Lightspeed Plans a $250 Million Fund for New AI Startups in India

Lightspeed wants to raise $250 million for Lightspeed India Partners V, a fund for young AI companies in India and Southeast Asia, TechCrunch reported on Sept. 24, 2026.
The new fund would be half the size of the last one. Lightspeed raised $500 million for India and Southeast Asia in 2022, and 80% of that money is now committed, according to a September 2026 investor letter seen by TechCrunch. Lightspeed first listed the new India fund in a U.S. regulatory filing in late April 2026, without giving a target size then.
Lightspeed plans to start investing from the new fund within two months of Sept. 24, 2026. The plan is to make first investments over about two and a half years. Starting with this fund, the firm will raise money for its India funds on the same schedule as its global funds for the first time.
The firm has backed Sarvam AI in India. Sarvam was chosen by the Indian government to help build AI models kept under national control. Globally, Lightspeed is an investor in Anthropic, xAI and Databricks.
The broader context here is simple. The 2022 fund was put to work steadily over four years, and the smaller new fund with a shorter window is built for backing companies earlier, like planting seeds rather than watering full-grown trees. That means more help with basic technical choices.
In my view, focusing only on AI is less about trends and more about being clear. AI startups today include tools adapted from large models, systems that work in Indian languages, helpers that find information, and workplace assistants tied to normal business software. A separate fund lets investors judge them fairly, without comparing them to shopping apps or other software with very different costs.
Looking at what this means for builders, early money matters most. Choices about which AI model to use, the cost of running it, and access to computers shape a product from the start. If the fund closes as planned, it should give new teams in India and Southeast Asia another clear place to get first serious backing, plus a link to lessons from older AI companies.


