AI Competition and the Software Investment Narrative

From an investor’s perspective, AI (Artificial Intelligence) has introduced uncertainty into enterprise software by inserting a powerful new layer between users and traditional applications. The core concern is not that software disappears, but that AI agents increasingly become the primary system of engagement, orchestrating work across multiple applications and relegating incumbent SaaS (Software as a Service) vendors to commoditized systems of record with reduced pricing power. This fear, combined with several years of post-Covid growth deceleration, has driven broad multiple compression across application software, even though there is limited empirical evidence of revenue disruption to date. 

At the same time, a growing body of analysis argues that the market is overestimating this risk both in scope and timing. AI monetization remains nascent, enterprise adoption is constrained by governance and reliability requirements, and incumbent vendors retain durable advantages in data ownership, workflow depth, compliance, and trust. The emerging debate is therefore less about “AI versus software” and more about where value ultimately accrues in a hybrid, agent-driven architecture.

Bear Case - Disintermediation and Value Leakage: The bear case centers on the risk that AI agents become the dominant interface for enterprise work. As tools from large model providers mature, they increasingly coordinate tasks across CRM, ERP, IT service management, and analytics platforms. In this scenario, software vendors risk losing control of the user relationship and becoming back-end data stores, which could compress differentiation and weaken pricing power over time.

The lack of visible AI-driven growth acceleration in public software companies’ results amplifies this concern. Despite years of AI investment and excitement, aggregate application software growth continues to decelerate, and AI-related revenue remains small relative to total revenue. Investors reasonably question whether the economic rents from AI are flowing instead to infrastructure providers, hardware vendors, and the model layer itself.

There is also a structural seat-risk dynamic. AI-driven productivity gains are enabling businesses to reduce headcount in areas such as customer support and internal IT, which challenges traditional per-seat pricing models. While spending has not yet declined materially, bears argue that contract renegotiations and outcome-based pricing could pressure long-term sales growth and margins.

Finally, large AI model providers face strong incentives to move beyond tooling and into higher-value enterprise workflows. With ambitious revenue targets and rapidly improving capabilities, these providers may increasingly compete directly with established application vendors. Even if outright replacement is limited, the mere possibility of such competition justifies a lower terminal value for many SaaS businesses in a bear scenario.

Bull Case - AI as a Total Addressable Market Expander and Incumbent Reinforcer: The bull case argues that AI does not eliminate the need for enterprise software; instead, it amplifies its value. SaaS vendors do more than sell code, they deliver operational reliability, security, compliance, integrations, and accountability at scale. These attributes are difficult and costly for enterprises to replicate internally, even with advanced AI models.

Rather than displacing SaaS, bulls believe AI agents are more likely to increase engagement with existing systems of record. In practice, agents typically query, update, and act through incumbent platforms. As a result, value can be captured through higher usage, premium AI features, and expanded scope, even if the user interface shifts away from traditional dashboards.

Importantly, early customer anecdotes weaken the most extreme bear narratives. Many enterprises are reducing headcount while increasing spending with core application vendors because AI-enabled modules deliver incremental value. In these cases, AI reallocates labor budgets toward software rather than eliminating the software layer altogether.

AI also introduces new complexity around governance, identity, security, and data control. As autonomous agents proliferate, enterprises require tighter oversight, auditability, and guardrails. This dynamic structurally favors large, established platforms and security vendors that already sit at the center of enterprise workflows.

Bottom Line: The market is currently pricing a negative outcome well ahead of observable fundamentals. While AI clearly introduces competitive risk, particularly for undifferentiated, seat-based SaaS, we believe the most likely medium-term outcome is coexistence rather than displacement. AI expands the opportunity set, reshapes monetization models, and redistributes value unevenly across the stack, but it does not eliminate the need for enterprise software. The key investment question is not whether SaaS survives, but which business models and platforms are positioned to capture value in an agent-driven world.

Which Software Vendors Are Best Positioned to Adapt? Software business models best positioned to adapt to AI-driven disruption are those that monetize control, context, and outcomes, rather than simple user access. The most resilient platforms function as systems of record with deep workflow embeddedness, where the software serves as the authoritative source of truth for mission-critical data and processes. In these models, the user interface is not the primary value driver; instead, value resides in reliability, auditability, compliance, and integration into core operations. As AI agents proliferate, they must operate through these trusted systems rather than replace them, reinforcing switching costs and sustaining pricing power even as interaction patterns evolve.

Closely related are vendors that already employ outcome-based or usage-based monetization models, which align naturally with AI-driven productivity gains. As automation increases and headcount requirements decline, software priced on delivered value, transaction volume, or workflow intensity benefits from higher activity rather than suffering from seat compression. In this framework, AI acts as a demand multiplier: more automated decisions, interactions, and events translate directly into higher software usage and revenue, insulating these models from traditional per-seat pricing pressure.

Data platforms and enterprise context providers are also structurally advantaged. AI systems are only as effective as the data they can access, govern, and interpret, making platforms that aggregate, normalize, and permission enterprise data indispensable. As agentic workflows scale, demand rises not only for raw data access but also for governance, lineage, observability, and performance. These platforms increasingly evolve from passive repositories into active AI-enablement layers, capturing incremental consumption and becoming more central as AI workloads grow.

Another category with strong adaptive positioning includes security, identity, and governance platforms. AI agents introduce new operational and liability risks, as autonomous systems require authentication, authorization, monitoring, and audit trails. Each additional agent expands the enterprise attack surface and compliance burden, driving incremental demand for software that governs access and behavior rather than executing tasks directly. In this sense, AI growth structurally expands the total addressable market for security and control layers, making spend in this category durable and often non-discretionary.

Vertical SaaS platforms with deep domain specialization are similarly well positioned. These businesses embed industry-specific workflows, regulatory logic, and operational nuance that general-purpose AI models struggle to replicate reliably. In regulated or high-liability environments, enterprises prioritize accuracy and consistency over experimentation, favoring platforms that already encode domain expertise. AI enhances these systems by improving efficiency and insight, but it does not substitute for the underlying operational framework, reinforcing customer dependence rather than undermining it.

Finally, the most advantaged long-term models are those that control orchestration across systems, humans, and agents. These platforms coordinate work rather than perform isolated functions, giving them decision authority over how processes execute across the enterprise. As organizations seek to simplify technology stacks in an AI-heavy environment, orchestration layers gain strategic importance by reducing fragmentation and enforcing consistency. In an agent-driven future, value accrues not to tools that execute single tasks, but to platforms that manage, sequence, and govern work at scale.

These are the types of discussions and factors our research team contemplates with every investment decision. If you would like additional information on particular securities as it relates to this white paper, please reach out.

Disclosures:

Crawford Investment Counsel (“Crawford”) is an independent investment adviser registered under the Investment Advisers Act of 1940, as amended. Registration does not imply a certain level of skill or training. More information about Crawford, including our investment strategies, fees, and objectives, can be found in our Form ADV Part 2 and/or Form CRS, which is available upon request.

The opinions expressed are those of Crawford. The opinions referenced are as of the date of the commentary and are subject to change. Crawford reserves the right to modify its current investment strategies and techniques based on changing market dynamics or client needs. 

Material presented has been derived from sources considered to be reliable, but the accuracy and completeness cannot be guaranteed.

CRA-2602-4