科技的未来：我们所知道的软件世界会终结吗？

to software,welay out our 5+yearviewon how Almight impactthesoftware stack andwhatthe industry,layerbylayer,will looklike beyond 2030 witha focus onlaaS/PaaS andSaaS.We believe that GenAlwill increasethe TAMof laaS/PaaSdriving solid growth at mark.moerdler@bernsteinsg.comRichard Nguyen +331 42 13 54 22richard.nguyen@bernsteinsg.comFiroz Valliji, CFA thehyperscalers and parts of PaaS(e.g.database).Within laaS,both GPU/ASiCs andCPUusagewill increase,especiallyasAgenticAlbecomesmoreprevalent.Wealsobelievethat PaaS usage will grow, but certain functionalities willchange, and new vendors takethelead.Within PaaS,database usage will growand on-prem database (e.g. Oracle) willcontinuemoving tothecloud, withmore specialized database likelytobe created.On thewillincrease as traditionally non-IT personnel could use these tools to create agents. +1917344 8316firoz.vallj@bernsteinsg.comShelly Tang, CFA +1917344 8342shelly.tang@bernsteinsg.com structural tailwind for these layers (as compared to parts of SaaS that are cloud-maturing)whichwebelievewill lastbeyond2030. are going to be far more than they are today. Al is going to allow more personalization,customizationand is goingto support automation of currentlymanual operations andchange certain business processes.Thismeans that Al must be deeply embedded intotheapplications andappswill alsobecomemanagementlayersforcomplex Alagents andprocesses. Whilethere are sub-marketsthat willbe disrupted /replacedby Gen-Al, we disagree withthebroad stroke bearthesisthat Aleats software.Webelievethat beneaththe Alfear,someof the recent slowing in certain segments was in fact driven by cloud saturation more thananything else Although not all SaaS vendors areas insulated from Al disruption, we believethatincumbents have anumberofadvantages and vendors thatare"quick"in innovatingandkeepingupwithcompetitionwillsurvive/prosper.Complexdeterministicprocessescreatesamoataroundanenterpriseapplicationasdoesdomainexpertiseanddeepsemanticknowledge.Webelieve that themost protected areas within enterprise softwareshouldbeERP,followedbyHcMandthenfull-suiteCRM,ifthevendorsinnovateandexecute. Finally,within the next 5+ years, we believe that Generative Al is not an end state or the finaliteration of Altechnology. Unlike the Cloud tech stack that has solidified, Gen Al is very likelygoing to be disrupted both by internal technology waves (e.g.reasoning, Agentic Al) andpotentiallycompletelynewAltechnologies,someofwhichwillaugmentGenerative Alandsome that might replace parts or all of the present day Generative Al tech stack.This, webelieve, is a really important point for anyone investing in what they believe is the tip of thespear in Al -the spear may be replaced by the gun. INVESTMENTIMPLICATIONS By 2030, we believe that overall IT spending, which in some cases will not come from traditional IT, will grow strongly with theincreasing adoption of Generative Al technology.Over the last 7O+years the speed of innovation has accelerated with eachtechnology cycle and Al notjust add to the rate of innovation but the rate of innovation adoption. Within Saas,twokey factors that determine disruption risk are speed ofcatch-up innovation and the complexity of functionality,especially around the deterministic processes.In addition, incumbents must leverage their domain expertise,semanticknowledge of the interrelationships between data, customer relationships.Those that are not innovating aggressively will bedisruptedasaremostpure-playon-premisevendors. WithinlaaS/PaaS,webelievethatAlworkloads willincreasetheTAMforlaaS/PaaSvendors as: To leverage Al, workloads are going to need to be in the Cloud and by 2030 a much larger portion of critical workloads will We seelaaS/PaaS vendors as beneficiaries of Al revolution.We see hyperscalers (MSFTand ORCL within our coverage), MDBand SNow (to some extent) as beneficiaries from incremental database usage driven by Al. Within SaaS,we see SAP and Oracle as best positioned and HubSpot as relatively unaffected by the Al headwinds investorsare concerned about.Workday should be fine.Al should be both a small headwind and a mostly offsetting small tailwind(Agentforce)from Al for Salesforce. What is the right time horizon?Whathappens tothe Hyperscalers (laaS/PaaS)overthenext5+years?Does Al negatively impact laaS/PaaS and the hyperscalers?Does Al increase the TAMfor the hyperscalers.10Does the Enterprise applications industryget disrupted by Al?....11WhataboutthebearconcernsThen why are so many Cloud companies'growth slowing?Does the technology get disrupted before it completely disrupts the software world?. site selection); hiring (can Ireplace my employees by autonomous agents); or even the software industry.While we would lovetopontificate on the global ramifications of Al (andto a small extentI will in this note)I willfocus on where wehave themostexpertiseand that is look athowAl willimpact the software/Cloud industry over the next 5+years. Al in general and