Fundamentum Frontier Advisors (F2A), a new investment platform under the Fundamentum Partnership umbrella, has launched a ₹3,000-crore fund targeting growth-stage AI, consumer tech, and frontier-technology startups in India.
The announcement comes at a time when investors in the country have become bullish on both deeptech and artificial intelligence (AI).
"At this point, we're seeing a critical mass of companies which are ready to scale and they need access to a little bit more capital," said Ashish Kumar, co-founder of Fundamentum and general partner at F2A, in an interview with Mint. "There's a large 'missing middle' gap for AI and deeptech. That's the space we're trying to occupy."
Investors are writing larger cheques as deeptech companies begin to show maturity and growth. With AI, investors have become more selective but are also more willing to write larger cheques to startups, as long as their core intellectual property is sustainable in the long term.
The investment vehicle is structured as two parallel instruments - a ₹2,000 crore SEBI-registered category 2 alternate investment fund called Fundamentum III- AI and Deep Tech, and a ₹1,000 crore offshore vehicle for investors who prefer to participate outside the Indian fund structure.
The fund plans to build a concentrated portfolio of 12 to 15 companies, writing first cheques of between ₹50 crore and ₹150 crore, with average investment of approximately ₹90 crore. F2A expects to complete raising its targeted amount within the next three to four months. It plans to invest in companies ranging from Series A to Series D.
Kumar has roped in Debraj Banerjee - former senior fund manager at Small Industries Development Bank of India Venture Capital Ltd - to co-lead the fund as a general partner. His portfolio of investments include companies like quantum computing platform QpiAI, cybersecurity startup Pantherun, space-tech company Digantara, healthtech SigTuple, and AI algorithm startup Myelin Foundry.
Given the firm's growth stage focus, the tenure of the fund has been set at eight years, instead of the traditional 8+2 or 10+2 lifecycle that most deeptech investors follow. F2A's point of entry into deeptech companies is at a stage where they're already generating revenue or are close, thus minimising risk to the larger portfolio. "We will only invest in companies where we have a reasonable chances of them hitting the IPO market in a 5 to 6 year time period," said Kumar.
Splitting focus
The fund is split broadly into three categories of investments - consumer AI, led by Kumar; sovereign tech, overseen by Banerjee; and enterprise technology, which the two will do together.
"We're thinking of roughly 40% to sovereign tech, 40% will be consumer and 20% will be enterprise," Kumar said regarding how F2A would be allocating capital from the fund.
With deeptech, the firm isn't looking to spread itself too thin, instead opting to focus on specific sectors, with room to make investments in emerging sectors down the line. Its targeted sectors include semiconductors (particularly vertical, industry-specific chips for sectors such as electric vehicles (EVs), smart meters, and surveillance, drones, energy, space, and vertical robotics.
Quantum remains on the watchlist but is unlikely to feature among the fund's earliest deals, given its limited breadth of investable companies in India today.
On the consumer AI front, the firm's interests lie in companies that are asset-light and don't require physical infrastructure. Instead, Kumar said he's more bullish on companies that are able to do what he calls "productization of services" citing companies like Urban Company, Snabbit and Pronto as an example.
This is in addition to the 'House of Apps' approach that many consumer AI companies have begun moving towards. It's where startups create a single platform that leverages their propietary user data to iterate and build multiple products under the same banner while targeting the same pool of customers. "You can always build a wrapper and build on consumer insights unique to your platform. The other thing you can do is play on the model layer of AI and create some sort of small language model," said Kumar. "The best, more defensible companies will take advantage of both."