2026年展望：代理型人工智能与企业软件的未来研究报告

Enterprise Software ASHLEY MACNEILLManaging Director,Co-Head of Capital Markets MONTI SAROYASenior Managing Director,Co-Head of Flagship Fund K E Y TA K E AWAY S We believe Agentic AI will transform enterprise software from tools that support workinto systems that perform work on behalf of employees. In our view, this represents a History shows that major technology transitions have consistently driven software expansion.We believe AI will follow the same pattern, and that the amount of software consumed will In our view, widespread enterprise adoption of Agentic AI is not a question of if, but when.Roughly 20 to 40 percent of workers are using AI in some capacity in the workplace, but only1% of enterprise data is currently incorporated into AI solutions,1suggesting the large-scale With an estimated $11 trillion in cumulative value projected to flow to AI software andinfrastructure by 2030,2we do not see this transition as a zero-sum game between modelproviders, hyperscalers and software. We expect durable value to accrue to software that can AI is expanding addressable markets for software and may raise the ceiling on revenue percustomer, potentially improving unit economics. This shift in the business model may unlock Not every software company will make the transition. Some will evolve into Agentic AIplatforms, while others will use AI primarily to drive margin expansion and operationalefficiency. Those with generic workflows, limited proprietary data, and shallow customer More than 96% of enterprise software companies are private.3Public markets therefore reflectonly a fraction of AI-driven activity in the sector, and we believe private markets exposure is animportant consideration for investors seeking to participate in the Agentic AI opportunity. In a transition of this complexity, manager selection may matter as much as asset classexposure. Investors should evaluate whether their partners have the operational depth and The emergence of Agentic AI has introduced uncertainty in public markets and prompted afundamental question from investors: what role does software play in a world where AI can take Introduction Capital has flowed aggressively toward semiconductors, infrastructure providers, hyperscalersand foundational AI model companies. In some quarters, the prevailing narrative is that AI will We believe that narrative is materially incomplete. Historically, major technology transitionscreate periods where public markets struggle to price changes to software delivery and businessmodels, but in each of these transitions, demand for software ultimately expanded. We believethe transition to Agentic AI will follow a similar arc. In our view, software is not being replaced but In this paper, we examine the evidence behind that conviction and explore: •Why the current dislocation mirrors prior transitions•How Agentic AI can expand the economic opportunity for software•Which structural attributes could determine the winners in this new landscape•How AI is reshaping the software business model and what that means for valuations SoftwareHas EvolvedThrough PriorDislocations Public software markets have experienced dislocations during prior technology transitionperiods. Since 2014, the software sector has experienced nine downturns of meaningful A notable example was 2014–20, when SaaS valuation multiples compressed as investors askedwhether cloud platforms like Amazon Web Services would displace traditional software vendors.What followed was a decade-long expansion. Ultimately, enterprises didn’t abandon theirsoftware, they consumed more of it. Cloud infrastructure made software faster to deploy, easier We see a similar dynamic at work today. The concerns shaping the current narrative — AI-native disruption, changing revenue models and uncertainty around enterprise adoption — arenot irrational. But they are getting ahead of the evidence. Importantly, approximately 96% ofsoftware companies are private.2Public equity indices therefore capture only a fraction of the AI Infrastructure toEnterprise Execution:The Next Phase ofValue Creation Prior technology transitions have followed a recognizable pattern of value creation. Forexample, the chart below shows relative stock performance during the mobile internetera. Semiconductors captured value first. Hardware and infrastructure followed. Waves of Value: Relative Stock PerformanceDuring Mobile Internet Era Today, we’re seeing a similar story unfold where the initial phase of value creation wasconcentrated in semiconductor manufacturers and compute infrastructure providers. Asecond phase has accrued to foundational AI model providers, hyperscaler platforms W H AT D O W E M E A N Workflowsare the step-by-stepprocesses that companies followto get work done. They definewhat happens, in what order andwho is responsible. For example,how an invoice gets approved,how an employee is onboarded orhow a customer issue is resolved.In enterprise softwa