In stark contrast to the global trend of surging AI funding, the Indian AI startup ecosystem is facing a significant downturn, raising concerns among investors, industry leaders, and the government.  

happyllama

A recent Tracxn report, shared exclusively with AIM, reveals that total funding for AI startups plunged by 53% year-on-year. It decreased from $305.9 million in FY 2023-24 to $143.6 million in FY 2024-25, alongside a 44% drop in the number of funding rounds. 

Meanwhile, data from analytics firm Dealroom shows that global AI startups secured $110 billion in 2024, a 62% increase from the previous year, with US-based startups accounting for around 42% ($46.2 billion) of the total funding. Not to forget, OpenAI recently announced that it has raised $40 billion at a $300 billion post-money valuation.

“The initial AI funding boom was driven by hype, with many startups claiming AI capabilities without substantial technological depth and real AI/ML implementation,” Surabhi Sanyukta, VP of Investments, BlackSoil, told AIM in an exclusive email interaction. 

We at AIM  wrote last year that India was likely the new hub for GenAI startups. It fueled high hopes that these startups would create revolutionary products and models to put India on the global map. But fast-forward to today, and the landscape looks surprisingly similar—no game-changing innovations have surfaced.

Is the Government under pressure from global peers?

This lack of innovation debate was further amplified by a recent WhatsApp forward that compared the startup ecosystems of India and China. Commerce Minister Piyush Goyal cited this to make a scathing attack on startups for their lack of production innovation in India. 

The Minister took a dig at some of the thriving startups and criticised them for their overemphasis on instant convenience services while lagging in deep tech and high-impact industries. He criticised them for turning unemployed youth into cheap labour instead of promoting significant technological progress.

Goyal pointed out that while Chinese firms are leading advancements in semiconductors, robotics, electric vehicles, and battery technologies, many Indian startups focus on “fancy ice creams” and grocery delivery. This comment ignited a furore within the startup community even as he appealed for greater participation from the investor community.

Investors felt that the startups offering solutions were often tailored for a narrow set of use cases, lacking scalability or genuine impact. This disillusionment, combined with broader market corrections and tighter capital flows, has led to more selective and measured funding, Sanyukta explained. 

She said India largely remains a consumer of AI rather than building a creator economy.  For instance, the Ghibli image generation trend, which flooded the internet, is a classic case.

Even today, amid the government’s push for AI innovations, several startups focus on building applications on top of existing global models instead of developing foundational AI infrastructure. This limits their strategic value in a world where countries with decisive AI strategies are emerging as global leaders.

Romit Mehta, an investor at Lightspeed India, dismissed Goyal’s statements in a post on X and termed them as one of the most odious takes on the ecosystem in a while. He argued that blaming founders conveniently shifts the focus away from the government’s failure to foster a meaningful ecosystem for manufacturing and deep tech in India. 

Worse, he added, it undermines the grit of those building for and in India, especially in deep tech. 

What are the so-called GenAI startups up to?

However, several founders of GenAI startups whom AIM spoke with disagreed with the minister’s comment and explained how the ecosystem is making progress. 

Pranav Mistry, founder of TWO AI, told AIM that the company has expanded from building foundational models like SUTRA into a full-stack GenAI infrastructure provider—tailored for enterprise-grade security, localisation, and scalability. “We also launched India’s first reasoning model, SUTRA-R0, earlier this year. Our revenue has already crossed a $10M run rate this quarter, and we’re on track to reach our $20M (170 crore INR) target by the end of 2025,” he said.

Amid the tightening of funds, Mistry was hopeful to raise another $50 million to scale infrastructure. This is in line with the strong demand from its growing enterprise customer base and to expand its tech and GTM teams, especially in India. 

That said, China’s DeepSeek, which made waves in early January and brought down US chip stocks, is already gearing up for its next reasoning model, which is reportedly set to launch in the coming weeks. 

“The round will also support further R&D in LQMs, secure inference, and new agentic AI solutions, helping us deliver India-first AI globally,” Mistry added.

On the contrary, Harneet SN, founder of Rabbitt.ai, told AIM that application-based generative AI startups in India do not need to raise more funds to sustain themselves. He opines that AI is a very revenue-positive domain. The company is likely to hit a $10 million ARR.

The company pivoted from improving AI models, data annotation, and deploying open-source LLMs for specific use cases to developing full-fledged GenAI applications rather than just selling a single AI product. Harneet explained that companies now demand comprehensive AI solutions rather than just customising or deploying models.

On similar lines, Pratik Desai, founder of KissanAI, told AIM that since last year, their company has focused on building an agriculture knowledge platform with a multi-agent orchestration engine. “We are currently raising funds to scale globally and expand our team, as we are seeing strong global demand and significant organic inbound traction for our platform,” Desai said.

Besides, there are several AI startups in India exploring vertical SaaS. Nurix AI, the latest startup founded by Mukesh Bansal, secured $27.5 million in seed and Series A funding last September. The financing round was co-led by General Catalyst and Accel, with participation from Bansal’s venture studio, Meraki Labs. It specialises in building custom AI agents for enterprises, focusing on solutions such as AI-driven customer service with human oversight for call centers and outbound sales.

AIM also spoke to Madhav Krishna, founder of Vahan.ai, which is backed by Khola Ventures. He said that this year, his company will focus on voice agents for the hiring process, assisting with job discovery, matching, onboarding, and training. Krishna further explained that the already existing AI recruiter voice bot will be part of it.

“There’s been this buzz around vertical SaaS AI a lot in the US, and we are, in that sense, highly verticalised because we’re utilising it for a very specific use case in India,” he added.

Require More Investment?

However, in a previous interaction with AIM, Debarghya (Deedy) Das of Menlo Ventures said that India faces a considerable challenge in competing with Western labs. Although not entirely impossible, he admits, as an investor, he is unlikely to make such a bet. 

“If you look at AI in India today, the market size doesn’t quite justify the level of R&D investment needed to make it a success,” he said. Das views localised AI models, particularly those supporting Indian languages and voice applications, as a natural and achievable focus area for India’s AI ecosystem. 

Tracxn reports that customer service software, particularly in self-service and conversational AI, is the dominant business model among AI startups in India. Conversational AI platforms, including chatbots and virtual assistants, are playing a crucial role in this transformation by automating responses, handling queries, and improving overall customer interactions.

The data matches reality, as the majority of Indian AI startups, including Corover.ai, Sarvam AI, Yellow AI, LimeChat,  Kogo AI and Gnani.ai, are building voice-based chatbots for customer service in Indian languages. It would not be wrong to say that India loves AI voice agents. 

So far, only Sarvam AI and Krutrim claim to be building an LLM from scratch. At the same time, everyone has high hopes for the government’s IndiaAI mission, which has called for proposals to build an indigenous LLM within the next eight to ten months. However, IIT Madras professor Balaraman Ravindran has cast doubt on its execution. 

Earlier this year, Ola chief Bhavish Aggarwal announced Krutrim AI Lab and launched several open-source AI models tailored to India’s unique linguistic and cultural landscape. This includes the launch of Krutrim 2, the startup’s second LLM, consisting of 8 billion parameters. He also said that the lab has committed an investment of ₹2,000 crore into Krutrim, with a pledge to increase this to ₹10,000 crore by next year. 

The company became one of India’s first AI unicorns after raising $50 million in early 2024, led by Matrix Partners India, which valued it at $1 billion. 

Meanwhile, in 2023, Sarvam AI raised a total of $41 million in a round led by Lightspeed Venture Partners, with participation from Peak XV Partners and Khosla Ventures. The startup recently partnered with the Unique Identification Authority of India (UIDAI)  to improve the user experience for Aadhaar services. 

It should be noted that foreign investors, not Indian ones, back Sarvam AI. However, there are several Indian VCs who have invested in Indian AI startups.

Speaking of strategy of Indian VCs, Sanyukta said that investors are now prioritising startups with strong AI fundamentals, differentiated IP, and real-world applications that solve critical challenges, especially in sectors like healthcare, agriculture, and education. 

In India, Accel is supporting AI startups through its Accel Atoms 4.0 program. This program will offer up to $1 million in funding for startups focusing on AI and Bharat, specifically catering to middle-income households in Tier 2, 3, and rural India.

Similarly, WTFund—led by Zerodha co-founder Nikhil Kamath—is back with its second cohort, featuring 22 founders across nine startups in its initiative for entrepreneurs under 25. The fund offers up to ₹20 lakh in grant funding, along with mentorship and strategic support to help early-stage startups scale their impact.

Who is to blame? 

Zepto founder Aadit Palicha took Goyal’s criticism personally, stating that it is unfair to blame food delivery apps for the lack of deep-tech startups in India. He pointed out that nearly 1.5 lakh real people earn their livelihoods through Zepto today—a company that did not exist 3.5 years ago.

“₹1,000+ crores in tax contributions to the government per year, over a billion dollars of FDI brought into the country, and hundreds of crores invested in organising India’s back-end supply chains, especially for fresh fruits and vegetables. If that isn’t a miracle in Indian innovation, I honestly don’t know what is,” he said.

Moreover, he blamed the lack of a foundational AI model in India on the country’s failure to build great internet companies. “Most technology-led innovation over the past two decades has originated from consumer internet companies. Who scaled cloud computing? Amazon (originally a consumer internet company). Who are the big players in AI today? Facebook, Google, Alibaba, Tencent, etc. (all started as consumer internet companies),” he said.

However, his argument falls short as OpenAI, Anthropic, Cohere, and DeepSeek did not start as internet companies.

Shaadi.com founder Anupam Mittal also dismissed Goyal’s statement, saying that in the last few months, he has met a few deep-tech companies that have absolutely blown him away. He noted that from AI and space tech to material science, Indian entrepreneurs are ready to take on the world. “But capital and the ecosystem for growth and commercialisation are severely lacking. Founders can do most things but not everything,” he added.

On the other hand, Zoho founder Sridhar Vembu took Goyal’s statement positively. In a post on X, he said that the government cannot invent a better operating system or a smarter robot.

“The government should not even fund such things—it is not usually good at picking winners and losers. At best, it can conduct competitions where companies participate and then purchase the best Indian products,” he added.

He said that he sees Goyal’s call as a challenge to our engineers and technologists, not as pointing fingers.“What we need are smart engineers who roll up their sleeves and get it done.” 

He explained, using an analogy, that Indian deep tech startups should ship vitamins and painkillers to fund their business, even as they work on a cancer cure. “How we pay for our big tech ambition is very much part of the engineering problem that smart engineers must solve. We can do this.