturing the next marginal dollar of that spend, rather than the most crowded parts of the trade. Power is emerging as one of the key constraints on AI infrastructure. We own generation assets positioned to benefit, including a nuclear operator contracted directly to a major hyperscaler.

Outside AI, we are focused on electrification, healthcare, consumer and financials, backing the dominant profit pool in each industry. In batteries, that is CATL, still the only meaningful profit pool in cells. In healthcare, we have been steadily adding to Eli Lilly and the broader obesity theme as oral treatments expand the addressable market. In consumer, we own Asian internet leaders trading well below historical multiples. In financials, we are long the highest quality US bank and short the Australian majors. Where we believe AI is more likely to destroy value than create it, we stay out.

We also hold two small private investments, Innovaero (more below) and Powerplay AI. Separately, we have taken a position in a copper company listing in July, providing exposure to a commodity where long term supply remains constrained.

We remain concentrated where conviction is highest, hedged where appropriate, and maintain cash to deploy when opportunities emerge. Individual holdings will change, but the process will remain the same.

Year one reinforced our confidence in both our investment process and our ability to compound capital over many years. Year two is about sharpening our edge, deepening our research and improving our ability to identify and act on the best opportunities globally.

Every market cycle teaches us something and every investment decision adds another piece of evidence. The AI Engine continues to improve, our research coverage keeps expanding, and our hedging framework is more refined than it was twelve months ago. We expect that evolution to continue for many years.

In August, I will be in the United States meeting management teams, visiting companies and touring assets across several industries. On the ground research complements what we do from the desk year round. A key part of the trip is the Canaccord Boston Growth Conference, where around 400 global growth companies, including more than 50 private companies, are expected to present. Trips like this help us deepen industry knowledge, identify emerging trends and engage with businesses early.

One private investment we continue to monitor closely is Innovaero, a drone supplier to the Australian Defence Force, which is now targeting an IPO in early September after delaying its original timetable from late July. While the listing has moved back slightly, the additional time should provide greater visibility into customer contracts, the order book and growth pipeline before the company comes to market. We believe this additional information will allow the market to better assess the quality and value of the business.

Markets will always surprise us. Themes will evolve and individual holdings will change. That is exactly as it should be.

We are not trying to predict every move in the market. We are trying to back opportunities where the upside materially outweighs the downside, protect against permanent loss and give winners time to compound.

Laser Beam is the fund I had wanted to build for years: a modern global fund built for today's markets, combining deep fundamental research with technology developed in house.

Last week, after 11 years at Endeavor Asset Management, I stepped away from my other investment responsibilities to focus exclusively on Laser Beam. Eleven years is a long time, and I am grateful for every one of them.

The decision was simple. I believe this fund is something special, and it deserves my full attention. Thank you again for your trust during our first year. It won’t be forgotten.

Please email or call any time if you'd like more detail.

Regards