Agefi Luxembourg - juin 2026

Juin 2026 23 AGEFI Luxembourg Fonds &Marchés By Monica DEFEND, Head of Amundi Investment Institute T he comparison is almost inevitable. As artificial in­ telligence continues to drive equitymarkets higher, investors in­ creasinglywonder whether they arewitnessing a repeat of the dotcombubble that burst at the start of themillennium. The similarities are clear.Arelatively small group of technology compa­ nies is responsible for an outsized shareofmarketreturns,whileenthu­ siasmforatransformativetechnology continues to fuel growth expectations. Yet a closer look suggests that today’sAIdrivenmar­ ket differs fundamentally from the speculative ex­ cesses of the late 1990s. The more relevant question may not be whether AI represents a bubble, but whether investors fully understand the risks associ­ atedwith an increasingly concentratedmarket. FundamentalsMatter This Time The defining characteristic of the dotcom era was thedisconnectbetweenvaluationsandbusinessfun­ damentals.Investorspouredcapitalintocompanies with little or no revenue, often based solely on future growth expectations. When those expec­ tationsfailedtomaterialise,valuationscollapsed. Today’sAI leaders tell a different story. CompaniesattheheartoftheAIecosystem,includ­ ing Nvidia, Microsoft and Alphabet, gen­ erate substantial revenues, strong cash flows and significant profits. Impor­ tantly,earningsgrowthhaslargelykept pacewith shareprice appreciation. As a result, valuation multiples have actually declined despite strong mar­ ket performance. The pricetoearn­ ings ratio of a broadAIrelated equity universefellfromaround40in2023to approximately 34 in 2025. This suggests investors are paying less for each unit of earnings than they were two years ago, a sharp contrast with the late stages of the dotcom boom. Currentvaluationsremainmuchmorecloselylinked to underlying fundamentals. Concentration Is the Real Risk While valuations appear more grounded, market concentrationhasbecomeagrowingconcern.AIre­ latedcompaniesnowaccountforroughly38%ofthe S&P500.Althoughthisremainsbelowthepeaktech­ nology weighting seen during the dotcom era, ex­ posure is concentrated inamuch smaller number of firms, around 44 companies compared with nearly one hundred at the turn of the century. This means overall market performance is increas­ ingly dependent on a limited group of dominant technology companies. Many investors gain expo­ sure throughpassive strategieswithout fully appre­ ciating the concentration embedded in their portfolios.As a result, diversificationmaybeweaker than headline allocations suggest. Massive InvestmentsMust Deliver Another defining feature of the current AI cycle is the unprecedented scale of investment. Technology giants are committinghundredsof billionsof dollars to data centres, computing infrastructure and spe­ cialisedhardware. Relative to their asset base,AIre­ latedcompanies are investingatmore than twice the rate of the broader market. The challenge is straightforward: these investments musteventuallygeneratecorrespondingprofits.IfAI adoption develops more slowly than expected, or if infrastructure capacity expands faster thandemand, investors could begin questioning the returns on these enormous capital expenditures. At the same time, balance sheets remain healthy. Debt levels amongAIrelatedcompanieshavefallensignificantly in recent years, suggesting that much of this invest­ ment boom is being funded through internally gen­ erated cash flows rather than excessive borrowing. Conclusion The AI boom is often compared with the dotcom bubble, but the analogy only goes so far. Unlike the late 1990s, today’s market leaders are highly prof­ itable businesses with strong balance sheets and earnings growth that largely supports current val­ uations. The evidence suggests investors arefinanc­ ing real technological progress rather than speculative promises. The greater challenge lies elsewhere. Financialmar­ kets are becoming increasingly dependent on a small number of companies andon the assumption that today’smassive infrastructure investmentswill generate tomorrow’s profits. The key risk is there­ fore not a replay of the dotcom collapse, but whether a highly concentrated group of market leaders can continue delivering on exceptionally high expectations. For investors, that distinction may prove more im­ portant than the bubble debate itself. AI is not the new dot-com bubble V enture capital is shifting fromgrowthdriven ex­ pansion to a concentrated, resiliencefocusedmodel, where AIled deals, strategic sectors, and realworld deployment redefine investment dynamics. When the recovery narrative breaks down At first glance, 2025 seemed to confirma stabilization in venture capital markets, supported by resilient capital deploy­ ment and a limited number of highpro­ fileexits,largelydrivenbyAI.Yetbeneath these aggregates, themarket told amore constrained story: rather than a broad­ based recovery, activity remained highly selective, with fewer deals, stable capital levels, and increasing concentration around large transactions and latestage technology. Exit markets reflected the same asym­ metry, with a limited number of initial public offerings (IPOs) and strategic M&A transactions capturing a dispro­ portionate share of value. Taken together, these trends suggest that 2025didnotmarka return togrowthled venture dynamics, but rather the consol­ idation of a deeper structural shift. 2025 & early 2026 in review: fewer deals, larger tickets, structural concentration A stagflationary and barbell venture market Across both Europe and the United States, 2025 was characterized by a con­ tinued decline in deal volumes along­ side broadly stable total capital invested. This apparent paradox reflects a stagfla­ tionary market dynamic, where capital remains available but increasingly con­ centrated. In this context, the large and latestage transactions sustained overall investment levels, while earlyandmid­ stage activity continued to contract, par­ ticularly in Europe. As a result, the market has evolved to­ ward a barbell structure: fewmegadeals absorbing a disproportionate share of capital,coupledwithasustainedpullback across the broader venture ecosystem. Rather than signaling renewed momen­ tum,thisconfigurationreflectsaflightto­ wardperceived certainty. Investors are prioritizing scale, balance sheet resilience, and clearer deployment pathways over breadth and experimen­ tation, contributing to a more selective and polarized investment landscape. A structural inflection rather than a cyclical pause Early 2026 confirms an acceleration in activity, but not a normalization of mar­ ket conditions. While both deal volume and total cap­ ital invested have increased, this re­ bound remains highly concentrated, with artificial intelligence capturing a growing share of total deal value and continuing to shape investment flows across adjacent sectors. Beyond AI and a narrow group of strategically positioned sectors such as advanced manufacturing, robotics, de­ fenserelated technologies and energy infrastructure, much of the venture uni­ verse continues to experience subdued activity. Investors are reassessing scala­ bility assumptions, capital intensity, and exit visibility, particularly for early and midstage companies. Exit dynamics mirror this asymmetry with the overall liquidity dominated by a handful of transactions, while the broader exit environment remains constrained. Increasingly, exit activity favors, but increasingly reflect a qual­ itative shift: liquidity concentrates around a small number of transac­ tions, with a growing preference for companies embedded in critical infras­ tructure, industrial systems, or regu­ lated environments, where strategic relevance and deployability outweigh traditional growth metrics. Taken together, these trends point to a deeper structural shift in venture cap­ ital. The market is not undergoing a temporary slowdown, but transition­ ing toward a model defined by con­ centration, selectivity, and a renewed focus on resilience and realworld de­ ployment. What’s next: a structurally redefined venturemodel where innovation is structured under constraints Continuedconcentrationacrossstrategic sectors Recent geopolitical and economic de­ velopments have reinforced the role of venture capital in addressing strategic and critical needs. States and suprana­ tional institutions are increasingly en­ couraging and, in some cases, relying on earlystage companies to respond to crisisdriven and sovereign priorities. Support to defense industry is showing strong and niche development into subvertical involving independence and unmanned technologies: USV (un­ crewed surface vehicles), UGV (un­ manned ground vehicles) and UAS (unmanned aircraft systems) are struc­ turing this highly specialized market. Support for defense and security ecosys­ tems has intensified. For instance, the launch of a defensefocus structure by a French public institution illustrates a clear policy shift toward financing un­ listed companies operating in strategic areas such as cybersecurity, aerospace, and defense technologies. More broadly, capital is increasingly di­ rected toward technologies with direct linkstoindustrialandstrategicinfrastruc­ ture.Swarmroboticsexemplifiesthisshift: innovation is no longer centered on indi­ vidualsystems,butonthecoordinationof distributed, autonomous units capable of operatingunder extreme constraints. Initially developed for hostile and de­ fense environments, these technologies are nowexpanding into industrial appli­ cations from infrastructure maintenance to disaster response and mining where they offer greater efficiency and en­ hanced safety. Alongside robotics, nuclear fission tech­ nologies, particularly small modular re­ actors (SMRs), are emerging as an increasingly coherent investment theme. Unlike fusion, which remains distant from commercial viability, SMRs offer clearerdeployment timelines andarepo­ sitioning themselves as infrastructureen­ ablers rather than speculative bets. Across these sectors, a common pattern is emerging: operational, technological, and energy resilience is replacing growthatallcosts as the primary driver of capital allocation. This shift implies longer development cycles, higher capital intensity, and a renewed focus on realworld deployment. The return of a familiar capital pattern Venture capital is increasingly reverting to investment patterns historically ob­ served in sectors suchas space or energy. Private capital is no longer operating in­ dependently, but in conjunction with public demand and institutional frame­ works. This results in a shift toward more patient capital, capable of support­ ing longer time horizons andmore com­ plex development pathways. Institutional actors are playing an in­ creasing role in shaping this ecosystem. Initiatives involving public capital and strategic funding frameworks are pro­ viding direction, reducing earlystage risk, and facilitating capital allocation into complex sectors. The innovationrelated fund initiatedby a supranational organization illustrates how public capital is adapting to this new landscape. This fund functions less as a traditional financial return vehicle andmore as a strategic signaling instru­ ment, validating spend categories and derisking early participation by private investors. At the same time, newfinancingdynam­ icsareemergingwithinAIdrivenecosys­ tems. Recent initiatives suggest a shift toward supporting startups through ac­ cess to infrastructure notably compute and AI capabilities rather than relying solely on traditional equity financing. Thisevolutionreflectsabroadertransfor­ mation:accesstotechnologicalinfrastruc­ ture is becoming as critical as access to capital, reshaping earlystage funding models and competitive dynamics. Rather than replacing private investors, these actors are enabling a more inte­ grated investment environment, where institutional and private capital operate incombination tosupport theemergence of strategically relevant technologies. Taken together, these trends confirm that venture capital is not only returning to familiar patterns but evolving toward a hybrid model one where capital allo­ cation, industrial constraints, and plat­ formdriven dynamics are increasingly intertwined. Conclusion The venture capital market’s previous equilibrium is now shifting towards in­ dustrial execution. Recent examples make this shift tangible: an established nuclearenergycompany’sexpansioninto theU.S. (1) , supportedbyapartnership in­ cludingupto$2billionininvestmentinto nuclear fuel infrastructure (LesEchos), il­ lustrateshowventurebackedcompanies are moving rapidly toward largescale deployment and strategic integration. At the same time, a leading defense technology company’s (2) +87% surge in orders and €1 billion+ backlog (Reuters) highlights amarket increasingly driven by state demand and operational neces­ sity, particularly in defense and critical infrastructure. The direct implication is clear: venture capital is no longer about how fast companies grow, but how swiftly and effectively their products are marketed. In that sense, the industry is not shrink­ ing, it is being reindustrialized. Géraldine CANETTE, Assurance Partner Laurent CAPOLAGHI, Managed Services and Private Equity Leader Vincent RIETSCH, Assurance SeniorManager, EYLuxembourg 1) Nucléaire : la startup européenne Newcleo décrocheuncontratcapitalauxEtatsUnisLesEchos 2) Underwater drone maker Exail’s profit leaps as ‘highintensityconflicts’spurdemand|Reuters From AI exits to strategic resilience: How venture capital is being re-industrialized