OPINION | SaaS isnt dying, it's getting a stronger foundation

For more than a decade, organisations have adopted best-of-breed applications at unprecedented speed, under the guise of 'agile' transformation, which resulted in flexibility in the moment but at the cost of long-term fragmentation. This is what we call integration tax. What CIOs had once dismissed as a manageable inconvenience—this includes application sprawl, disconnected data, and integration debt—has now metastasized into a strategic liability. This fragmentation has impacted companies across sizes and industries. CRM systems fail to integrate with financial data, HR databases don’t connect to operational systems, and isolated platforms lead to an architectural setup where AI models generate false information, contradict workflows, or deliver conflicting insights, resulting in chaos.

This year, enterprises have realised that AI demands organised and integrated database to produce dependable results, ones that can provide them the ROI on the expensive technology. Platform consistency, once just a buying consideration, has now become essential for use of AI, governance, and security.

Therefore, the takeaway from 2025 is clear: the age of haphazard SaaS tool usage is being replaced by a phased consolidation and architectural discipline owing to the advent of AI.

2) Bigger LLM might not be the better or greener LLM for your enterprise

As generative AI captured global attention, a compelling narrative emerged, that companies using the largest foundation models would achieve a significant competitive edge, the more the parameters meant greater the capability and higher the value. However, the ground reality suggested otherwise. Gartner predicts that by 2027, over half of enterprise AI deployments will rely primarily on small or medium-sized models driven by efficiency gains, privacy requirements, and the need for context-specific reasoning.

The immense computational power required to train and run massive LLMs translates into a significant technological carbon footprint. Sustainable AI demands that enterprises prioritise model size, which is what led us to 'right size' the model and architecture to reduce power consumption dramatically. Choosing an appropriately sized, highly efficient model is not just an efficiency gain, it is also a critical environmental responsibility.

Additionally, enterprise workflows rarely require internet-scale inference. They require context, deep understanding of process logic, data definitions, compliance requirements, roles, permissions, and organisational nuance. Zoho's LLM models—1.3B, 3B, and 7B parameters—trained on business use cases and deployed in dedicated cloud or on-premise environments, can run efficiently, on much lower compute cost and give results for work that actually matters.

3) For large enterprises, SaaS adoption fails without co-creation

Despite sustained growth in SaaS spending, adoption challenges persist. The root cause is simple, enterprises treat procurement as a license transaction, not an opportunity for co-created capability.

The last mile of innovation cannot be purchased off-the-shelf, it requires collaboration. The next phase of SaaS adoption demands SaaS players who act as strategic partners with deep engineering and workflow expertise, moving beyond simple software provision.

Enterprises that treat SaaS as a joint co-creation effort will build more resilient technology ecosystems and achieve significantly higher ROI.

Gearing for SaaS X.O: The pillars that will define the next wave of SaaS

The SaaS industry is fundamentally changing: exiting a decade defined by rapid expansion and entering a new era grounded in discipline, interoperability, and licence optimisation. Some structural shifts are already underway that will clearly separate the market leaders from the rest of the pack over the next few years.

In addition to building a strong foundation for adding contextual AI layer, businesses will need to relook at the way the data is stored, processed, and used. Recent geopolitical events have shown the criticality of data privacy and technological sovereignty. In fact, a recent Gartner study underscores this urgency: 70% of enterprises adopting generative AI will cite sustainability and digital sovereignty as top criteria for selecting cloud AI services by 2027.

At the same time, enterprise expectations around value creation are shifting. The traditional model of evaluating software based on seat counts or feature breadth is becoming obsolete. CXOs are now looking at platforms that deliver outcomes:from reduced implementation cycles, heightened productivity, faster decision-making.

As platforms evolve from static “systems of record” into dynamic “systems of action,” agentic AI will shape operational flows in ways that interfaces alone never could. Platforms built on deeply integrated stacks, like ours, are uniquely positioned to support this transition, because cohesion and context across the tech stack strengthen both automation quality and data governance.

The next decade of SaaS will reward companies that build with intention, and keep business context at the epicentre. Fragmented portfolios and short-term engineering cycles are giving way to unified ecosystems, long-term R&D investments, and trustworthy AI architectures. The future belongs to platforms that operate cohesively, respect user data, and deliver measurable impact. The industry is moving from an era defined by acceleration to one defined by accountability, and in that shift lies meaningful opportunity for those committed to building responsibly for the long term.

(Praval Singh, Vice President, Zoho.)

Views are personal, and do not represent the stance of this publication.