分析师注：SaaS已死，SaS万岁

EMERGING TECH RESEARCHSaaS Is Dead, Long Live SaS Institutional Research Group Rudy TorrijosDirector, Industry Researchrudy.torrijos@pitchbook.com The enterprise AI super-cycle begins now Derek HernandezSenior Research Analyst,Enterprise SaaS andInfrastructure SaaS PitchBook is a Morningstar company providing the most comprehensive, mostaccurate, and hard-to-find data for professionals doing business in the private markets. Published on February 9, 2026 Key takeaways •The “SaaS-pocalypse” is a false prophecy:Incumbents are not standing still,they are becoming the largest AI companies and greatest consumers of tokensin the world. The market is pricing SaaS for obsolescence just as those vendors Contents Key takeawaysExecutive summaryEnterprises wake upInfrastructure precedes applicationProductivity to replacementStructural moats and the enemies at the gatesThe future, todayRecommendations to investors and vendorsPE and private creditVC and corporate VCPublic equitySaaS foundersIT managers and chiefinformation officers •The software TAM will merge with the labor TAM in the next 20-year enterpriseAI super-cycle:Outcome-based pricing shifts vendor revenue capture from anannual charge of around $1,200 per seat to an annual charge of about $10,000 per •ARC-AGI-2 scores now surpass 60% at $1 to $10 per task:60% is widely regardedas the labor-replacement milestone. Costs per task are declining incrementally •Hyperscaler AI capital expenditure is approaching $1 trillion annually:Amazon, Google, Meta, and Microsoft are building expert intelligence inferenceenvironments for SaaS incumbents to deploy agentic AI. Infrastructure investment •Fear is the greatest accelerator:Market turmoil is forcing the SaaS-to-SaStransition roughly 18 months ahead of when agentic AI infrastructure will be ready •Incumbents retain 10 structural moats:distribution, data context, switchingcosts, workflow integration, security, compliance & governance, AI as a revenuelayer, outcome pricing, vendor consolidation, investment scale, and deep business •Investors must reassess every software position:PE and credit investors shouldaudit portfolios for workflow-wrapper risk and restructure covenants with grossmargin triggers below 60%. VCs should avoid funding UI wrappers and pursue Executive summary The market is suffering from a fundamental misunderstanding of the AI technologyinvestment cycle. A wise man once said, “the market is always right.” The same is truetoday and the disruption represents a decisive opportunity for investors and vendors SaS is the realization thatthe end service is thevalue-added deliverable Software as a service (SaaS) is dead. Long live service as software (SaS). SaS is the realization that the end service is the value-added deliverable and thatsoftware is the vehicle by which it is implemented. The paradigm shifts fromenterprises renting productive tools on a subscription basis to value-based solutions We are witnessing a seismic transformation in advanced computing and enterprisesoftware, culminating in the simultaneous valuation implosion of public softwarecompany market caps and the explosion of capital expenditure on AI infrastructure Concurrently, Amazon, Google, Meta, and Microsoft reported dramatically increasedcapital investments for fiscal year 2026. Investors have become worried thathyperscaler capital expenditures on datacenters are growing too fast and will quicklyeat through available operating cash flow and into capital balances. We believe the The software TAM is lining up to merge with the much more expansive TAM for humanlabor, and the SaaS-pocalypse market turmoil was the starting gun to align softwarecompanies on this trajectory. For investors, AI-era winners will be those companies Enterprises wake up Enterprises are now fully engaged. For specific business processes, AI has achievedexpert, Ph.D.-level capabilities. Current AI intelligence benchmarks like ARC-AGI-2show scores are now surpassing 60%, with visibility to 70% at costs per task between $1 and $10.2Below $1 per task, these scores have approached 30%. 60% is widelyregarded as the labor-replacement milestone, especially as costs per task are We are witnessing the precise moment when enterprise AI transitions from AI pilot-program curiosity to a competitive workflow imperative. The simultaneous explosionin hyperscaler capital expenditures and the rapid ascent of these intelligence •Advanced Micro Devices reported that the number of large businesses deployingits high-performance server chips on-premises more than doubled in 2025, and thecompany exited the year with record server sell-through.3 •SAP reported that more than two-thirds of its Q4 cloud order entry includedits Business AI platform, with that figure rising to 90% for the company’s 50 •ServiceNow’s centralized governance platform exceeded targets by more than 4x,and the company noted that customers are expanding entitlements by 13x upon •Amazon, Googl