1980sTwo or three languages, and everyone knew them. Across 30 to 35 years in the industry he watched programming go from a handful of languages in the 1980s and 1990s to a count nobody tracks, until the work itself became integration rather than writing code.
Late 1990sTelecom hardware was a big business. Through the late 1990s and early 2000s the market was huge. Today he counts about five companies in the world selling that kind of hardware, and calls the industry very hard to enter.
Jan 2016Ideaspring says the word technology out loud. At a press conference towards the end of January, Doraswamy and his partner told the room they would fund entrepreneurs who love technology, at a moment when everybody else was talking SaaS. 2016 was also the year India was moving from B2C into B2B.
2016Fund one goes out under a name they had to invent. Refusing to file SaaS under deep tech, they coined product innovation as the umbrella wide enough to hold both, and sold LPs a patient-capital thesis of five to eight years. The fund would put ₹120 crore, about $18 million, into 16 companies on a life written as seven plus one plus one.
2021-24The world order changes under the portfolio. In the three years before this conversation, Covid, the war in Ukraine, the war in Israel and China's role broke the global supply chain, valuations overshot, and rounds at Series C and beyond went quiet. Exits, in his word, became ridiculously painful.
2024India gets an official definition of deep tech. The government's test is recreation time: a rival should need two to three years to come close, even throwing people at it. He rates it as good as policy gets on something this hard to pin down.
TodaySix exits out of sixteen, nothing shut down. Fund one still has about two years to run, every product from the six exits is in use somewhere including Numocity at ABB, and 45% of capital is already back. Fund two is ₹265 crore and carries twice as many real deep-tech companies, with 28 companies now across the two. He now says the asset class needs twelve years, not the seven plus one plus one he wrote.
2025Fund three hits the market. Targeted for around the same time next year on the same agnostic mandate, pre-seed to Series A, once the two things that have to happen first do.
2029-34Semiconductors, entered at the boring end. He puts India's chip opening five to ten years out and would start at 40 to 50 nanometre parts that need no leading-edge fab, not at a high-end AI chip that has to be sold in the US anyway. Defence buying from startups through iDEX is the other unlock.
