The Infusive Consumer Global Leaders Fund delivered 5.3% for July. This brings year-to-date performance to 0.5% as at 31 July. This is net of fees for our USD A Class.
The Fund’s top contributors for the month were Microsoft, Amazon, and Apple, while the largest detractors were Tesla, Home Depot, and Walmart.
July was one of the most turbulent months for investors in years. At Infusive, we were encouraged by how we navigated this period.
The sharp rotation away from crowded AI proxies exposed how quickly enthusiasm can turn into fragility when leverage, concentration, and narrative collide. The Goldman Sachs Hedge Fund VIP index fell 12% relative to the S&P 500 – its worst relative month since inception in 2001 – with pressure amplified by the reported liquidity event at Situational Awareness, a large AI-focused hedge fund.
For us, the lesson was clear: it is not enough to own a theme; investors must understand the theme. We remain constructive on AI, but measured, disciplined, and selective. We believe AI can enhance the value of incumbents with distribution, data, customer trust, balance-sheet strength, and infrastructure.
The July dislocation also widened the opportunity set. Beyond hyperscalers, we see attractive mispricings in high-quality payments, consumer staples, discretionary, and digital businesses where select share prices appear to reflect excessive pessimism.
Our portfolio is deliberately broad, covering the consumer landscape, and our posture is active but disciplined: avoid crowded enthusiasm, stay close to fundamentals, and use volatility to add to strong businesses when fear creates opportunity.
The AI debate: expectations versus reality
The market has spent much of the year debating whether AI will destroy incumbent profit pools or reinforce them. Our view has been more nuanced.
We have used AI internally for several years at Infusive, giving us practical experience of both its power and its implementation challenges. The technology is clearly transformative, but adoption is not automatic. Organisations must make difficult decisions, including:
- Which technology stack to use
- Which models to rely on
- How to manage compliance and governance
- How to organise data so AI tools can retrieve and use it effectively
- How to bring colleagues along culturally and operationally
The last point may be the most important, in our view. AI adoption is not only a technology problem; it is a people, culture, and management problem. Companies whose leaders personally understand and champion AI appear to be moving faster than companies where AI strategy has been delegated to middle management.
This matters because it shapes who benefits. The companies most likely to capture value are those with:
- Leadership teams that think through an AI lens
- Well-organised, centralised, retrievable data
- Existing customer relationships and distribution
- The balance sheet to invest through uncertainty
- A culture willing to change workflows rather than merely add new tools
AI may lower the cost of creating new products, but it does not eliminate the need to distribute them, integrate them, sell them, support them, and persuade large organisations to change how they operate.
The “SaaSpocalypse” thesis – and why we disagreed
Earlier in the year, a popular market narrative was the so-called “SaaSpocalypse”: the idea that AI would commoditise software by dramatically reducing the cost and time required to build new products.
The argument was directionally understandable. A software product that might once have required significant capital, time, and headcount can now potentially be built far more quickly and cheaply using AI tools. In simple terms, what may have taken $100 million, three years, and a large team in 2010 might now be prototyped in hours for a fraction of the cost.
However, this framing misses what we believe is the more important point: product creation is being commoditised, but distribution is not.
In an AI-enabled world, the ability to create product may become abundant. The ability to place that product in front of billions of users, persuade enterprises to adopt it, integrate it into workflows, satisfy regulatory and compliance obligations in a secure environment, monetise it, and retain the customer relationship remains scarce.
That is why we believe the value has shifted from product alone toward distribution, trust, data, and scale. A captive and engaged customer base may be more valuable after AI, not less.
This is a crucial distinction. We do not believe it is easy for new AI companies to immediately capture the full customer base of incumbent platforms. Nor do we believe incumbents are immune from disruption. Instead, we see a more likely intersection: AI-native model providers partner with large incumbents because select incumbents offer the fastest route to distribution, monetisation, and enterprise adoption. This is the opportunity for stock pickers.
Hyperscalers as AI toll roads
The hyperscalers – Microsoft, Amazon, Alphabet, and Meta – sit at the centre of this debate.
These companies have collectively spent more than $1 trillion on AI infrastructure, including data centres, semiconductors, power, and related infrastructure. Many investors feared this spending would depress returns on invested capital for businesses that have historically produced high margins, strong cash generation, and formidable barriers to entry.
Our view has been different. We believe these businesses are becoming toll roads in the AI economy.
Every AI model relies on infrastructure. OpenAI, Anthropic, and other model providers need compute, data centres, networking, energy, and chips. The hyperscalers are among the few companies with the scale, balance sheets, technical expertise, and existing customer relationships to provide this infrastructure globally, and so far they have kept pace with insatiable demand.
At the same time, they own distribution assets that are difficult to replicate:
- Meta has more than 3.2 billion monthly active users
- Microsoft is embedded across a substantial majority of Fortune 500 companies
- Alphabet has multiple products with more than a billion users
- Amazon is deeply integrated into both enterprise infrastructure and household consumption
The latest earnings season provided meaningful evidence that the market’s concerns may have been too pessimistic. Cloud revenue backlog for hyperscalers is growing at approximately 130% year-on-year. Microsoft, our largest position, rose around 25% after earnings. Amazon, a top-five position, rose around 20% after earnings. We did not have a position in Meta over earnings, but at the time of writing, hold a modest position.
This is not simply a share-price observation. It reflects a shift in market understanding: hyperscalers are not just spending into AI; they are increasingly monetising the demand for intelligence, with a strong pipeline ahead.
The role of model providers: partners, not necessarily conquerors
The rise of companies such as OpenAI and Anthropic is extraordinary. Their growth rates and market impact are evidence that AI is creating new categories and changing the competitive landscape quickly.
However, the question is not whether these companies matter. They clearly do. The question is how they scale commercially and economically.
Our view is that model providers need paying customers at scale. The fastest route to those customers is likely through partnerships with large incumbents that already control enterprise and consumer distribution. This creates a more complex outcome than the simple disruption narrative.
Rather than AI companies destroying incumbents outright, we believe many AI companies will enhance the value of incumbents by:
- Embedding better intelligence into existing workflows
- Improving productivity across large customer bases
- Helping incumbents monetise their data and distribution more effectively
- Creating new reasons for customers to remain inside existing ecosystems
This is why we see AI as more likely to enhance many high-quality incumbent businesses than destroy them. That view remains differentiated versus parts of the market.
Adoption risk and customer behaviour
A key risk debated by investors is whether AI-related infrastructure will depreciate faster than expected. If data centres, chips, and related assets become obsolete quickly, the hyperscalers’ capital spending could prove less attractive.
We acknowledge this risk, but we do not think it is the whole story.
Not every AI task requires the most advanced model or the most expensive compute. A tiered model ecosystem is likely to develop. More complex tasks will justify premium models and premium pricing, while simpler tasks – summarising notes, drafting emails, retrieving information, basic workflow support – may be handled by cheaper or older models.
This matters for infrastructure economics. If different workloads can be routed to different models and different levels of compute, then not all infrastructure becomes obsolete at the same pace. We are already seeing more sophistication in how users and enterprises direct workflows based on task complexity and cost.
There will still be customers willing to pay for the best model across all tasks, especially where they view model quality as a competitive advantage. But cost sensitivity should create a broad spectrum of demand rather than a single winner-takes-all model environment.
Our conclusion is balanced: adoption and spending risk is real, but the market may be underestimating the durability of demand across different tiers of AI compute. We think it is demand that will surprise positively, which may in fact increase the value of legacy infrastructure over time.
Mispricing outside technology: payments and consumer opportunities
The AI rotation also created opportunities outside the technology sector.
In payments, companies such as Mastercard and Visa were trading at 20–30% discounts to multi-year valuation averages, despite retaining attractive structural characteristics. These businesses benefit from global consumption, digital payments adoption, scale, network effects, and strong profitability.
We are also seeing more compelling valuations in traditional consumer segments, including restaurants and other discretionary categories, such as luxury. Some of the best consumer businesses in the world are now reflecting valuations rarely seen outside black swan periods. For example, Coca-Cola delivered strong numbers and raised its guidance. Ferrari surprised the market with strong sales of the new EV that commentators had panicked about. The world keeps turning and consumers keep voting with their wallets.
This does not mean everything is cheap or that cyclical risk should be ignored. It means the opportunity set is widening. Fear has moved from speculative AI areas into selected high-quality consumer businesses where the fundamental deterioration implied by the share prices may be too severe.
What’s next?
The portfolio is intentionally diversified across different ways consumers and enterprises spend money.
This breadth matters. It aims to position us in the AI-driven growth opportunity while also maintaining exposure to businesses with defensive characteristics, habitual consumer demand, and long records of compounding. We receive feedback from our clients that they like this complementary composition.
Consumer behaviour is often more resilient than market narratives suggest. People continue to seek convenience, value, aspiration, status, comfort, and better lives for themselves and their families. Wars, pandemics, technology shifts, and economic cycles may alter the path of demand, but they do not eliminate many underlying human behaviours.
We are looking for durable brands, strong distribution, resilient customer behaviour, proven management teams, and valuations that compensate us for uncertainty.
Conclusion
We believe AI is and will be one of the most important forces shaping business over the next decade. But the market’s early interpretation of that force has been too narrow.
AI does not simply destroy incumbents. It increases the value of data, distribution, customer trust, infrastructure, and management quality. In many cases, those advantages sit inside large existing “boring” businesses.
At the same time, the market’s fixation on AI has created mispricings elsewhere. High-quality consumer businesses have at times been marked down despite durable long-term characteristics. This gives us a broader opportunity set.
Our portfolio is built around that breadth. We own businesses exposed to the broad consumer landscape. We are seeking to compound capital through the cycle while remaining sensitive to valuation and volatility.
We are optimistic about the long-term opportunity, alert to the risks, and focused on using market dislocation to capitalise in a measured way.