Morgan Stanley Report: Agentic AI Will Not Commoditize Software; Competitive Advantages Shift to Data and Control

Written by: Rita

13.2% versus 10.9%. Gartner projects a compound annual growth rate of 13.2% for application software spending from 2025 to 2030, and 10.9% after accounting for AI's cross-functional displacement. The 2.3 percentage point difference reflects how AI has reduced coding costs, but not the costs of production deployment, integration, governance, and maintenance.

In its report dated October 20, 2026, Bank of America concluded that agentic AI will not turn software into a commodity. The moat is shifting from interfaces and standardized implementations to trusted data, industry expertise, AI control planes, and partner networks, with value moving from the functional layer to the execution and governance layers.

The decline in coding costs has not eliminated the complexity of production deployment; seat-based pricing is under pressure, and spending growth has been slowed by only 2.3 percentage points. Whether software vendors can elevate AI revenue from low single digits to double digits is the next variable to be tested.

In early October, a tool called REA (Reverse Engineer Anything) connected AI coding agents to Ghidra, Hopper, and IDA, deconstructing installed software into simplified blueprints and enabling the AI to replicate their functionality. ArtCraft followed a similar approach to release seven open-source applications mimicking Photoshop, Illustrator, Premiere Pro, and Lightroom, with interfaces and tool layouts closely resembling Adobe’s products.

Bank of America believes these demos amplify the narrative of "software for everyone." Replicating menus, panels, and basic functions is far easier than matching the depth, reliability, performance, and file compatibility of mature commercial software. ArtCraft itself labels multiple applications as early alpha or under development. A similar case exists in the computer-aided design (CAD) field: the open-source SolidCraft offers parametric modeling workflows, but simulation, advanced surfaces, computer-aided manufacturing (CAM), and 2D drawings remain incomplete.

Agent-based coding falls into four categories: pure prompt-based generation, conversational development, AI-augmented coding, and specification-driven development. These four types of tools transform natural language into a software creation interface, but enterprise-level deployment still relies on experienced engineers, product judgment, and domain expertise. AI agents excel at replicating, adapting, and iterating on known designs, but still struggle to generate truly novel concepts.

The moat of user interface and user experience (UI/UX) is weakening. Agent-based AI interacts with applications primarily through APIs, allowing agents to operate clunky legacy interfaces on behalf of users, reducing the importance of interface familiarity and design quality. Bank of America is shifting its defensive focus to trusted business data, embedded industry knowledge, AI control planes, and partner networks.

The data moat depends on how the data is generated. Customer-imported data is portable, but approval histories, anomalies, escalation paths, human overrides, and final outcomes produced during workflow execution are much harder to replicate. ServiceNow highlights 85 billion production workflows and 7 trillion transactions as AI business context; SAP generates similar context within its core financial and supply chain processes. The quality of business context matters more than the volume of raw data.

The AI control plane is becoming a new moat. SAP requires AI agents built on third-party models to access systems through its governed platform; ServiceNow’s AI Control Tower discovers, monitors, and governs AI agents across enterprises, capable of shutting down unauthorized agents. Bank of America expects no single company to control all agents; instead, layered control towers will coexist beneath enterprise-level platforms: Microsoft manages productivity tools, SAP manages enterprise resource planning (ERP), ServiceNow manages IT and workflows, and CPaaS providers such as Twilio and Sinch manage customer communications.

System integrators and value-added distributor networks are underappreciated. Enterprise deployments rarely end after software sales—integration, migration, process redesign, localization, training, and ongoing optimization all require partners. Agentive AI can automate routine configuration and coding, while introducing new demands around data preparation, governance, control, auditing, and cross-system coordination.

The seat-based pricing model is facing direct pressure from agent-based AI. Agents perform more tasks, while user counts remain stagnant, capping the base for per-user pricing. Bank of America observes software vendors shifting toward consumption-based and outcome-based pricing, charging by tokens, actions, conversations, or completed outcomes. Pricing is still in experimentation, and customer resistance may slow the transition away from subscription models.

SAP’s monetization model combines subscription and consumption-based pricing, with Joule Copilot and Joule Studio supporting approximately 40 domain agents, enabling customers to combine and customize workflows. SAP disclosed that over two-thirds of its cloud deals include AI, and 90% of its top 50 transactions include Business Data Cloud (BDC) or AI. Less than 60% of cloud customers have adopted AI, and $2 billion in BDC contracts have been signed. Bank of America has placed SAP in a defensive stance with a target price of €226 and a Buy rating.

CAD software is more sensitive to seat-based pricing. The moats of Autodesk, Dassault Systèmes, and Nemetschek lie in their accurate digital records spanning design, testing, manufacturing, and product lifecycle—AI-generated geometries must still satisfy engineering constraints, undergo simulation, and integrate with downstream processes. All three companies continue to rely primarily on seat-based pricing, and agent-based AI’s pressure on seat counts directly impacts their revenue models. Bank of America assigns Autodesk a Buy rating with a target price of $300, Dassault Systèmes a Neutral rating with a target price of €25.6, and Nemetschek a Buy rating with a target price of €93.

Gartner projects a compound annual growth rate of 13.2% for application software spending from 2025 to 2030. When accounting for the erosion caused by cross-functional agent applications, the growth rate for traditional functional markets declines to 10.9%, a difference of 2.3 percentage points. Cumulative spending of approximately $116 billion will shift from traditional functional categories to cross-functional agent applications between 2026 and 2030, with annual erosion increasing from $400 million in 2026 to $2 billion in 2027, $11 billion in 2028, $25 billion in 2029, and $79 billion in 2030.

Bank of America believes these 2.3 percentage points reflect a shift in spending between categories, not an equal loss in total spending. Platforms that can deliver both core applications and cross-enterprise orchestration can recapture spending across categories, while risk is concentrated among narrow vendors focused on single functions. Growth rates for Customer Relationship Management (CRM) declined from 14% to 11%, ERP from 12% to 9.5%, Supply Chain Management (SCM) from 13.5% to 12.5%, and content services from 10% to 6%. Cross-functional agent applications are expected to account for over 10% of software spending, with value concentrated in vendors controlling orchestration, enterprise data foundations, and workflow execution.

AI investment decisions are more often driven by business units. According to a survey by IDC, 73% of AI investment decisions are jointly owned by business and IT or led by the business. Some AI spending may come from business unit productivity budgets, reducing reliance on centralized IT software budgets.

Bank of America has positioned SAP as a defensive play. SAP’s ERP system covers finance, human resources, supply chain, procurement, and customer experience; its deep data models, embedded controls, and regulatory workflows create high switching costs that generic alternatives struggle to match. While generative AI may displace SAP in edge cases, regulatory requirements, mission-critical coverage, and switching costs tip the balance in SAP’s favor.

The top of the travel IT distribution funnel will be reshaped by individual agents. Bank of America rates Amadeus as "Underperform" with a target price of €53, rates Sabre as "Buy" with a target price of $52.50, and rates HBX as "Buy" with a target price of €12. Individual agents are unlikely to become a mainstream booking channel due to supplier control over inventory, limited user trust, and ongoing privacy constraints—but agents can still alter how travelers discover and compare options without executing transactions.

Agent-driven communications in CPaaS will drive long-term volume growth. Bank of America assigns Twilio a Buy rating with a $320 price target, and Bandwidth a Neutral rating with a $60 price target. Data platforms and observability software benefit from increased demand for governed data and more complex workloads driven by agents. Bank of America assigns Snowflake a Buy rating with a $470 price target, MongoDB a Buy rating with a $540 price target, Datadog a Buy rating with a $305 price target, Dynatrace a Buy rating with a $65 price target, and Elastic a Neutral rating with a $108 price target.

The distinction between development tools and robotic process automation (RPA) has become more pronounced. JFrog has a Buy rating with a target price of $110, GitLab has a Neutral rating with a target price of $54; AI is transforming the software development lifecycle, increasing complexity and security demands, thereby elevating the strategic importance of development tools. UiPath has an Underperform rating with a target price of $15; probabilistic AI agents are competing with deterministic RPA for future workloads, and improvements in AI capabilities or reductions in cost could erode RPA’s value proposition.

Agent-based AI drives value toward control and data; whether software vendors can raise AI revenue from low single digits to double digits is the next variable to be validated.

Disclaimer

This article is a compilation and interpretation by Chaoxiang Research of a third-party brokerage research report (Bank of America, October 20, 2026), combined with publicly available market information. The ratings, price targets, earnings forecasts, and related judgments cited herein reflect the views of the brokerage’s analysts and represent the position of their respective institution, not the views of Chaoxiang Research, nor do they constitute any investment advice.

The market carries risks; make decisions independently. This article should not be used as a basis for buying or selling any securities.