July 2026

The fund returned +1.0% in July. The MSCI ACWI (AUD) declined 1.9% and the Nasdaq 100 fell 6.6%. The S&P 500 was flat.

July was our most active month since launch and tested both sides of the strategy: pursuing the best opportunities while protecting the downside.

AI stocks fell sharply in July. At its lowest point the Nasdaq 100 was down 10.6%, with the S&P 500 and MSCI ACWI (AUD) down 3.4%.

Our index hedge triggered on 16 July, helping protect the portfolio. As prices fell, we put capital to work.

Contributors

Detractors

The hedge did what it was designed to do. It reduced the impact of the market decline and provided funds to invest in a pullback. AMD and Dell did not work, and we cut both rather than carry the losses.

We closed July 85% invested, with our hedge reducing the market exposure to 49%.

We made significant changes through the selloff. We added Hewlett Packard Enterprise as AI spending moves into corporate data centres, increased our highest conviction software positions, bought back the hyperscalers, and invested in Praemium, an undervalued platform administrator with operating leverage. We exited AMD, Dell and Talen Energy due to stop losses.

We returned to gold equities for the first time since April, after gold fell from US$4,780 to US$4,050/oz, buying Northern Star at $20.6 and opening a position in St Barbara.

The portfolio is now broader: Alibaba and Tencent at 19x and 14x forward earnings; SailPoint and ServiceNow, priced for disruption while still growing earnings; Lilly in healthcare. We also own selected deep value cash generators and Innovaero ahead of its proposed IPO, while retaining our Australian bank short.

Investment cases for our themes and top holdings are at laserbeamcapital.com/thematic.

We enter every hedge expecting to pay for downside protection. It is an insurance policy. Our backtested system aligns index hedges with the market regime.

The system is a guide. It is deliberately blind and cannot see our holdings, earnings or macro calendar. Market judgement will always be required. We often add or remove the hedge early, within defined portfolio limits.

Tested across two decades of major crashes, the system roughly halves drawdowns at an average cost of 2% p.a., with years like 2022 recovering several years of those costs.

The GFC: a 52% index drawdown became 25%, and it finished up 11% over a period the index lost 11%.

COVID: a 27% decline became 15% but the hedged strategy finished the year up only 30% against 47% unhedged.

The 2022 to 2023 bear market: 35% decline became 17%, and it finished up 13% against the index down 6.0%.

The hedge helps us hold our investments through the months the market does not want them. And when it works, it provides profits to invest into oversold opportunities.

S&P 500 Q2 earnings growth reached 47% year on year by the end of July, from 25% two weeks earlier. The 2027 earnings estimate has cracked $400 for the first time, against $274 actual in 2025. This rally is supported by earnings more than a broader valuation re-rating.

As a global fund, we can invest wherever the risk reward is strongest. Eli Lilly is a core position and is up 11% on our entry in May.

Lilly just reported revenue up 48% and raised guidance to $85 to $87 billion. The bigger news was retatrutide: three more positive Phase 3 trials, with filing planned for early 2027. It is the main reason we own the stock.

Phase 3 trials show retatrutide delivered substantial weight loss and improvements in blood sugar, cardiovascular risk factors, knee pain and sleep apnoea severity. We believe it could become one of the world’s largest medicines. Lilly cost 27bp in July. We held 6.0% at month end and would add on weakness.

Open source intelligence is getting ridiculously cheap. Alibaba's Qwen3.8-Max-Preview now benchmarks close to leading US models at roughly a tenth of the price. DeepSeek is cheaper still.

We use open source and frontier models throughout our research and operations. Chinese models closing the performance gap at lower cost helped build our conviction in Alibaba [9988 HK] and Tencent [700 HK], now our two largest holdings.

Alibaba was a top July contributor. It trades at 19x and Tencent at 14x forward earnings, well below US peers. Cheaper models lower the cost of intelligence, expanding demand for compute, networking and power. Which feeds into our Hyperscaler and Neocloud investment cases.

We sold our hyperscaler winners into strength earlier this year as the trade got crowded. In June and July we bought them back: Amazon 16% off its high and Microsoft down 20% from the high.

Wall Street keeps underestimating the capex. Consensus 2026 hyperscaler capex has risen from US$546 billion in January to US$793 billion, up 45% in seven months. Analysts keep expecting a slowdown; the hyperscalers keep raising spending.

This is growth capex, building capacity for new revenue rather than replacing existing infrastructure. The four largest clouds have roughly US$2.3 trillion of contracted work between them. While the measures are not directly comparable, they show that much of the demand is already contracted years ahead. That matters because the usual bubble risk is supply being built before demand arrives. Here, demand is already waiting for the capacity.

The money they are spending is now generating a return. The economics are being proven. US hyperscaler cloud revenue is growing 43% on a $364 billion run rate, and all three clouds run positive and rising operating margins: AWS near 40%, Google Cloud from -15% in 2021 to +35% today. Demand for compute still exceeds what anyone can build.

Amazon delivered the result of the season so far: revenue rose 20%, AWS grew 37%, its fastest in 18 quarters, and backlog increased US$132 billion. Microsoft: Azure grew 43%, passed US$100 billion in annual revenue and has US$678 billion contracted work ahead. Alphabet: Google Cloud grew fastest and carries a US$520 billion backlog while Alphabet trades below peers. We bought all three before the results.

Power remains the bottleneck across generation, grids, switchgear and cooling. This underpins our small private holding in Powerplay AI, which targets 400 MW of powered land in West Texas.

Goldman Sachs estimates AI agents could push global token consumption (AI use) to 24x current global capacity by 2030. Token economics turned positive in the first half of 2026, making investment at this scale increasingly economic.

Whilst we're on private holdings, our other pre IPO position Innovaero, Australia's sovereign armed drone manufacturer, is scheduled to list in the first week of September. More on that next month.

We ended July with a stronger portfolio, adding to high quality businesses at more attractive prices. The hedge meant we could do this while finishing the month up.

Earnings are driving the market and we own companies delivering that growth. We also respect price action and will reduce exposure when stop losses are hit, the facts change or we hedge in response to market conditions.

Our approach is unchanged: concentrate capital in the best ideas, cut losers quickly, let the winners ride, and only invest when the setup is compelling.

Live performance and holdings are always updated at laserbeamcapital.com/performance.

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

Regards