Hans Tung

The through-line

The consistent idea across everything Hans Tung says is that speed of adjustment beats accuracy of prediction. He describes the habit that has served him best as "the ability to make decisions uh even when there's imperfect information available," because once you act you can watch the impact and correct: "if you make start making decisions and see the impact and then make adjustments you're in the driver's seat you can dictate the tempo you can control the flow you control the pace" [1]. He is explicit that this is not a claim to superior insight. "i don't think we're smarter than anyone else but the more adjustments we get to make we have better chance of succeeding" [1]. Framing a choice as a trade-off rather than a verdict is what makes it bearable: he says knowing he can adjust later "somehow release a lot of stress remove a lot of burden" and lets him commit and iterate [1].

The second thread is global range. He came to venture from a childhood absorbed in the histories of different regions and a teenage arrival in California where he first encountered "asian american" as a category, and he traces his style directly to that: someone "more global trying to bridge the gap and bring more different people uh together" [1]. That instinct shows up as an investment thesis, in a firm built across geographies and in a conviction that the best teams anywhere eventually travel [4].

On making decisions with imperfect information

He is unusually direct that perfectionism is the enemy of progress, and locates his own immunity in having been a founder twice: as a startup "you have a scrap scrape for everything," everything you ship is far from perfect, users give you mostly negative feedback early, and you win them back by improving fast [1]. The practical version is participatory: "getting stuff out getting people from the community iterate with them include them in the process so they feel they're part of your success" [1]. He extends the same logic to careers, arguing against buyer's remorse and in favour of momentum, with sports as the model, since the champions "don't let the weight of the previous game where this loss or bad play affect them they just move on to the next" [1]. He also insists no experience is wasted, that "everything you have learned up to this point is going to be useful to you," pointing to a decade of apparently lateral moves across eight or nine cities that became an asset once the world got more global [1].

On what product-market fit actually looks like

He distinguishes two things he underwrites: product-market fit and founder-product fit. Airbnb is his example of the first, a company he kept tracking even after it missed its numbers between rounds, backing it at growth stage in 2015 and standing behind it publicly during COVID on the argument that people would take staycations 150 to 200 miles from home and Airbnb could add inventory in the middle of nowhere faster than hotels could build [4]. He later described the same platform as having the most flexible supply chain in the world [1]. Max Levchin is his example of the second: a product person so suited to fintech that Tung retains confidence in him to build something new even when a firm is out of favour [4].

He treats pivots as evidence rather than failure. Probing Wispr Flow's move from a silent-speech wearable to software, he asked the question he thinks matters most, when to hold the original vision and when to make what he calls "in-game adjustments" as field feedback arrives [3]. And he watches for signals that a product's real market is wider than its category label: a TechCrunch reporter in India told him Wispr Flow was effectively giving people access to good business English, and Tung concluded "that secondary impact is actually could be as great if not greater than original value proposition" [3]. He had tested the product himself by speaking Mandarin homophones into it and found "the output in English caught that nuance and translated it correctly in different words" [3].

On building products people love enough to talk about

His consumer instinct is that intensity of affection in a small group beats mild approval in a large one. He cites Brian Chesky's philosophy as the standard: "let's not have a million people like you" but "focus on getting the first 10,000 or 100,000 people just love you" [3]. Word of mouth and community show up repeatedly as the growth mechanism worth engineering, whether in Wispr Flow's developer and founder evangelists [3], in Jam's use of developer community to scale to 150k builders [11], or in the brand recognition he noticed walking into fast food restaurants wearing a StockX t-shirt, where strangers would come up and talk about the platform [2]. He is drawn to marketplaces that solve the experience problem structurally: StockX sitting in the middle of the transaction to guarantee authentication, at eleven centres worldwide, is the case he draws out as genuinely innovative versus policing listings after the fact [2]. He also probes whether brands come to the platform to launch, and treats marketplaces as the next stage of brand distribution strategy alongside omnichannel and direct to consumer [2]. The longer-term version of this interest is the keyboard-less interface, making interacting with technology feel as natural as talking to a close friend [10].

On investing through cycles and managing exits

He was on record early in 2023 saying "I don't mind seeing down rounds," and returned to the upside of down rounds as the market reset [4][8]. His counsel to founders in that climate is survival and efficiency: "you need to be around to be relevant," manage cash carefully, and understand that "learn to be efficient and find the pain Point that's relevant today and to be solved and do that efficiently is the most sure way to build something lasting" [4]. He grounds this in portfolio arithmetic, noting that of roughly 150 North American companies, half had three years or more of runway and another quarter had 24 months or more, while several were doubling or tripling without heavy burn because they had real fit [4].

Exiting he calls the second hardest part of the job, harder than raising capital once you have been around a while. The temptation is the casino table: "things are looking good you want to go on the market is hot and then next thing you know Market cracks" [4]. His discipline is staged: take 10 to 20 percent off the table pre-IPO in a hot deal, then sell 5 to 20 percent of the position each quarter after lockup, never all of it if you still believe [4]. The third hardest thing is internal, convincing a skeptical partner, which he frames as absorbing the feedback while keeping the conviction that comes from being closest to the company [4]. On valuations he acknowledges that heat rotates from crypto to AI and that people now price in anticipation of unicorn status, but insists the durable question is whether a company fixes a big pain point for customers representative of a much larger TAM [4].

On where the value accrues in AI

He divides AI into three layers: the foundational layer of hardware and LLMs, dominated by big players; a middleware and devtools layer increasingly built on API models rather than SaaS, which his firm covers with a dedicated infrastructure team drawing on its open source and cloud experience; and the application layer, where his consumer background is most relevant and where he sees vertical opportunities in fintech, retail and e-commerce, construction, real estate, proptech and customer service [4]. Most VCs, emerging or established, will end up at the application layer, and he calls the whole shift "probably the biggest trade we'll see" over the next ten to twenty years [4]. His differentiated claim is that models will not become commodities, because durability comes from the feedback loops built around users rather than from having the best model; he points to Anthropic standing out early through its developer ecosystem, and to network effects that emerge inside AI systems [7]. Adjacent preoccupations include physical AI, prosumer behaviour, immigrant founders and the psychological traits required to build category-defining companies [7][12].

On investing globally and across emerging markets

He rejects the old view that China only copied, noting that SHEIN, Temu and TikTok changed that perception, and regrets that geopolitical tension ended the practice of taking US founders to China to see what was happening, "because a lot of learnings and competing makes everyone better" [4]. On Latin America he is bullish on fundamentals, a large population and growing middle class, a bigger economy than Southeast Asia, essentially two language markets, and sharply improved teams over five years, while naming the real constraints honestly: FX risk, and LPs burned in both Latin America and India who now need a few Nubank-style outcomes to come back [4]. The pattern he trusts is patience through the cold part of the cycle, since Nubank itself was funded when the region was out of fashion [4]. Notable Capital's footprint reflects the same logic, investing in the US alongside Europe, Israel and Latin America [8].

On how a venture firm earns its way in

"i believe that all battles are won before they're ever fought" is how he explains firm design [4]. Three assets do the work: global data points on the same industry across multiple geographies, which is why he joined a firm already spanning Singapore, Silicon Valley and China in 2013; a platform team built since 2016 covering BD, marketing, talent, recruiting, people and cultural management, running a founders and leaders program that companies send VPs and directors to, not just C-suite; and organisation by industry group across consumer, fintech, enterprise infrastructure and cyber security so partners can be genuinely relevant [4]. The proof he cares about is referenceability: "the founders that we have helped before become the best reference points for us," which he says wins roughly eight out of ten contested deals [4]. Internally he is proudest of the compounding, investment, platform, data and finance teams working in sync, and assets under management going from a bit over a billion when he joined to just under ten billion, "the second decade ends up being like 5 6x the first decade" [1].

On backing founders through the hard parts

Having been a founder twice across three or four years, he speaks about the loneliness of it, the worries that keep you up at night and the need to face a hundred people the next morning with an inspirational speech [1]. That produces a deliberately restrained style of support. The mistake he sees ex-operator VCs make is trying to fix things: "you can't fix the issues for you for for your founder everyone has to learn at their own pace," so his job is encouragement, sequencing and patience while founders find what works for them [1]. He accepts that this is sometimes painful to watch, and argues the payoff only exists if you stay: "unless you are patient and you're willing to stick to stand by them when things are tough" they never get the chance to show you what they can do [1]. That patience is what he credits for the Airbnb outcome [1][4].

On food tech, B2B and the next decade of commerce

His food tech thesis, running since 2016, rests on three areas: supply chain, distribution and especially last-mile delivery; food waste management through AI software and data mining; and upstream production such as vertical farming and alternative proteins in Israel and Singapore, all framed against a world population heading to 10 billion [5]. The cross-cutting trends he names are that "efficiency is the name of the game," that "localization plays a very important role in serving the community effectively," and that delivery remains a huge opportunity despite DoorDash and Uber Eats, a conviction reinforced by watching Meituan become a super app and by the Grab investment in Southeast Asia [5]. Most pointedly, after a portfolio of consumer names including Wish, Poshmark, StockX, Ibotta and Peloton, he argues "b2b could be even bigger over the next decade" [5]. With Bowery Farming he pushes on what makes indoor farming defensible, drawing out that LEDs made it viable but not scalable, and that computer vision, AI, sensors, robotics and a proprietary operating system are what turn it into a network of farms that get better as they multiply [5].

On identity, speaking up, and there being no single path

After the Atlanta spa shootings he and colleagues put a matching pledge on Twitter and LinkedIn to fund community groups fighting anti-Asian hate, expecting little and instead helping catalyse a five million dollar raise with co-leads across the venture community [1]. The lesson he takes is about visibility rather than charity: unless people speak up, others do not know the significance of the cause, and "the more inclusive the stronger we are and the more vocal we are the more people understand why we feel the way we do" [1]. He wants that inclusiveness to run across East Asian, Southeast Asian, Indian, Pakistani and Pacific Islander communities rather than fragment, and he notes with concern the online instinct to differentiate rather than stand together [1]. For younger people he pushes back on the achievement script he was raised with: "there is no one path to success and there's no one path that fits at all," so the task is figuring out the right fit for yourself and having the courage to resist typecasting [1]. He also treats small first steps as generative, pointing out that a podcast nobody expected anyone to listen to multiplied into several shows and vlogs [1].

Takeaways

• Act on imperfect information and correct as you go: "the more adjustments we get to make we have better chance of succeeding," and framing a choice as a trade-off removes the paralysing weight from it [1].
• In a hard market, survival is strategy: "you need to be around to be relevant," and companies with genuine product-market fit can still double or triple without heavy burn [4].
• Sell in stages, not in one decision: roughly 10 to 20 percent pre-IPO in a hot deal, then 5 to 20 percent of the position per quarter after lockup, keeping exposure if you still believe [4].
• Underwrite either product-market fit or founder-product fit, using Airbnb and Max Levchin as the respective templates [4].
• Optimise for a small group who love the product rather than a large group who like it, and watch the second-order effects, which can exceed the original value proposition [3].
• Models will not commoditise into irrelevance; enduring AI companies are built on feedback loops and ecosystems around users, as Anthropic showed early with developers [7].
• Do not try to solve founders' problems for them, because "everyone has to learn at their own pace"; the value of a patient investor shows up exactly when things are worst [1].
• B2B may be a bigger opportunity than B2C over the next decade, with efficiency and localization as the operative trends in food tech and commerce [5].