At Carmignac, we offer pure-play Private Equity exposure through open-ended funds focused on LP-led secondaries, complemented by direct investments1 in high-conviction companies. Within this direct allocation, we may include, on a highly selective basis, exposure to AI in specific parts of the value chain.
Artificial intelligence is undoubtedly reshaping technology investment, and while we at Carmignac are firmly convinced of its integral role and excited about its potential, we are also acutely aware of the potential risks associated with the hype and “FOMO”2 it tends to create.
As enthusiasm has matured, concerns around excessive capital expenditure and elevated valuations have begun to temper the earlier exuberance. Against this backdrop, our Private Equity range neither treats AI as a single, all-encompassing investment theme nor shies away from AI-related exposure altogether.
Instead, we pride ourselves on being highly selective, focusing on businesses that stand to benefit from the broader adoption of AI without requiring us to predict which individual companies or technologies will ultimately emerge as winners. This is particularly important in private equity, where capital is committed for the long term. When we do invest in an AI platform, we require a unique and defensible value proposition, such as critical data security and technological sovereignty.
This discipline in our investment selection is illustrated by three current portfolio companies, each offering a distinct route to participate in technology- and AI-driven growth.
AHEAD: monetising the conviction in long-term AI adoption rather than predicting the winning model
AHEAD illustrates our preference for broad-based AI exposure. The company helps large US enterprises modernise IT environments through cloud migration, infrastructure optimisation, data management, automation and managed services. As organisations move from AI experimentation to implementation, they require specialist engineering, integration and ongoing support across increasingly complex hybrid environments. AHEAD can therefore monetise the overall growth of enterprise AI adoption without needing to predict which individual solutions or providers will emerge as winners.
The risk/reward profile is reinforced by its scale, diversified blue-chip customer base, engineering-led workforce and multiple growth levers across organic expansion and M&A. While AHEAD remains exposed to IT spending cycles, integration requirements and skilled-labour constraints, its contract-based, services-focused model, together with its broad vendor relationships, makes it a resilient way to participate in enterprise technology adoption.
Mistral AI: scarce, direct exposure with a defensible proposition
We are particularly proud that our proprietary sourcing secured access to a highly sought-after, oversubscribed financing round in one of Europe’s pre-eminent AI champions. While the team was well aware of the highly competitive dynamics, we were convinced of Mistral’s unique, defensible and sustainable value proposition, backed by already-secured partnerships and contracts that provide evidence of its competitive edge.
Mistral’s open-weight AI models address increasing demand from governments, regulated industries, and large enterprises for data security, control over sensitive information and reduced dependence on non-European technology providers. Its capabilities span frontier models, enterprise software, customisation and compute, while blue-chip deployments in industrial design, engineering and mission-critical operations point to use cases that extend beyond generic chatbots and may support deeper integration and higher switching costs.
At the same time, our team explicitly recognises the risks: a rich entry valuation before considering forward revenue potential, substantial future capital needs, fast-moving competition, and uncertain exit liquidity. Mistral is therefore positioned as a source of asymmetric upside and strategic optionality, supported by a defensible proposition around data security and technological sovereignty rather than a generic wager on AI valuations.
Project Soar: servicing the infrastructure behind AI
Project Soar is a direct co-investment in a US mechanical, electrical and plumbing (“MEP”) services company benefiting from strong growth in data-centre demand. This provides another form of secular exposure to the broader adoption of AI: its services are required as AI workloads expand, irrespective of which individual companies gain market share. The team is therefore confident that Project Soar can benefit from the industry’s broader advancement without relying on a single technology bet.
The investment case is not dependent on AI alone. It also rests on consolidation in a fragmented market, a proven acquisition engine, and the expansion of recurring service and maintenance revenues. Customer concentration and any slowdown in data-centre investment remain risks, but diversified end markets and multiple value-creation levers reduce reliance on a single technology, customer or spending cycle.
Selective participation, not generic, thematic exposure
At Carmignac, we seek highly selective exposure to AI’s advancement without relying on any single technology or perceived winner. We are proud to offer our clients access to, what we consider, quality companies, including hard-to-access companies such as Mistral AI, through our Private Equity fund range: Carmignac Private Evergreen and Carmignac ELTIF Evergreen.
1Target exposure of c.15% for direct investments.
2“FOMO” refers to “fear of missing out”: the impulse to follow the latest trend or hype, sometimes without fully understanding its underlying fundamentals or drivers.
Carmignac Private Evergreen A EUR ACC
- Recommended minimum investment horizon
- 5 years
- Risk indicator*
- 6/7
- SFDR - Fund Classification**
- Article 8
*Risk Scale from the KID (Key Information Document). Risk 1 does not mean a risk-free investment. This indicator may change over time. **Sustainable Finance Disclosure Regulation (SFDR) 2019/2088. The SFDR classification of the Funds may change over time.