Emilio Lapiello of Arthur D. Little explains how AI is reshaping software economics, pricing models, quality control, and the future of investable SaaS businesses.
The rapid rise of artificial intelligence is reshaping the software industry in ways that go far beyond product innovation. Traditional SaaS economics, built on fixed costs and recurring subscription revenue, are now being challenged by AI’s transaction-based cost structure, growing pricing pressure, and the continued need for human oversight in quality-sensitive environments.
Emilio Lapiello, Partner and Head of Digital & AI Solutions, Americas at Arthur D. Little, explores how AI is disrupting long-standing software business models, why hybrid pricing structures are likely to emerge, and what will define the most resilient and investable companies in this next phase of transformation.
Interview Excerpts
How is AI changing the core economics that made the SaaS model highly scalable and profitable?
AI is changing the core economics that made the SaaS model highly scalable and profitable because the AI cost structure is transaction based. The SaaS model was built on fixed costs and recurring revenue. AI changes that because of variable costs. AI revenue models will have to follow suit, shifting toward usage-based pricing, outcome-based contracts, or entirely new revenue streams.
Do you see transaction-based AI pricing replacing seat-based SaaS subscriptions, or are hybrid models more likely to emerge? Hybrid models are likely to emerge: subscription for baseline access, plus consumption-based pricing tied to actual usage. But at current inference cost levels, prices will likely need to rise to make the business viable, and rising prices will push some users out of the market. This pricing pressure could also trigger a few important trends: First, AI companies may need to tap into alternative revenue streams, including advertising. AI tools are gathering an extraordinary amount of behavioral data, not just what users click, but how they think, reason, and work. That data is a new asset class, and the temptation to monetise it will grow as margin pressure increases. Second, enterprise AI subscriptions create a window into how a company actually operates: its workflows, decision patterns, and internal processes. This data is enormously valuable, and raises serious questions about data governance and competitive sensitivity that most companies haven’t fully grappled with yet. Third, as prices rise, we’ll likely see a push toward smaller models hosted in-house. This mirrors what happened with cloud computing: first, everyone moved to the cloud, then many brought workloads back on-premise for cost and control reasons. The same dynamic could play out with AI, especially as open-source and smaller specialised models become more capable.
How should software companies manage the tension between AI-driven efficiency and the continued need for human oversight in quality-sensitive domains?
AI agents are delivering real efficiency gains, but there is likely a widespread underestimation of the quality gap in their outputs. The productivity numbers look impressive: fewer people, faster turnaround, lower cost. But in domains where accuracy, compliance, and nuance matter, AI agents still often produce outputs that require significant human review and correction. AI increases the volume of output while potentially decreasing the average quality per unit, which means the total burden of quality assurance goes up, and the overall productivity might still not see any improvements. For software companies specifically, the development pressure is shifting from code creation to validation and testing. The bottleneck is no longer “can we build it?”, it’s becoming “can we verify it’s correct?” Companies managing this tension successfully will be those that position AI as a productivity multiplier for their human workforce.
“The value of AI today is too focused on accelerating throughput, rather than on guaranteeing quality. I think successful companies will have to push on quality as a differentiator and rebalance where they focus their investment.”
What advantages do established SaaS players retain as AI-native challengers begin to disrupt the market?
SaaS companies have been slow to adopt AI, but they still might hold significant advantages over pure-play AI competitors. They have a deep understanding of customer needs, existing infrastructure, established customer relationships, and access to vast amounts of proprietary data that can be used to train and fine-tune models. They also carry something harder to replicate: customer trust and brand recognition. The strategic question is whether they can translate these advantages. A promising path is the “certified AI” approach: offering AI agents that are domain-specific, auditable, and integrated into existing workflows. In regulated industries, especially, customers could pay a premium for AI they can trust and explain to their compliance teams. There’s also a potential cost advantage hiding in plain sight: as AI agents increasingly replace direct human interaction with applications, incumbents could significantly reduce their investment in traditional user interface and user experience development, which might boost their profitability.”
In your view, what characteristics will define the most resilient and investable software companies in this new AI-driven phase of disruption?
The most resilient and investable software companies will be those that can demonstrate clear economic value. In a world where every AI interaction carries a real cost, the margin equation that separates winners from losers comes down to whether AI can deliver high value while keeping inference costs manageable. I suspect we’ll see specialised AI companies emerge as the first success stories before larger, general-purpose platforms figure out how to monetise effectively. Specialisation allows companies to train smaller, more efficient models on domain-specific data, which directly attacks the cost problem. It also makes it easier to demonstrate measurable, differentiated benefits to customers and to price accordingly.