Lightspeed halves its India fund, and 64% of the money never came from it

Lightspeed is raising a new India fund dedicated to early-stage AI, and the target is $250m. The thesis in the investor letter is expansive and the vehicle behind it is smaller than the last one. Both facts come from the same document.

The numbers as they stand:

• Target: $250m, exactly half the $500m raised in 2022, with 80% already committed.
• Reported target before the letter: $300m to $350m, so the real figure is 17% to 29% lower.
• Accel's comparable India fund in August: $550m, raised inside a coordinated $3.5bn effort.

“AI will create more value in India than the internet did.”

— Lightspeed, Investor letter, reported by TechCrunch

From the Lightspeed investor letter, as reported

The firm's own explanation is reasonable and worth stating before anything else: the size is matched to how fast it is currently deploying and to a shorter investment period of about two and a half years, and a smaller fund lets it focus on individual deals and return to investors sooner. That is a coherent argument for halving a fund, and it is not the same as retreating from a market.

Where the India money actually came from

The figure that reframes the whole announcement sits further down the letter. Lightspeed's dedicated India and Southeast Asia funds have deployed roughly $900m into the region, while its global funds have put another $1.6bn into companies from the same portfolio. So 64% of the capital that reached the India portfolio did not come from the India funds. The dedicated vehicle is the smaller half of the cheque book, and always has been.

Scale explains why a $250m fund is not a signal about conviction. Lightspeed manages more than $65bn, raised $9bn across several funds last December in its largest haul ever, and that round included a single early-stage venture fund of $980m. The new India fund is 0.385% of assets under management and 3.9 times smaller than one global early-stage vehicle. Read as a share of the platform, it is a line item. Read as a regional commitment, it is half what it was.

One comparison for proportion

The market context is the part investors are actually betting on. India has not produced a frontier model developer and attracts far less AI investment than the US or China; the argument for it runs through the application layer and a very large pool of software engineers. Lightspeed has backed Sarvam AI, one of India's leading LLM developers and a company selected by the Indian government to help build sovereign models, alongside a portfolio that is otherwise quick commerce, consumer internet and household services. Dedicating a whole regional fund to one theme is new for the firm.

One comparison puts the sums in proportion without diminishing them. The compute contracts behind the datacentre buildout come to $1.5tn over two years, which makes this entire fund 0.017% of what the industry has already committed to spend on hardware. Application-layer investing does not need those sums, which is the point: OpenAI has been selling advertising in India since before this week's Asian expansion, and value at that layer accrues to whoever owns the interface, not the silicon.

Informational material, not investment advice. The fund is still being raised, the figures come from an investor letter seen by TechCrunch, and Lightspeed declined to comment.