Why the Future of Venture Is Hard Tech, Drones & Physical AI | Ian Rountree

Summary

• José frames Roelof Botha’s “return-free risk” charge around the median fund and notes that Ian voluntarily put a hurdle rate in his fund. José’s caveat is that median venture can underperform public markets or T-bills, while the top 1–2% put up “mind-boggling numbers.” Ian’s answer is persistence: top venture firms repeat across vintages because “the asset picks the investor rather than the other way around” — anyone can buy Micron, but a pedigreed founder shortlists Sequoia — making outlier-fund selection the relevant LP question, not the median.
• Cantos’ answer to the mega-platforms is deliberate smallness: Fund IV grew only from $50M to $70M so it can fly “under the floor of economic rationality for the big funds.” Writing $1.5–4M first checks (“if the company has a website, it’s too late for us”), everyone on the investing team can tell a founder “your company is a meaningful percentage of my net worth” — alignment the multi-stage giants may not offer at inception.
• He now prefers “hard tech” to “deep tech,” having “come to question the myth of the breakthrough.” Most value is “a thousand little things compounding” — engineering and execution rather than primarily science risk — and he skewers fusion’s $10-trillion narratives: “what happens if you have two fusion energy companies succeed? Do you understand that competitive arbitrage is a thing?”
• Howard Marks’ consensus/non-consensus matrix “doesn’t apply to early-stage venture,” and AngelList data from Abe Othman shows price positively correlates with performance. Insolvent startups must become consensus before the next round, so edge is access plus speed — SAFEs can be signed “same day if we meet them before 2:00 p.m. Pacific.” His biggest misses: passing on price, and passing on decks like Castelion’s pre-seed, which Cantos later bought at Series A “a heck of a lot more” expensive.
• The core thesis: don’t sell technology to staid incumbents — vertically integrate and compete with them. Earth AI applies for its own mineral rights rather than selling AI leads to miners; Shinkei sells branded fish, not robots; the lumber startup designed a bigger saw than the industry had built and now runs its own Mississippi mill, selling “a literal commodity” off “a preferential cost structure.”
• Neros went from a pre-seed check into Soren, a drone-racing world champion, in August 2023 to America’s highest-volume drone maker at 1,000 drones a week, targeting a million a year. A Pentagon official’s tell: “we’d give a billion dollars to someone who could make a million of anything” — and Ian’s corollary, “if you have a drone that’s second best but you can make 10,000 more of them… I know which one I’m picking.”
• The strategic risk is China’s grip on the physical-AI supply chain: 50–70% of a humanoid robot’s BOM is actuators and motors, “and we don’t really make cheap motors in the Western world.” Rare earths, processing chemicals, copper and motors are concentrated in China — Ian says Beijing’s leverage already forced the U.S. to back down on sanctions — and if physical AGI arrives on that stack, “they’ve got all the power.” He invests in the U.S. and its allies as a principle, not as a protest; without that principle, “I would find a way to invest in Unitree.”
• On momentum and meme-company dynamics, Ian is torn, while José proposes being “half stubborn and contrarian” and “half” acknowledging “the game on the field.” Path-dependent, unprofitable startups “can’t afford to be non-consensus for more than a few years,” and José argues secondary liquidity now makes venture trade more like crypto — but Ian’s stated endgame remains “something enduring that compounds forever,” not dumping on retail.

Deep dive

1. Venture is “return-free risk” for the median — and mind-boggling for the persistent top

• José frames Roelof Botha’s provocation that the median fund underperforms public equities over long periods and, in certain vintages, the risk-free rate. He also notes that Ian voluntarily put a hurdle rate in his fund. José’s self-aware caveat: “How do you know a VC is talking their book? Their mouth is moving.”
• José’s counterpoint is that the top 1–2% “put up mind-boggling numbers.” Ian’s rebuttal is persistence: top-10 hedge funds one year barely correlate with the next year’s list, while venture’s top 10 are “highly correlated” vintage to vintage. The mechanism: “the asset picks the investor rather than the other way around” — no one stops you buying Micron, but a hot founder shortlists Sequoia. José characterizes Botha’s implication as: don’t invest in venture funds, invest in Sequoia.
• Why Ian does it at all: he grew up an only child reading science fiction, likes thinking about the future and having a scoreboard, and sees technology as a major measure of societal progress. “There’s a job where you can just sit there and think about what’s going to happen in the future and then you go monetize that.”

2. Stay small enough that alignment is the product

• Ian’s edge philosophy: “I’m a very competitive person, but I’m lazy enough to want to be unfairly competitive” — structural advantage over effort. He grants the platforms have never been stronger: capital, media teams, lobbying, and people “one or two calls away from the very top” of labs, hyperscalers and government.
• The Cantos counter: Fund IV grew only from $50M to $70M so its $1.5–4M checks stay “under the floor of economic rationality for the big funds,” and everyone on the investing team can look a founder in the eye: “your company is a meaningful percentage of my net worth… I’m not looking at you as some call option.”
• His warning to new Series A/B entrants: without a truly unique advantage — say, the former Palantir or SpaceX management team backing old colleagues — you can’t out-compete Sequoia, Founders Fund, Thrive and Andreessen for the young founder’s shortlist.

3. Deep tech was never a decision — Silicon Valley was “living in flatland”

• The origin: joining SoFi in 2012 when “fintech was a new word” and applying software to non-software industries was novel — his startup was the only one at the Vegas asset-backed-securities conference “immortalized in The Big Short.”
• The drift was gradual, not binary: vertical-software founders kept saying the product improves with a sensor at the edge, then a GPU, then “maybe we hack a KUKA or Yaskawa robot arm” — “things went three-dimensional… one day they were so physical in nature people started calling it deep tech. That was more a labeling shift than a decision.”
• Why it took an independent small fund: at big firms “one job is investing and the other is playing politics in your own organization.” Ian says he is bad at the second job and might have been stopped by senior partners.

4. “I’ve come to question the myth of the breakthrough”

• He now prefers “hard tech” because deep tech “denotes a lot of science risk” and puts technology first; the canonical deep-tech deal is “some PhD… a hammer looking for nails — that’s not what a business is.” Occasionally a Transformer/GPT-3 moment changes everything, but mostly it’s “a thousand little things compounding” — engineering and execution rather than betting primarily on new science.
• The fusion takedown: eye-popping valuations on 20-year timelines justified by $10-trillion talk. “What happens if you have two fusion energy companies succeed? Do you understand that competitive arbitrage is a thing?” — price wars, licensing-versus-building ambiguity, and duration risk: his mentor backed a fusion company in 2002 that is still working on it. “What do you do with that venture fund?”
• His old dichotomy — software asks “will they come,” deep tech asks “can you build it” — is now too simplistic: a little of both risks is better than “a ginormous amount of one or the other,” with execution (recruiting, retention, culture, capital, and go-to-market) the biggest factor of all.

5. Howard Marks doesn’t apply at inception; speed and access do

• José’s tension — contrarianism versus adverse selection — gets a structural dismissal: the consensus/non-consensus matrix comes from Oaktree’s world of liquid debt, more information, quantitative analysis and solvent, profitable issuers. “We over-apply that to venture.”
• The inversion: funding unprofitable businesses means “by definition you need to be somewhat consensus before they run out of money” so a bigger checkbook can do the next round. And not every VC is looking at the same asset at the same time — at inception the company may not even be incorporated.
• The operational edge: pre-seed SAFEs mean “we can shake hands and sign and wire same day if we meet them before 2:00 p.m. Pacific… if another investor meets them next week, sorry, we beat you.” By Series A/B pressure builds publicly and “when they fire the starting gun, everybody’s competing.”

6. Price is signal, not noise — and his misses prove it

• Abe Othman’s AngelList analysis finds a positive correlation between valuation relative to stage and performance — “seems VCs are actually kind of good at pricing things.” Ian’s confession, seconded by José: “a lot of my biggest misses were saying no when I liked everything except the price.” Telling a founder the valuation is too high can be cope, although Ian says he has also genuinely passed on valuation and regretted some of those decisions.
• Second miss category: judging decks instead of founders — “I was looking at a 7- to 20-page pitch deck pretending there was an existing business to underwrite,” when the compelling entrepreneurs either represented most of the information he was missing or would tack until something worked.
• The Castelion specimen: he passed on the hypersonic-missile pre-seed deck before defense tech was cool (“who’s going to do the next round?”), partly fearing overlap with another hypersonics investment. “That was a profound miss. Had I spoken to them in the first five minutes, I probably would have been banging down their door trying to give them a SAFE.” Grant led the seed investment while at In-Q-Tel; Cantos finally invested at Series A after Grant joined the firm, paying “a heck of a lot more.”

7. Six attributes, one gravity test, and knowing in five minutes

• Cantos scores founders on six attributes: velocity/drive, talent gravity, narrative building, growth mindset, financial mindedness (“do you understand the point of a business is to produce profit margins above its cost of capital… or do you just think having a startup is cool”), and encyclopedic industry knowledge. Calibration is everything — Grant’s rule, “your bar for excellence is the best you’ve ever seen”; in the $4M demo-fund days, “everybody seemed smart to me.”
• Verdicts come fast, and first negative impressions almost never reverse — the exception is low baseline, steep slope: bet on the bacterium evolving every few seconds over the primate not evolving at all. For a SpaceX-veteran team like Castelion’s, he even questions the growth-mindset weighting: “if your y-intercept is already exceedingly high, do I really care?”
• The gestalt test is gravitational: do the founders bend spacetime — “if there’s not some small part of us that’s like, oh my gosh, I kind of want to quit and join this company… then they don’t have it.” Roughly a third of diligence time goes to founder psychology, which informs most of the score; the business discussion also assesses industry knowledge, financial and strategic thinking, narrative-building, and the separate market or “wave” score.
• José’s probe on courage — the willingness to “burn all your bridges and do something completely nonlinear” — gets folded into the existing attributes, though both agree it could be a seventh. José’s suggested diligence question is: “when were you right when everybody thought you were wrong?”

8. Rider and wave — and half-surrendering to the momentum game

• The framework: “if you have the best surfer in the world but there’s no swell, you ain’t going anywhere.” The newer, awkward internal debate: with the biggest wave ever behind a deal, do you need the world’s best surfer? “Probably not” — so how much should the founder bar drop?
• On meme-company dynamics — José cites Project Prometheus and ex-founder mega-raises — Ian is torn: “it’s been rational strategy… I can’t blame them,” since insolvent businesses “can’t afford to be non-consensus for more than a few years.” “I can’t figure out if this is an aspect of the game we’re finally acknowledging rationally or if it’s a mind virus.”
• José’s settlement is to be “half stubborn and contrarian and half acknowledge that this is the game on the field,” which Ian says he likes. José also argues that secondary liquidity makes venture trade like crypto — momentum through liquidity gates rather than underwriting an IPO or acquisition. Ian’s guardrail: he takes carry in kind and refuses to fund his kid’s school by “dumping a crappy asset on less-informed investors” — “the endgame is something enduring that compounds forever.”

9. Process: sleep on it, keep an ace, let Claude write the memo later

• Hard rule: no same-day commits — “let us sleep on this and get back to you tomorrow.” Each of the three investors has an ace to use if the other two disagree, knowing “your neck’s on the line”; decisions land within a week, sometimes two.
• The memo comes after the investment decision: calls, texts, Slack threads and Granola notes get dumped into a channel — “that is the raw data,” the time series — and Claude later turns it into a memo, with the decision articulated in the next quarterly LP update.

10. Don’t sell to the incumbents — become them: mining, fish and a bigger saw

• José’s setup — the S&P 500 is roughly 40% tech, headed toward 100% — meets Ian’s sharper version: selling technology to staid industries “is rarely enough to change the world… it’s not ‘please, Mrs. Boeing executive, buy my actuator.’” Be the Tesla to Ford, the SpaceX to United Launch Alliance, the Anduril/Castelion/Neros to Northrop, Lockheed and Boeing.
• In practice: Earth AI buys the drill, applies for the mineral rights and owns the deposits rather than licensing AI-driven leads “for some small sliver”; Shinkei keeps its slaughtering-and-processing robots invisible behind a consumer fish brand — “you won’t see anything about the robots on the website.”
• The lumber company’s operators come from Blue Origin, Planet Labs, SpaceX, Anduril and the lumber industry. After the pandemic, they discovered that mills, not forest owners, captured the price spike — local mills are monopsonies, and shipping logs 300 miles to the next one eats all the owner’s profit.
• The saw story is the thesis in miniature: tree prices fall past a certain size because “it doesn’t fit in the saw” — only two companies make the saws, and those are the biggest available. “Can you just design one a little bigger?” “Huh, never thought of that,” after decades. Result: a fully operational Mississippi mill with two more planned — “our end product is a literal commodity… we’re using our technology to have a preferential cost structure.”

11. Shenzhen is a motherboard, and the West doesn’t make cheap motors

• The physical-AGI vulnerability: with 50–70% of a humanoid’s BOM in motors and actuators, “we’re potentially in trouble because we don’t really make cheap motors in the Western world.” His tell: “sometimes you’ll meet a defense-tech startup that’s got a little drone — go look at the motor. I guarantee it will say Made in China.” Rare earths are mined and processed in China, the processing chemicals are there, and copper and motors are also produced or assembled within that supply chain; Shenzhen from above “might look like a motherboard.”
• The leverage is already proven: Ian says the U.S. gave up on sanctions on China after Beijing flexed its control of rare earths and Trump backed down. If humanoids reach homes, industries and battlefields on that stack and China “can flip off a switch, then they’ve got all the power.”
• José’s challenge — then why not invest in China? Ian says, “We’re an American venture capital firm… I’m not investing in China in protest” is not his framing; he wants to build capabilities in the U.S. and allied countries and is concerned about Xi Jinping’s CCP. He lived in China in 2008, when “it was really opening up.” The honest kicker: “if I didn’t have that principle and I wanted an investment in robotics, I would find a way to invest in Unitree.” Neros, meanwhile, is onshoring or friend-shoring all parts of its military-drone stack.

12. Neros: parents’ backyard to 1,000 drones a week in under three years

• August 2023: Cantos led the pre-seed into Neros when it was just Soren — a high-school dropout who won the drone-racing world championship, alarmed that Ukraine’s war was being fought with consumer-style quadcopters. Two weeks after backing him, Ian received a prototype video “doing 130 miles an hour, pulling 5–6 Gs in his parents’ backyard.” Two months later, Soren and the team were meeting Ukraine’s Ministry of Defense.
• Cantos introduced him to Shaun Maguire at Sequoia, who led the seed. Less than three years on, Neros is the highest-volume American drone manufacturer — it has crossed 1,000 drones a week, with a stated goal of a million a year.
• The Pentagon’s real demand signal, via an assistant undersecretary’s game (“you tell me what you think I want and I’ll tell you how wrong you are”): “we’d give a billion dollars to someone who could make a million of anything.” Ian’s corollary: “if you have a drone that’s second best but you can make 10,000 more of them than a drone that’s a little better, I know which one I’m picking.”
• Proof point: in the Pentagon’s Drone Dominance flyoff, Neros sits number 2 overall — edged by a joint Ukrainian-British entrant — is the number 1 American supplier, and is far and away the one on the list that has actually shipped the most drones.

13. Class-1 drones are a land-war product; sovereign new primes are the rare exception

• Market structure: FPV/Class-1 quadcopters “exploded onto the scene” because of Ukraine’s land war — open plains, where many wars have been fought — and are here to stay: “we are going to make millions of them” for deterrence and to lend to allies. Japan, Korea and Taiwan are different theaters favoring longer-range Class-2/3 drones; commercial counter-UAS on vessels is coming “in the fullness of time” but is too small a market today.
• Portfolio doctrine: pick one vital thing and do it better than anyone (Castelion in hypersonic missiles, Neros in FPV) over a portfolio approach — but Heaviside, which came out of stealth as it closed its Series A, is the deliberate exception: handheld RF sensors for the post-drone battlefield, loitering munitions, and submersibles. “We made an exception to our own rule because we found a team in Heaviside that’s an exception” — and he wouldn’t recommend the portfolio path to most defense-tech startups.
• A trend he flags but hasn’t played: sovereign new primes — “you probably will have a sovereign Anduril/Heaviside equivalent in every jurisdiction,” including a New Zealand defense startup. Cantos’ advantage stays in its backyard, with indirect exposure via Heaviside’s Norway office and ITAR-supervised allied sales.

14. Physical AI is the next ChatGPT moment — and the backlash will be worse

• His taxonomy: all technologies produce energy, information or movement. Energy has several portfolio companies (Radiant, Cache Energy, Arbor Energy, plus unannounced names); movement is the open frontier. “If there’s some ChatGPT-3-equivalent moment with robotics, that’s going to transform the world in a way we’ve never seen” — hedged as “possible, I’m not saying it will happen.” A cheap robot that could act as a full-time nanny and maid would be “absolutely transformative for society.”
• The warning he saw on a Mission District wall: Terminator posters with lasers for eyes and a protest QR code — “don’t build this.” “People are this mad about chatbots? Oh boy, you just wait.” Hollywood’s negative sci-fi has primed overcorrection, and ignoring the alienation “is a huge mistake, as the frontier labs have seen.”
• José’s question about faith and meaning in finance draws the most personal answer: “the act of creation is sort of divine in and of itself,” while the precautionary principle “is basically absolute conservatism… I can’t stand that.” Take risks, steward carefully, accept mistakes — “to not try would be almost sinful in my framework.”

15. The best founders are people you don’t quite want to hang out with

• Ian’s working hypothesis, which he attributes to another excellent investor as well: “if you find yourself wanting to hang out with a founder too much, they might not be good to invest in” — the very best are “kind of weird” and push things to an uncomfortable degree (Elon, Jobs), though Jensen Huang is an “understudied” counterpoint showing that kindness can coexist with greatness. He has still passed on founders he thought would succeed because he did not want to work with them for 10 years.
• On José’s psychedelics-and-introspection provocation, Ian won’t over-index on trauma: what he wants is antifragility — his wife, who competed at the Olympic marathon trials and placed in the top 75 in the country, and her elite runner friends are “a little bit crazy… they kind of enjoy the pain.” From bingeing David Senra’s Founders podcast: history’s great entrepreneurs consistently regret not spending enough time with their children — “I don’t know that that’s avoidable.”
• Succession as principle: “your obligation is to make sure that you are the worst GP that your firm has ever had.” When he becomes the weak link, Benchmark-style, “100% of the economics will go to my partners” while he stays on boards and helps fundraise — hopefully “20-plus years from now.”