Recently, Zoho cofounder Sridhar Vembu raised concerns over the economics of building software in an AI-driven market. In a post on X, Vembu said server prices have risen 4-5X in a year, while AI usage and token expenditure are pushing up operating costs.
Zoho, however, plans to respond by cutting its marketing budget, freezing headcount for 18 months and pausing its usual salary hikes this year.
Vembu’s concern is that while AI has made software easier to build, running a software business is getting more expensive. At the same time, AI coding tools and agents are challenging traditional subscription and per-seat pricing models.
Another concern, as per Vembu, is that AI-driven productivity gains in software development currently stand at 25% to 50% when measured across the entire product development process. While developers can build prototypes faster, testing, verification and other stages of taking a product to market continue to require human involvement.
When AI Costs Start To Bite
As AI eats into SaaS revenue, it is also making software more expensive to run. Every AI query requires computing power, adding to operating costs. At the same time, growing demand for AI infrastructure is driving up hardware prices, making it more expensive for companies to run their software.
According to ICONIQ’s 2026 data, AI products have an average gross margin of around 52%, compared with the 80% benchmark for traditional SaaS. For established SaaS companies adding AI features, gross margins can fall by 12-17 percentage points, depending on how efficiently they manage AI computing costs.
The pressure is also forcing software companies to reconsider how much profitability they are willing to sacrifice to build AI products.
On October 15, 2026, SaaS giant Figma’s CFO, Praveer Melwani, said the design software company might not return to its previous gross margins of over 90% as it invests in AI features. Its margins were in the mid-80% range, with the company prioritising long-term growth and competitiveness over an immediate return to earlier profitability levels, according to Reuters.
Meanwhile, rather than treating every decline as a problem to be eliminated, some software companies are accepting higher costs, expecting AI features will attract more users, strengthen their products and create future revenue opportunities. Similarly, some are adding more capabilities without proportionately increasing subscription prices.
Krish Ramineni, cofounder and CEO of Fireflies, said that the company has not raised its subscription pricing in the five to six years since launching its meeting note-taking product. It is now expanding beyond meeting notes into tasks such as summarising emails, responding to workplace messages and preparing users for meetings, while aiming to include more tools within a single subscription.
The Push To Rein In AI Spending
The pressure to manage AI expenditure is also showing up in the financial results of Indian companies. Groww, for example, reported a 28% year-on-year increase in total expenses to ₹556 Cr in Q2 FY27, even as revenue from operations grew 47% to ₹1,501 Cr.
In its shareholder letter, the company identified inflation in compute costs due to AI adoption as one factor behind higher serving costs. Cofounder and CFO Ishan Bansal said AI-related costs began rising as the company increased its focus on AI, alongside higher compute usage on trading days and a one-off operational expense. Groww expects inflation-driven costs to remain part of its operational baseline going forward.
This is prompting companies to re-evaluate their technology investments. Instead of relying entirely on large, general-purpose models, some are fine-tuning smaller models for specific tasks, optimising infrastructure and using more expensive models only when necessary.
Perfios, a financial technology company serving banks and other financial institutions, is taking this approach to manage the cost of deploying AI.
Sumit Nigam, chief technology officer at Perfios, said the company is fine-tuning smaller language models and optimising model and GPU architecture to make specific workloads cheaper to run. According to him, GPU utilisation costs for some use cases have become significantly lower than the cost of relying on external models such as Google’s Gemini or Anthropic’s Claude.
Companies are also looking to improve visibility into how AI expenditure builds up across teams, applications and workflows. On October 12, US-based software company Freshworks signed an agreement to acquire Torii, a software management platform. The deal is intended to extend Freshworks’ Freshservice product into software management and AI spend governance.
Torii tracks software applications, licences and AI consumption and attributes spending to the people and teams driving it. The platform is designed to help IT and finance teams understand what is being used, identify cost spikes, set limits and forecast expenditure.
Perfios, meanwhile, is combining cost optimisation with closer scrutiny of business returns. Sridhar Narayan, chief business officer at Perfios, said the company tracks profitability at both the product and customer levels. The focus is on optimising costs and directing spending towards areas where it can deliver better business outcomes, rather than cutting expenditure indiscriminately.
These approaches point to two parts of AI cost management: making usage visible and controllable, and reducing the cost of delivering each AI-powered task.
While AI is making software cheaper to build, the real value lies in data, workflows, integrations and the ability to run business processes reliably at scale.
Adapting To A New AI Reality
As Indian technology companies head into 2027, they face a two-sided challenge: customers are demanding more productivity from AI, while companies must figure out how to earn sustainable revenues.
For IT services firms, this is changing how clients pay for work. A Reuters report found that AI is pushing clients to link payments to productivity gains rather than the hours billed by IT services companies, forcing the industry to rethink its pricing models.
For SaaS companies, the challenge is different. As AI makes software easier and cheaper to build, products with limited differentiation risk becoming easier to replace.
According to Prashant Singh, the cofounder and COO of LeadSquared, “If AI makes software cheaper to build, then software that was only ever an interface over a database becomes harder to defend. That pressure is real.”
But Singh believes the threat is not uniform across the industry. “What is being commoditised is the ability to build an application, not the ability to run a business process reliably at scale,” he said. Proprietary data, integrations and industry-specific workflows will continue to help companies retain their competitive edge.
Going into 2027, the challenge for Indian tech companies will be to adapt to AI-driven changes in pricing and customer expectations while protecting their margins.
[Edited by: Shishir Parasher]
[Creatives by: Abhyam Gusai]
The post Rising AI’s Costs Put SaaS Companies Between A Rock And A Hard Place appeared first on Inc42 Media.