Sarvam AI funding accounted for about 87 per cent of the money raised by India’s language AI startups in 2026, according to a new report from market intelligence firm Tracxn. The report, titled Indic AI: Building AI for Bharat and released on 28 October 2026, tracks 34 companies building artificial intelligence for Indian languages, using data up to 15 October 2026.
The sector raised 399 million dollars across nine funding rounds in 2026, up from just 23 million dollars in 2025. The round count stayed the same at nine, so the jump came from larger cheques, not more deals. The two Sarvam AI funding rounds of 2026, valued at about 255 million dollars and 90 million dollars, together account for roughly 346 million dollars of the year’s total. Outside Sarvam, the rest of the group raised 54 million dollars across seven rounds in 2026, up from 23 million dollars across nine rounds in 2025.
Since 2021, the 34 companies have raised 657 million dollars across 56 rounds, with all time equity funding of about 687 million dollars. Deal activity peaked in 2024, which saw 15 rounds, the most of any year. Every round in the dataset sits at seed or early stage, and Tracxn has not recorded a late stage round, a sign the sector is still establishing business models rather than showing mature operations.
Two kinds of Sarvam AI funding rivals
Tracxn splits the 34 companies by how central Indian language capability is to their product. The first group holds 18 companies for which language AI is the core offering, building speech recognition, translation systems, language models and training data. It includes Sarvam, Gnani, Soket AI and BharatGen. Together they account for about 452 million dollars, or 66 per cent of the total.
The second group holds 16 companies where Indian language handling strengthens a broader enterprise product. Firms such as Skit.ai, Nuplay AI and GreyLabs sell customer support, voice automation and financial services software in which multilingual capability is a selling point, not the whole product. That group has attracted about 234 million dollars. Core language firms must spend heavily on model development, infrastructure and training data, while enterprise firms can sell into existing workflows and tie revenue to business outcomes.
How Sarvam AI funding compares with Krutrim
Sarvam AI funding stands apart even from the next best funded company in the set. Total Sarvam AI funding stands at about 389 million dollars, while Krutrim has raised 75 million dollars. Together the two account for 464 million dollars. Both were founded in 2023 and both have reached unicorn status.
The two companies have taken different paths. Sarvam pursues a full stack generative AI strategy across foundation models, language technologies and enterprise applications, and is one of the government’s four selected sovereign model builders. Krutrim has moved from foundation models to AI cloud infrastructure, paused chip design work in late 2025 and reported its first profit. Tracxn still counts Krutrim because of its earlier Indic AI funding and involvement, even though it no longer develops Indic language models.
Enterprise demand beyond Sarvam AI funding gives an early signal of real use: GreyLabs serves more than 50 financial institutions, including RBL Bank, AU Small Finance Bank and IDFC FIRST Bank. Adoption alone does not prove profitability or scale.
Why Sarvam AI funding concentration matters
Sarvam AI funding concentration also shows up in exits, or the lack of them. Only two acquisitions have been recorded among the companies studied, and both happened in 2019. Reliance Jio acquired about 87 per cent of conversational AI company Haptik in a deal valued at around 101 million dollars, including growth capital. Reliance Industrial Investments and Holdings acquired a majority stake in Reverie Language Technologies for about 27 million dollars. No further acquisition has been recorded since then. Exits matter because venture backed ecosystems depend on them to return capital to investors.
Global competition adds pressure: Google’s Gemini models already support multiple Indian languages, and OpenAI has built an Indian language evaluation benchmark with 261 domain experts. As global models improve in regional languages, domestic startups may need to compete on accuracy in specialised domains, deployment flexibility, data governance, integration and cost, rather than on basic language support alone. Tracxn also notes domestic capability does not remove dependence on global chips, cloud or foundation technologies.
Elsewhere, Singapore’s SEA-LION programme uses public funding, African language AI work has used grants, and Saudi Arabia and the UAE run larger state backed programmes. India’s Indic AI ecosystem, by contrast, is largely driven by competing venture backed companies, which ties it to investor appetite. Regional funding figures are not directly comparable, mixing equity, public funding and infrastructure commitments. Tracxn cautions its figures, including the Sarvam AI funding totals, cover only its 34 companies, not India’s whole AI industry.