AI is rewriting the software bill: what earnings say about SaaS disruption

Key takeaways

• AI threatens software built around simple tasks and seat counts, but can strengthen platforms tied to critical workflows and proprietary data.

• Recent earnings show monetisation matters: HubSpot is changing pricing, while ServiceNow, SAP and Atlassian still show strong AI-linked demand.

• Investors can track retention, pricing and workflow depth rather than treating the entire software sector as one disrupted trade.

Software-as-a-service (SaaS), cloud software sold by subscription, has spent much of 2026 under a simple fear: if artificial intelligence (AI) agents do more work, companies may need fewer human users and therefore fewer licences. But the risk is more nuanced than “AI kills software”. The real shift is in what customers are willing to pay for. Value is moving away from features and seat counts, towards outcomes, proprietary data, embedded workflows and transactions.

The seat is getting nervous

For years, SaaS had a convenient growth engine. A customer hired more people, bought more seats and paid more subscription fees. AI challenges that formula because fewer employees, supported by agents, may handle the same workload.

HubSpot is a useful case study. The customer-management platform grew second-quarter revenue 20%, yet slightly lowered its full-year outlook. Management said it was deliberately changing its products, pricing and sales model to make AI easier to adopt.

That can help adoption while delaying revenue. The old question was “how many seats?” The new one is increasingly “how much valuable work runs through the platform?”

Atlassian is preparing for that shift. Its new Flex model lets large customers commit to a fixed budget but move spending between seats, applications and AI usage. Salesforce is also reporting “agentic work units” alongside rapid Agentforce adoption. The billing unit itself is starting to evolve.

Workflow is the new moat

AI can recreate a feature surprisingly quickly. Recreating years of company data, integrations, permissions and workflows is harder.

That matters for ServiceNow, SAP, Workday and Salesforce. These companies sit inside important processes such as information technology, finance, human resources and customer management. Replacing them is less like swapping an app and more like rewiring an office building.

Recent earnings suggest these platforms still have leverage. ServiceNow grew subscription revenue by about a quarter and said its AI products crossed USD 1 billion in annual contract value. SAP reported 26% constant-currency growth in current cloud backlog. Workday continued to grow subscriptions while raising its margin outlook.

Adobe and Intuit offer similar evidence. Adobe said AI-first annual recurring revenue more than tripled and raised its full-year targets in June. Intuit raised its full-year revenue guidance after its May results.

None of this makes incumbents immune. It suggests that proprietary data and embedded workflows can make AI a reason to use the platform more, rather than remove it.

AI can enlarge the market too

Shopify and AppLovin sit slightly outside traditional SaaS, but they help define the boundaries.

Shopify earns from subscriptions, payments and merchant activity. Second-quarter revenue grew 34%, while AI-driven traffic and orders to merchants tripled. If AI helps shoppers discover and buy more products, Shopify can benefit even if the traditional software interface matters less.

AppLovin shows another risk. Its advertising engine depends heavily on AI model quality. Revenue still grew more than 50%, but results fell slightly short of expectations and progress on a new model slowed. When AI itself is the advantage, model execution becomes a business risk.

The distinction matters. AI can disrupt the interface while strengthening the underlying platform.

Where disruption can bite

The highest risk sits with software that solves narrow tasks, owns little unique data and is easy to replace. Warning signs include weaker customer growth, falling revenue per customer and heavier discounting.

A second risk is pricing. Moving from seats to usage or outcomes can improve long-term economics, but the transition may look messy. Investors can watch whether AI revenue grows without steadily weakening margins or cash generation.

The third is competition. AI makes it cheaper and faster to build software, so customers may have more alternatives. Incumbents need their data, distribution and integration advantages to remain valuable.

Investor playbook

• Separate systems running critical workflows from tools that mainly add convenient features.
• Track retention, revenue per customer and AI usage, not simply the number of AI launches.
• Watch pricing changes. Flexible models can expand adoption, but they must eventually support healthy economics.
• Diversify across business models rather than treating one software company as the entire AI disruption story.

The moat matters more than the model

AI is not cancelling SaaS. It is cancelling the assumption that adding seats automatically creates durable growth. This earnings season shows a more selective contest. Platforms with deep workflows, trusted data and strong distribution can use AI to become more useful. Simpler tools may have to defend why they exist at all.

The strongest signal is therefore not how often management says “AI”, a metric that remains conveniently unlimited. It is what customers do next: renew, expand, automate more work and pay for measurable outcomes. That is also the useful way to view Saxo’s software shortlist. These companies face the same technological wave, but very different economics. AI will not treat them equally, and neither should investors.