2025年一季度人工智能和机器学习风险投资趋势（英）2025

ContentsQ1 2025 AI & ML VC TrendsAI & ML landscapeAI & ML VC ecosystem market mapVC activityAI & ML VC deal summaryAI & ML VC deal summary 3451516 CONFIDENTIAL. NOT FOR REDISTRIBUTION.PG 2Institutional Research GroupAnalysisDimitri ZabelinSenior Research Analyst,AI/ML & Cybersecuritydimitri.zabelin@pitchbook.comDataMatthew NacionalesSenior Data Analystpbinstitutionalresearch@pitchbook.comPublishingReport designed byChloe Ladwig,Julia Midkiff, andDrew SandersPublished on July 8, 2025 AI & ML landscapeHorizontal platformsVertical applicationsSemiconductorsAutonomous machines AI & ML VC ecosystem market map VC activityAI captured one in every four VC dollars globally in 2024, a trend likely to accelerate as adoptiondeepens and infrastructure scales. Capital continues to flow into both horizontal platforms suchas OpenAI and Anthropic and sector-specific vertical applications built on top of them. Fallingdevelopment costs at the application layer are fueling record deal volume even as funding remainsconcentrated at the top. This report breaks down the split between platform and applicationlayers, recent exit activity, and shifting valuation dynamicsAI and machine learning (ML) startups raised a record-breaking $73.6 billion across 1,603 dealsin Q1 2025, marking the highest quarterly total on record by deal value. Deal counts were at theirhighest level since Q1 2024. Horizontal AI platforms captured the lion’s share, securing just under$50 billion across 425 deals, accounting for nearly 70% of total capital raised during the quarter.Horizontal platforms are general-purpose tools, such as large language models and infrastructuresthat support a wide range of industriesIn contrast, vertical application startups led in deal volume, notching 1,022 transactions, roughly60% of all deals, with an aggregate deal value of $19.2 billion. Vertical applications are tailoredto specific sectors and built on top of horizontal platforms, which enable domain-specificperformance and integration.The stark contrast between capital raised and deal count reflects underlying cost dynamics.Vertical models can be developed with just $3 million to $10 million in funding, while horizontallarge language model (LLM) development can cost hundreds of millions of dollars at the highend. This cost differential has lowered the barrier to entry at the application layer, where most AIstartups are building on existing horizontal platforms. Another driving force behind the surge invertical deals is the opportunity for startups to develop specific solutions with commercial value tolarge enterprises across industries and sectors. This contrast between horizontal and vertical platforms extends across a trailing 12-month period.Horizontal platforms attracted $107.3 billion in deal value across 1,673 transactions, nearly doublethe approximately $58 billion raised by vertical applications. However, verticals recorded 3,961deals over the same period, more than twice the deal volume of horizontal platforms, underscoringa persistent split between capital intensity and deal activity.Venture-growth deals generated the highest deal value in Q1 2025 at $49.4 billion, late-stage(series C and D) VC rounds reached $14.1 billion, and early-stage (series A and B) startups raised$7.3 billion. This distribution of capital by VC lifecycle stage was consistent across sub-verticalssuch as semiconductors, autonomous machines, and vertical applications.VC exit activity in Q1 2025 totaled 146 deals, a slight decline from 154 in the prior quarter. Despitethe dip in volume, exit value reached $27.4 billion, marking the strongest quarterly total sinceQ4 2021. Public listings accounted for $21.5 billion of that figure, while acquisitions and buyoutscontributed $2.7 billion and $3.2 billion, respectively. Exit counts remained skewed towardacquisitions at 105, with buyouts at 30 and 11 public listings. The concentration of exit value in IPOshighlights the impact of a few outsize listings driving quarterly exit value. The majority of activitywas propelled by AI hyperscaler CoreWeave, which went public on March 28 with a post-moneyvaluation of $14.5 billion. The second-largest IPO was Blocks Group, a China-based B2C company,which debuted on January 10 with a post-money valuation of $1.9 billion.Many acquisitions were driven by well-capitalized VC-backed companies pursuing bolt-onacquisitions to accelerate growth or expand capabilities. Acqui-hires remain common, particularlyin AI and software, where talent scarcity can justify premium valuations. Median VC pre-money valuations climbed to $12.7 million at the pre-seed/seed stage, $57 millionat the early stage, and $53.2 million at the late stage. Valuations peaked in the venture-growthstage at $325 million, reflecting sustained investor appetite for mature, scaled startups. Late-stageVC led all stages in deal count with 156, narrowly surpassing pre-seed/seed at 152. Overall, medianVC pre-money valuations climbed from $24.4 milli