2025年Q3先进计算领域私募股权与风投趋势及投资策略报告

INDUSTRY RESEARCH Launch Report:Advanced Computing PE and VC trends and investment strategies Contents Executive summary3Introduction6Market scenario8Market risks9Market growth areas10Recommendations to stakeholders12Overview and scope14Total addressable market18Growth drivers20Pivots and catalysts24Leadership matrix27Segment overview34Semiconductor supply chain34Semiconductor chip design35Quantum computing35Datacenter infrastructure38Deal and exit activity42Advanced computingPE and VC ecosystem market map61Advanced computingPE investor map62Advanced computingVC investor map63Taxonomy64Glossary70Methodology82 Institutional Research Group Analysis Rudy TorrijosDirector, Industry Researchrudy.torrijos@pitchbook.com Data pbinstitutionalresearch@pitchbook.com Publishing Report designed byJenna O’MalleyandMegan Woodard Published on December 10, 2025 Executivesummary This report launches our coverage of the advanced computingvertical. Advanced computing comprises all end marketsthat enable artificial intelligence (AI) to scale. This includesboth semiconductor and datacenter markets across bothoperational and information technology (IT) vendors.Advanced computing is one of two verticals within PitchBook’sadvanced technologies research domain. Coverage of PEsoftware will be launched in 2026. We see hyperscaler capital expenditures as the single definingmetric for investors and vendors to gauge the growth andsustainability of the advanced computing ecosystem. Webelieve the current expected hyperscaler capital expenditurelevel for 2025 and 2026 is realistic and indicative ofconservative strategic-planning expectations. We view powergeneration as the primary gating factor to revenue growth foradvanced computing vendors through 2030. Macroeconomicand semiconductor cyclicality are unlikely to affect capitalexpenditure deployment as reported end-customer return oninvestment (ROI) remains significant. Current capital expenditures are primarily focused onincreasing the fidelity of generative foundation models (training and inference). Without this achievement, end-user productivity from generative AI (GenAI) models will bemeaningfully limited beyond 2027. Agentic AI and physical AIdevelopment will be significantly curtailed as well. •Elasticity of intelligence:Demand for processors, projectedto generate $2.8 trillion in cumulative revenue, is effectivelyuncapped. Intelligence scales with computing power and islimited only by problem complexity. in cumulative datacenter chip revenue through 2030,effectively decoupling the industry from traditionalcyclical downturns. •Power constraints:Power availability has replaced siliconas the primary constraint on growth, driving $1.2 trillion incumulative spending on power and cooling infrastructurethrough 2030 to address three-to-five-year gridinterconnection queues. The primary beneficiaries of hyperscaler investments withinthe advanced computing sector are graphics processors,network processors, server microprocessors, AI accelerators,leading-edge-node semiconductor supply chain vendors,power-generation equipment, thermal managementequipment, and neoclouds. We do not see quantum computingas a commercially viable technology through 2030. •Architectural efficiency:The industry-wide transition toMixture-of-experts (MoE) architectures has drasticallylowered inference costs, making “expert-level”intelligence economically viable and triggering the currentinfrastructure boom. •Thermal obsolescence:Legacy air-cooled datacenters(limited to around 30 kW per rack) are now technologicallyobsolete for AI clusters, necessitating a massiveretrofitting cycle to liquid-cooling technologies to preventthermal throttling. •Fidelity risk:Language model fidelity (LMF) is the singlemost critical performance metric, as a failure to achieveperfect accuracy will relegate AI to creative tasks and stallthe high-value agentic and physical AI markets. Key takeaways To assist investors and stakeholders in navigating thistransformational shift, we offer these key takeaways that willshape the advanced computing market over the next five years: •The 60% threshold:Surpassing the 60% score on theARC-AGI-2 benchmark (average human intelligence) is thecritical technical milestone required to validate the labor-substitution thesis and justify the massive capital outlay. •Supply chain fragility:Geopolitical concentration remainsan existential risk, as the entire advanced computing stackdepends on TSMC’s capacity, with supply chain sovereigntyrequiring at least a decade to achieve. •Capital expenditure magnitude:Hyperscaler capitalexpenditures are projected to reach a cumulative $6.4 trillionthrough 2030, representing the single largest deployment ofinvestment capital in human history. •Semiconductor supercycle:The semiconductor sectorhas entered a secular supercycle driving $3.7 trillion •Not dot-com:Unlike the speculative dot-com era, currenthyperscaler capital expenditures are grounded in