Raju Rishi

The through-line

Raju Rishi's account of how venture returns actually get made comes down to two rules, and he states them as rules. "The first key to driving alpha in venture, I think, is to skate where the puck is going" [3], and "The second is have an unfair advantage" [3]. Everything else he says about sector research, geography, fund size and founder relationships is a working out of those two propositions. The first is an argument about timing and about being early to a place, a sector or a technology shift before the rest of the market prices it in. The second is an argument about access, which for him means proprietary deal flow rather than winning competitive rounds.

Underneath both sits a bias formed early: technology that never reaches a product is wasted. He describes Bell Labs as "really good at the tech side of things, but not really good at the product side of things" [3], and says he left in part because "a lot of the stuff I worked on didn't see the light of day, and I got frustrated" [3]. He remains sore about it, noting that "there were so many good patents in that company that uh other people exploited" [3]. That instinct, to ask who will actually commercialise a capability and when, runs through his investing in enterprise software, artificial intelligence and robotics [1].

On skating where the puck is going

The puck metaphor has two dimensions for him, geography and sector, and he treats both as bets on where value will sit in five to ten years rather than where it sits now. The geographic version is RRE Ventures planting a flag in New York City "before there was really anything in New York City" [3], at a time when venture was concentrated in Silicon Valley and Boston "because everybody needed to be headquartered where the tech talent was" [3]. His counter-argument was demand-side: New York holds 18 million people and is simultaneously the financial, media, healthcare and fashion hub, so "why would you actually start a company where you had tech talent as opposed to where you had the customers?" [3].

The mechanism that resolved that tension, in his telling, was cloud infrastructure. "The big inflection point was Amazon Web Services" [3]: once AWS removed the need to own a rack and outsourced engineering removed the need to sit next to your engineers, "if you could start your company anywhere because you have AWS and you have outsourced tech talent, you would start it where your customers are" [3]. He treats this as personally verified rather than theoretical, having started three companies in the New York market and migrated two of them to Boston "because we still needed that rack and that tech talent" [3]. Friendly, low-taxation state law for entrepreneurs was the second reason New York's rise was, in his word, inevitable [3], and DataDog is the example he gives of what proximity to that emerging ecosystem yielded [3].

On thesis investing, and knowing when to shelve

RRE runs formal sector research on a fund-by-fund cadence: "We pick one or two sectors, we have a partner do primary research, and go to universities" [3]. The output resolves into two verdicts. Either it is not a good venture bet, in which case, in his phrase, "we just destroy it. Keep the work but destroy it" [3], or it is a good bet whose timing is wrong, in which case the work goes on the shelf until conditions change [3]. The discipline is in the second category, because a shelved thesis is an asset to be picked back up rather than a rejected one.

Crypto is his proof case. RRE looked at it roughly 17 years ago, when the landscape was "Digicoin and you know, like weird esoteric types of models," and concluded it was not a good model [3]. Two years later the firm reopened the same research, judged the timing right, and began investing, becoming "one of the very earliest investors in Ripple" and "one of the very earliest investors in Digital Currency Group, DCG," with exits that came "before the world got into crypto" [3].

On the four inflection points

He organises technology history around three tectonic shifts, "PC, internet, mobile," each of which "created trillions of dollars worth of value" [3], and insists all three shared exactly three characteristics. First, a platform to develop on: Microsoft, Unix and Apple OS for the PC, browsers for the internet, and Palm Pilot, Blackberry, iOS and Android for mobile [3]. Second, a large open source developer base "allowing you to create applications rather quickly" [3]. Third, "radical enterprise interest" [3]. He asserts that a fourth inflection point is now arriving and treats its identity as obvious enough to leave unstated [3], which is consistent with his stated focus areas of artificial intelligence and robotics [1].

The more useful analytic move he makes with this framework is inverting it. Rather than chase the shift itself, he asks "Which industries haven't taken advantage of those three inflection points?" [3], on the theory that the backlog of unrealised value there is where the outsized returns sit.

On the industries that missed the wave

Healthcare is his primary answer, and his diagnosis of why it lagged is structural: "healthcare is highly segmented, it's proprietary, it's highly regulated" [3], which prevented it from exploiting platforms and open source the way other sectors did. His prescription eight years ago was sequencing rather than applications. "If healthcare is going to jump into the game and become really profound, the first thing we need to do is unlock the data" [3], and "if you want to really get healthcare into the modern era, you've got to create data interoperability" [3]. That thesis led him to Redox, founded by engineers who had left Epic and based in Madison, Wisconsin [3].

Satellites were the second category, an industry of "million-dollar boxes, old-school stuff" that had not absorbed Moore's law, until the economics collapsed from roughly $10 million for a satellite to about $100,000 for a CubeSat [3]. RRE has taken companies in that category public [3]. His continuing interest in unlocked healthcare demand also shows up on the consumer side, where he has argued for the growing appeal of functional medicine, its value to consumers and its benefits to the wider economy [2].

On fund size and portfolio construction

He is deliberate about staying small and explains it in terms of return mathematics rather than modesty. Firms historically start small and grow as they perform, and he declines to knock that, but "if you look at the metrics associated with fund performance, that $250 million bucket tends to be the one that can get the highest amount of multiples" because it lets you bet on seed and mostly Series A, "and if you make the right picks, you're going to get a 10x or a 50x or 100x" [3]. A billion or two billion dollar fund faces a fork: bifurcate the firm so seed is measured and operated differently, or move into B, C and D rounds where you must deploy $50 million or $100 million per company and "those aren't going to get as accretive," because there is much less inflection left after a Series C or D [3]. He accepts the cost of the early-stage approach openly, that some Series A bets fail because the thesis, the regulation or the tailwind changed, on the grounds that "if you get those series A's right and those seeds right, you can blow up a fund" [3].

Fund eight sits in the middle of the firm's 250 to 300 million range and is deployed over three to four years [3]. Pacing is a risk control, not an accident: "We love the time diversification as a way to avoid hiccups that happen. People that are investing all their money in two years, you're going to see a really bad vintage in there" [3]. Roughly 10 to 15 percent goes into seed, split into two kinds. The lightweight seed is around $500k where "we like this entrepreneur, we like the sector, but we haven't seen enough to really make a significant bet," taken without a board seat and with the aim of underwriting the Series A later [3]. The second kind is incubation.

On unfair advantage and recycled founders

His definition of unfair advantage is narrow and concrete: "the unfair advantage in this particular case is proprietary deal flow" [3]. He grounds it in longevity, noting that few venture firms have been around 30 years, and that duration produces repeat entrepreneurs who come back saying, in his rendering, "I loved working with you. I had a great outcome. I want to work with you again. I'm not shopping this deal around" [3]. Some arrive without an idea at all, in which case RRE incubates alongside them; others arrive with an idea and want RRE as the first check [3].

He also runs an idea inventory of his own. He describes "probably a list of 10 ideas that I think can be actually really really powerful, but you know I don't have time to do it," held for the right entrepreneur, and notes that one idea he pitched during incubation "is still on the shelf" because the founder did not love it [3]. The incubation process he describes is slow and conversational, meeting "once every couple of weeks," trading ideas both ways, with the founder going off "and do a week or two worth of homework" before coming back [3].

On architecture as an underpriced risk

The company he uses to illustrate all of this is Cato, built with Boris Bogatin, a founder he had backed before and served on the board of, and who had since built Sidon in over-the-top video for Asian markets [3]. The insight behind it is that boards now mandate penetration testing at Series B or C and beyond, because leaked data "can shut down a company," while "the one thing that people don't do is architecture vetting" [3]. Early architecture choices get locked in and become permanent, and "the cost of doing a bad architectural choice or technical debt can be millions if not billions of dollars downstream" [3]. The problem is worsening because busy CTOs are delegating by default: "small architectural decisions are being made by engineers now. They're like, 'Should we use this database or not?' Boom, they just picked one that they're comfortable with" [3].

Cato is an AI layer, a play on CTO and AI, that reads your software architecture and flags an unsupported open source module you may have to replace or that may get hacked, then tells you what it would change to optimise for performance, for reliability or for cost, "all of the illities and things like that" [3]. His first instruction to the founder captures how he thinks about early-stage risk: "your job number one is to recruit a great technical co-founder" [3]. The founder recruited the former VP of engineering from Splunk as co-founder [3].

Takeaways

• The two rules for outperformance in venture are to skate where the puck is going and to have an unfair advantage, which in practice means proprietary deal flow from repeat founders who do not shop the deal [3].
• Fund size drives return profile: the roughly $250 million fund produces the highest multiples because it concentrates on seed and Series A, while billion-dollar funds must write $50 million to $100 million cheques into later rounds where less inflection remains [3].
• Deploy over three to four years rather than two, because compressed deployment guarantees a bad vintage [3].
• Sector research should end in one of two verdicts, kill it or shelve it for timing; crypto was rejected roughly 17 years ago and picked back up two years later, leading to early positions in Ripple and Digital Currency Group [3].
• Every major inflection point so far, PC, internet and mobile, combined a development platform, an open source developer base and radical enterprise interest, and the best opportunities lie in industries that missed those waves [3].
• Healthcare lagged because it is segmented, proprietary and regulated, so the unlock is data interoperability first, which is the thesis behind the Redox investment [3].
• AWS plus outsourced engineering severed the need to sit near tech talent, which made New York's rise as a startup hub inevitable once companies could locate near their customers instead [3].
• Architecture decisions are the underpriced risk in software: boards mandate penetration testing but not architecture vetting, and delegated database and framework choices create technical debt worth millions or billions later [3].