Venture capital firm Accel, which has backed Flipkart, Swiggy and Freshworks, says generative artificial intelligence (AI) has fundamentally changed the metrics Indian investors have relied on for over a decade to assess and value startups, with the focus now shifting to whether AI companies can deliver durable margins rather than simply grow quickly.
During India's consumer-internet boom between 2011 and 2019, investors could use measures such as repeat purchases, customer-acquisition costs and contribution margins to gauge whether a business was achieving scale. For software-as-a-service (SaaS) companies, annual recurring revenue (ARR) and net revenue retention became central markers of product demand and efficient expansion.
AI does not make these metrics irrelevant; it changes when they become useful. the US-based investment firm believes that, at the earliest stage, investors must first assess the problem being solved and the economics of delivering the product before applying conventional growth metrics to evaluate a startup.
“The first thing we assess while evaluating AI pitches is whether the founder has deeply understood the customer or consumer, and whether that insight has been built into the business,” Shekhar Kirani, partner at Accel, said in an interview.
Accel announced its ninth India fund on Wednesday, with $550 million in commitments, as part of a $3.5-billion global fundraise. The firm has identified AI, consumer internet, financial technology, advanced manufacturing and deeptech as focus areas for the vehicle.
The old scorecard breaks
Accel’s approach reflects early evidence from enterprise adoption and recent global investment trends, where investor interest has risen sharply, but the path from pilot to a commercially viable product remains uncertain.
A report titled The GenAI Divide: State of AI in Business 2025, published by MIT’s Networked Agents and Decentralized AI (NANDA) initiative, found that about 95% of generative AI pilots at enterprises had failed to deliver a measurable impact on profit and loss, while only 5% reached production with meaningful value.
Generative AI can make it faster and cheaper for startup founders to build an initial product. But every customer interaction creates a recurring cost. Each time an AI tool reads a prompt, analyses a document or produces an answer, it must run the underlying model, a process known as inference.
AI providers generally charge according to “tokens”, or small units of text processed by a model, while startups must also pay for cloud infrastructure, data storage, security and, in some cases, human review.
The investment question is whether a startup can charge its customers enough for the AI service to cover those recurring costs and still retain healthy margins as usage rises.
Kirani said a second test is whether founders are using AI to create genuinely new products or simply placing models on top of existing workflows, a pattern investors often describe as “wrappers” around foundation models.
“The next question is whether the business can be built with positive unit economics after accounting for AI and token costs, and whether customers can be reached efficiently,” he said.
For enterprise AI, the key calculation is the value of work the software actually replaces. “The relevant margin pool is the gap between what a customer will pay and the labour cost the software removes,” Kirani said.
Where Accel is investing
Accel has built its India AI exposure through early positions, including via its Atoms programme that. The firm says 70% of its Atoms portfolio comprises founders building AI-focussed businesses, and that it has invested in more than 27 AI companies over the past two years. Its Atoms 4.0 programme offered selected companies investments of up to $1 million.
Among Accel’s recent AI bets in India, it co-led Arivihan's $4.17 million pre-Series A round after backing the AI tutoring startup through its Atoms programme. It also led Oolka's $14 million Series A round, backing the fintech company’s AI agents for consumer credit and financial decisions. In another consumer-fintech bet, Accel led Bachatt’s $12 million Series A round for its AI-led savings and wealth platform aimed at merchants and self-employed users.
Indian AI companies raised $1.07 billion in the first half of 2026, up 33% from $802 million a year earlier, and nearing the $1.6 billion raised through all of 2025, according to Venture Intelligence data. AI’s share of Indian venture funding rose to 12.3% in 2025 from about 4.5% in 2020, according to the India Deep Tech Alliance.
The test beyond the AI boom
The broader question for investors is whether AI can develop into a durable revenue category or repeat parts of the funding-and-hype cycle seen in Web3 (technologies like blockchain that decentralise data ownership and control on the internet) and crypto. India's Web3 startups raised about $1.4 billion in 2022, but funding fell 81% to about $270 million in 2023, according to Hashed Emergent’s India’s Web3 Landscape 2023 report. The number of deals also fell to 47 in 2023 from 62 a year earlier.
The correction also brought shutdowns. Separate Tracxn data from January 2023 showed that at least 13 Indian Web3 startups shut down in 2022, while the number of new Web3 startups founded fell to 95 in 2022 from 261 in 2021.
The comparison between AI and Web3 is not perfect, but it points to a familiar venture-capital risk. Early excitement can bring funding, product launches and pilots before it is clear whether customers will continue paying for the product, use it often enough to support recurring revenue, or give the company enough traction to raise its next round. For AI startups, the test is especially important because a product can attract initial users quickly, but may still struggle to convert experimentation into sustained contracts and profits.
“Venture investors are risk-takers, and it is acceptable to lose money on some ideas. But when an idea works, it needs to have the potential to become very large,” Kirani said, adding that the firm typically makes five or six early bets in a category before committing large sums to the companies that show the most promise.