Enterprise software grows 13%, but AI gains are wildly uneven

The global enterprise software market grew by 13% in the first half of 2026, up from 11.5% in 2024 and 12.1% in 2025, according to IDC research. But that headline figure conceals a sharp divergence: workflow automation and collaboration vendors are posting double-digit growth, while customer service and human capital management software are stagnating or shrinking. Atlassian grew 28%, ServiceNow 24%, and BILL Holdings 16% in core revenue, while Sprinklr grew just 6.8% and guided to roughly 1% next quarter; LivePerson’s revenue fell 12% year-on-year.

The uneven impact of artificial intelligence is not a question of whether AI is reshaping enterprise software—it clearly is—but of where and how that value is being captured. For executives, investors and technology buyers, the aggregate market number can be misleading. The 13% headline growth, while an improvement from 11.5% in 2024 and 12.1% in 2025, obscures a market in which some vendors are growing well above the average and others are contracting. Understanding category-level dispersion is now essential for decisions about procurement, pricing and strategy, because the commercial momentum of AI is highly uneven across software segments.

The dispersion behind the aggregate number

IDC analyst Eric Newmark argues that the overall market growth rate hides more than it reveals. “The aggregate number is lying to you,” he said, pointing to significant differences across software categories. The real signal, according to Newmark, lies in category-level dispersion.

Workflow-heavy and automation-focused applications are leading the expansion. Atlassian reported 28% growth, ServiceNow 24%, and BILL Holdings 16% in core revenue. Collaboration software is also strong: Monday.com grew 22%, and enterprise portals neared 17% growth in 2025. These categories share a common trait—they embed AI directly into the flow of work, automating tasks, routing decisions and reducing manual overhead. That integration makes AI a feature of the existing workflow rather than a separate add-on.

In contrast, customer service and human capital management software show clear stagnation. Sprinklr’s 6.8% growth and its guidance for approximately 1% next quarter stand in stark contrast to the double-digit leaders. LivePerson’s revenue declined 12% year-on-year. Workday’s growth held at 13.5%, which is in line with the 13% average for the HCM category but far below the momentum seen in workflow automation. The gap between the fastest and slowest categories is widening, even as the overall market accelerates. That divergence is the central fact behind the headline number.

Pricing shifts, not a lack of AI value

The slowdown in AI-exposed sectors such as customer service and HCM may not reflect a failure of AI to deliver value. Newmark suggests it may instead reflect a structural shift in how software is priced and sold. Usage-based or outcome-based pricing models are replacing traditional per-seat licensing in many AI-heavy applications. That transition can temporarily suppress revenue growth even when the underlying technology is creating real efficiency gains.

For example, a customer service platform that automates responses may reduce the number of human agent seats a company needs. If the vendor charges per seat, its revenue can decline even as its AI delivers more value. If it shifts to usage-based pricing, revenue becomes tied to outcomes or consumption, which may grow more slowly than seat count did in the past. The result is a period in which new AI revenue generation has not yet outpaced the decline of legacy licensing models.

This dynamic explains why some AI-exposed categories appear weak in revenue terms while still being strategically important. It also means that investors and buyers cannot simply read top-line growth as a proxy for AI success. A vendor with slower revenue growth may be undergoing a deliberate pricing model transition that positions it better for long-term AI monetisation. The shift is not uniform across all AI categories, but it is most pronounced where AI directly reduces the need for human seats or where outcomes can be measured and priced.

The execution gap in enterprise AI

Beyond pricing, a deeper execution gap is holding back AI value in many enterprises. Surya Gummadi, president of Americas at Cognizant, said that companies struggle to convert AI investment into outcomes because of legacy systems and fragmented data. A June 2026 Cognizant-Pearson study found that 63% of enterprises face a moderate-to-large gap between their AI ambitions and their actual capabilities, even though more than half are spending over $10 million annually on AI. That spending level indicates serious commitment, but it does not guarantee results. The gap between ambition and capability is not primarily a technology problem; it is an integration problem.

Gummadi argues that success requires “shaping technology around operational realities” and building an “enterprise harness” for AI integration. That means connecting AI models to the core workflows, data pipelines and decision processes that already exist inside an organisation. Without that integration layer, AI remains a set of isolated experiments rather than a driver of measurable business outcomes.

The categories that are growing fastest—workflow automation, collaboration, enterprise portals—tend to have AI embedded directly into the tasks users already perform. The categories that are lagging often require deeper changes to data architecture, employee roles or customer interaction models before AI can deliver value. The execution gap is therefore not uniform: it is widest where legacy processes and fragmented data are most entrenched. This explains why some vendors with strong AI capabilities still see slower revenue growth: their customers cannot yet absorb the technology into daily operations.

What leaders should watch

For international professionals, the lesson is that AI’s commercial value is not automatic. Leaders need to scrutinise vendor pricing models to understand whether AI features are bundled into existing licenses or metered separately, and how that affects total cost over time. They should also assess category-level growth, not just overall market trends, to anticipate where vendor pricing pressure may emerge. A vendor growing at 13.5% in HCM, where the category average is 13%, is essentially tracking the market rather than leading it. Leaders should compare individual vendor growth against the relevant category benchmark, not just the overall software market.

Key questions include:

• Is the vendor’s revenue growth driven by new AI capabilities or by legacy seat expansion?
• Does the pricing model align with how the enterprise actually consumes AI—per user, per outcome, or per usage?
• Can the AI be integrated into existing workflows without a costly data and systems overhaul?
• What is the vendor’s category-level growth trajectory compared with its peers?
• Does the vendor’s AI roadmap require the customer to replace legacy systems, or can it work alongside them?

“The future belongs not to those with the best models, but to those who can industrialise AI within core workflows.”

That statement, drawn from the research, captures the central challenge. Enterprise software growth will continue, but the gap between AI-native workflow platforms and legacy categories is likely to widen. Companies that can embed AI into operational realities—rather than simply deploy models—will capture the most value. For everyone else, the aggregate market number will remain a poor guide to where that value is actually being created. The 13% growth figure is real, but it is an average of very different trajectories. The winners are those who treat AI as an operational capability, not a standalone product.