Reasons to invest in Carmignac Portfolio Tech Solutions for the next 5 years

On June 2, 2026, Marvell Technology’s shares rose by around 33%¹ in a single session after Nvidia CEO Jensen Huang described the company as a potential “next trillion-dollar company.” This example, among many others, may look like a simple anecdote; however, it is, in our view, symptomatic of today’s momentum-driven technology market: a single narrative, endorsement, or data point can trigger a disproportionate move in share prices.

The semiconductor rally has become increasingly erratic beneath the surface: with some companies such as Micron, Intel, Arm and Marvell, all up several hundred percent in Q2 2026, whereas stock like Nvidia has lagged indexes. This illustrates both the strength and the fragility of the current market environment. The rally remains highly concentrated in AI-related themes, but leadership can rotate abruptly, and valuation discipline is becoming increasingly important.

The response to this momentum-driven market of Carmignac Portfolio Tech Solutions is to own technology winners without becoming hostage to momentum. To achieve this, a meaningful portion of the portfolio sits outside mainstream global indices: around 50%² of holdings are not constituents of the MSCI ACWI Information Technology Index. This reflects our willingness to look beyond the dominant technology names in search of differentiated return drivers.

Another defining feature of the strategy is its emerging-markets exposure, which stood at 36%³ at the end of June 2026. In our view, this broadens the opportunity set, reduces dependence on crowded developed-market leaders, and provides access to growth drivers shaped by different macroeconomic, geopolitical and company-specific dynamics.

In a market increasingly dominated by mega-cap technology stocks, we also believe less researched and less crowded small- and mid-cap companies can offer fertile ground for active management. By focusing on businesses with a durable competitive advantage, attractive growth prospects and disciplined capital allocation, we aim to build a portfolio that remains both agile and diversified.

Finally, portfolio construction is actively managed through factor exposure. In particular, we seek to ensure that the portfolio’s valuation bias remains lower than that of the benchmark⁴, helping us participate in long-term technology growth while limiting exposure to the most speculative areas of the market.

As concerns around rate levels and stagflation re-emerge, investors may reasonably question how a tech equity strategy such as Carmignac Portfolio Tech Solutions would behave during sharp sector rotations that can be unfavourable to traditional growth funds.
Our primary way of limiting the potential impact of a shift in the market regime is to place valuation at the heart of our investment opportunity analysis process.

At the stock level, we focus primarily on EV/FCF (Enterprise Value to Free Cash Flow), which provides a more comprehensive perspective than simple earnings multiples by incorporating balance-sheet structure, financing costs and capital intensity. This reflects our broader philosophy: combining a traditional value discipline with a quality-growth mindset.

At the portfolio level, this disciplined approach creates balance between high-growth companies with lower free cash flow yields and lower-growth companies with more moderate free cash flow generation.

Beyond valuation discipline, we are also willing to adopt a genuinely contrarian stance when market expectations implied by valuations diverge from our assessment of intrinsic value. Being “on the other side” is not a permanent posture, but a state-dependent response to sentiment-driven mispricing—a useful portfolio construction tool, particularly in the current environment where the market regime is heavily influenced by momentum.

The core challenge for Carmignac Portfolio Tech Solutions is how to navigate a market where technology has become both the main source of optimism and the main source of concentration risk.

Markets remain highly momentum-driven, with narrow leadership and signs of excess in parts of the AI trade. In both the U.S. and Asia, equity indices have been carried by one dominant theme: semiconductors. What began as enthusiasm for the leading AI beneficiaries has increasingly spread to smaller and lower-quality semiconductor names, some of which have doubled or tripled in recent months. In several cases, valuations now appear to discount years of aggressive growth, despite limited visibility on fundamentals.

In the coming months, the AI debate is likely to shift. Investors are increasingly questioning whether hardware companies are capturing too much of the value created by AI. Semiconductor and hardware suppliers are benefiting from bottlenecks and pricing power, but rising infrastructure costs could eventually weigh on demand. Efficiency is also becoming a more important theme, with growing focus on cheaper models, open-source alternatives, and better routing of AI workloads to reduce token usage and infrastructure costs.

Tactically, this has led to a reduced exposure to hyperscalers and a move away from the most speculative semiconductor names. Instead, we are focusing on higher-quality leaders where valuations remain more attractive, such as Nvidia.

Within software, we are very selective. As AI moves from experimentation to production, value could accrue to some software layer that makes it usable at scale: orchestration, permissions, testing, security, deployment pipelines, and integration into enterprise systems. This is particularly relevant for developer software. If coding agents become one of the first large-scale enterprise use cases for agentic AI, companies will need stronger tools to manage AI-generated code. This supports incremental exposure to Atlassian and GitLab, which are embedded in developer workflows and could benefit from higher code-change volumes and greater governance needs.

Beyond this near-term caution, we believe the medium-term outlook for AI infrastructure remains robust. The cycle appears to be extending rather than peaking, supported by continued AI-related capex growth from the largest cloud platforms and by longer-dated infrastructure commitments from major AI customers. Importantly, demand is broadening beyond frontier model training. Agentic AI is moving into enterprise software, edge AI is creating new inference endpoints, and physical AI is adding another layer of compute demand through robotics, simulation, and real-world automation.

This suggests demand is not simply rotating from one chip generation to the next. Each new generation is likely to serve a broader base of workloads, customers, and deployment environments. As an illustration, demand for Nvidia’s current Blackwell platform could remain resilient for longer than a typical product cycle, even as customers prepare for the next-generation Rubin platform. The key point is that Blackwell and Rubin may not represent a simple replacement cycle, but rather successive layers of capacity absorbed by an expanding AI infrastructure market.