Most VCs Are Afraid to Be Wrong | Rohan Pujara
Rohan Pujara avoids seed categories where consensus capital creates inflated entry prices, instant copycats, and structurally weak margins. He separates seed from growth: following momentum can make sense at Series A and beyond, but at formation it puts startups into a “knife fight” with ten competitors before they have a foothold. Agents, infrastructure models, neolabs, data centers, energy, chips, and defense currently look crowded to him; Valhalla instead searches their neglected edges.
Pujara thinks venture capital’s professionalization has replaced risk-takers with careerists who fear looking wrong more than losing money. His phone contains only 21 people who have personally generated more than $1 billion in DPI, remarkably few against 50 years of venture activity. His conclusion is deliberately harsh: convexity gives many investors one lucky success, while repeatability exposes how few are genuinely good.
Valhalla underwrites distinctive people before it underwrites fashionable markets. Pujara wants founders with independent histories, courage, trustworthiness, social agility, and visible energy—not prestigious degrees—and treats polarized references as stronger evidence than lukewarm approval. “You should leave a lasting impression on anyone who interacts with you”; a founder everyone merely likes is often less interesting.
Valhalla’s concentration model turns this philosophy into an unusually focused portfolio. From an $80 million–$100 million fund, Pujara wants only two to four exceptional new investments annually and can write a $10 million seed check; if one becomes an unmistakable fund-returner, he plans to allocate 20%–30% of the fund. The opposing statistical case—roughly 30 checks to improve the odds of catching one or two outliers—remains José’s strongest challenge.
K2 Space and Jaza are the clearest demonstrations of Valhalla’s non-consensus playbook. K2 bet on large, power-rich satellites while the market copied small-satellite successes; three and a half years later, Pujara says it still has no real competitor, has signed more than $1 billion of business, and is raising at a $6.8 billion valuation. Jaza spent 2017–2023 finding a workable electricity model in sub-Saharan Africa, then raised $100 million and grew roughly tenfold year over year.
Aftersort[?] and Truffle show how Pujara approaches technical bets without pretending to possess perfect technical foresight. Aftersort[?] offered a principled answer to a real LLM limitation; Truffle wagers that token demand will exceed centralized AI infrastructure capacity, perhaps visibly by early 2027 and, in the worst case, 2028. Its personal AI device, custom weights, supply-chain work, and three-year lead make it “probably the perfect Valhalla investment.”
Pujara ties his risk tolerance to having repeatedly discovered that financial ruin was survivable, while tying Valhalla’s future to aligned incentives. Childhood scarcity, seven or eight moves, leveraged crypto losses, and taking a delivery job with $0 in the bank weakened the instinct to protect status: “Don’t gamble,” but do take underwritten risk. His institutional ambition is a flexible private-markets firm where managers and LPs have their own capital exposed, because “incentives rule the world.”
1. Consensus destroys seed economics before it validates the market
Pujara’s objection is not to large markets but to categories where every fund reaches the same conclusion simultaneously. Once consensus forms, startups confront copycats, Meta-scale incumbents, expensive entry valuations, and a “knife fight” that can drive profit margins toward zero from day one.
José’s Cursor challenge—worth keeping—is that crowded, obviously enormous markets can still produce extraordinary winners. Pujara’s answer is stage-specific: early Cursor was not yet a consensus coding-agent bet, while Series A and growth investing are different games where following demonstrated momentum can make sense.
Responding to Ian from Cantos’s argument that private information makes “consensus” a public-markets concept, Pujara says relevant information still circulates efficiently among Tier 1 funds and emerging managers through group chats and shared deal flow. Outside that network, information fragments: most Tier 1 firms would not even interview an African energy company like Jaza.
Geography reinforces the effect. San Francisco is, in Pujara’s words, “the home of consensus”: the same language, parties, investors, and companies. He prefers New York as a meeting hub and sees LA—especially around SpaceX, Anduril, K2, and El Segundo—as a growing refuge for ambitious people who want the West Coast without its dominant intellectual bubble.
2. Venture professionalization rewards reputational safety over risk
Pujara calls the change the “MBA-ification of venture capital,” though he does not limit it to MBA holders. As VC became prestigious, it attracted highly credentialed, status-conscious careerists who want to be right with everyone else and are especially afraid of being wrong when the group is right.
Earlier generations included investors such as Michael Moritz, Vinod Khosla, and Fred Wilson, who arrived through less standardized paths. Pujara doubts that many people who prepare for venture from undergraduate school are natural risk-takers—or even investors capable of operating in another asset class.
His private scorecard contains only 21 people who have produced more than $1 billion of DPI through personal investments. Against the number of people practicing venture over five decades, that scarcity suggests convex returns can make someone lucky once; consistently repeating that success is the rarer skill.
José supplied the starkest crypto contrast: extremely intelligent people made no money while less conventionally brilliant participants took risks confidently at the decisive moment. Pujara’s own self-description is intentionally deflationary—“on the left side of a normal distribution”—because he thinks avoiding elaborate forecasts helps him focus on people and durable businesses.
3. Proprietary sourcing begins with unusual people, not fundraising lists
Valhalla cultivates people who are “a little controversial” or polarizing, studies why they see the world differently, and remains close to them long before a financing. When one starts a company or says, “Hey, check this out,” the relationship produces access without monitoring every active seed round.
That approach often reaches founders who are not fundraising. In unconventional fields, having one investor genuinely believe them can matter more than manufacturing competition; Valhalla has rarely competed across multiple term sheets and may interpret an already crowded process as evidence that the opportunity no longer fits.
Pujara does not require outrageous public behavior. He wants someone who stands apart, thinks from first principles, chose an uncommon path, and will express a genuine view despite reputational pressure. A reference saying “I don’t like that person” or “I disagree with them” may be informative; indifference is worse.
His adverse-selection heuristic is similarly blunt: if someone contacts Rohan at Valhalla, that can be a warning. He would rather identify the person through research, understand why the idea belongs in Valhalla’s opportunity set, and initiate the conversation himself.
4. Courage, trust, and agency matter more than pedigree
The first underwriting question is whether the founder has taken real risks when prevailing advice felt wrong. The second is trust: Valhalla expects to own companies for 10 or 15 years, not sell magically after two, so Pujara will not enter that relationship with someone he cannot trust.
Intelligence remains necessary, but résumé proxies do not. Pujara often does not know whether a founder attended university, much less their grades; he values quick reactions, social navigation, proactivity, and the energy of someone visibly pursuing an objective more than raw IQ.
Motivation is the diagnostic center: “Why are you doing this?” A founder who noticed AI was fashionable after graduation differs from someone whose history leads naturally to the problem. Pujara argues that raising $5 million–$10 million has made startup formation a relatively safe career path for technically connected people.
Reference work tests specific suspicions rather than checking a box. If a compelling founder seems careless with detail or too excitable, Pujara finds people positioned to examine that trait. The desired founder should have left a clear impression—positive or negative—on almost everyone who worked closely with them.
5. Concentration forces conviction but raises the cost of being wrong
José discloses that Delphi is an LP in Valhalla’s second fund. Fund I began with a plan for 30–40 companies, was revised to 20, and ended with roughly 20; Pujara’s concentration model is only two to four exceptional opportunities per year.
With an $80 million–$100 million vehicle, Valhalla can lead a seed round with a $10 million check. It does not maintain systematic reserves for every Series A or B; Pujara instead wants the freedom to identify “our K2 Space,” then direct 20%–30% of the fund toward that evident return-driver.
José describes concentration as a “forced-conviction function,” but presents Michael Dempsey’s opposing arithmetic: about 30 checks may be required to include one or two major winners. A ten-company first-time fund can simply miss every outlier, leaving no engine for returns.
Pujara says selective follow-ons can be done in several ways, including the SPVs José raises, but does not call them reserves: his funds do not maintain reserves during the operating period. José warns against automatic doubling after six months of momentum, since private marks can rise vertically before a company goes to zero or is sold as preferred stock; his hypothetical obvious case is discovering the next Elon Musk.
6. Valhalla stays generalist while searching for future monopolies
Valhalla’s 2021 website emphasized digital assets, metaverse games, and psychedelic therapy. Pujara says he met the Valhalla partners at 20 and eventually formed a partnership with them in 2020; he spent roughly 2020 through late 2022 observing, building relationships, and developing judgment before gaining their trust to devote more attention to physical technology.
His reasoning preceded ChatGPT: software looked likely to commoditize because nearly every smart young person he knew studied software engineering, while mechanical, electrical, and aerospace engineering were neglected outside SpaceX. By 2023–2025 Valhalla made several physical-technology bets—but Pujara now thinks that field itself is becoming consensus.
That evolution made him reject sector boxes. LPs prefer legible mandates because they simplify allocation, yet Pujara wants carry rather than a management-fee business; being forced to buy defense during “the hottest defense market in history” would sacrifice the flexibility required to outperform.
His model is TCI, which he describes as Chris Horn with perhaps three or four other members or analysts managing nearly $100 billion across roughly nine to 12 public monopolies. Its task is simply to decide whether each moat endures. Valhalla’s translation is to find future monopolies at seed, support them with conviction, and hold them for the long term.
7. K2 Space inverted the small-satellite consensus
K2 was not raising when Pujara visited its facility. After several hours with the team, observing its ability to recruit strong SpaceX talent and checking the ecosystem, he asked to invest; the founders considered it for roughly a day and accepted.
The thesis began with a mismatch: Starship and broader launch capacity promised rapidly increasing orbital mass and power, while venture-backed companies kept imposing tighter small-satellite constraints. Pujara saw that pattern-matching Planet Labs and Starlink ignored Starlink’s vertically integrated launch advantage and the limited economics available to imitators.
K2 instead built much larger, more powerful satellites for a five-to-ten-year future in which launch constraints loosen and customers demand greater mass and power. Pujara says it still lacks a real competitor three and a half years later, has signed a deal worth more than $1 billion, and is raising at a $6.8 billion valuation.
8. Jaza proved that a hated market can conceal working economics
Corey from Fundomo introduced Pujara to Jaza; Rohan believes Corey was an angel investor. Founder Jeff’s path—growing up without affluence in northern Canada, forging his own route, operating in Africa, and eventually confronting household electrification—immediately marked him as unusually energetic, serious, and relentless.
More than one billion people in sub-Saharan Africa lacked home electricity, yet Jaza’s answer required years of iteration, beginning around motorcycle batteries. Between 2017 and 2023 it navigated governments, local politics, violence, communication barriers, and other execution hazards before discovering a small formula that actually worked.
José framed Africa as a venture “graveyard”; the nine-month diligence reflected that history. Valhalla interviewed investors in similar failed businesses and repeatedly tested whether it had missed a fatal flaw. It concluded that others’ losses explained the aversion better than Jaza’s economics; Jaza subsequently raised $100 million and, Pujara says, grew roughly tenfold year over year.
Fund I’s failures sharpened the hierarchy. Attractive markets paired with founders rated seven or eight out of ten generally returned little; now differentiation earns the meeting, an exceptional founder earns the work, and a “20 out of 10” founder may justify tolerating more business uncertainty—especially at a $10 million rather than $100 million post-money valuation.
9. Technical underwriting needs a bounded possibility test, not omniscience
José presses Pujara on whether a non-engineer outside frontier labs risks buying technically indistinguishable “call options.” Pujara concedes the concern—it contributed to passing on Oliks[?]—but argues that investors can distinguish a plausible engineering program from open-ended scientific experimentation without understanding every nuance ten layers down.
Aftersort[?] cleared that bar because Pujara saw a real LLM gap and a rational, principled approach to closing it. He did not need other VCs’ approval: the combination of a missing capability, a reasoned solution, and an exceptional thinker made the bet acceptable.
Truffle’s thesis is that token demand will eventually exceed feasible AI infrastructure investment. Pujara expects the constraint to become visible around the beginning of 2027, or 2028 in the worst case, against consensus forecasts of $4 trillion–$5 trillion of AI capex in 2028; power, supply chains, and centralized inference costs become the bottleneck.
Srikanth’s analogy is the migration from mainframes and terminals to personal computers: Pujara expects people eventually to have personal computing devices, potentially with large local memory, a dedicated operating system, and continuous learning. Truffle has its own device and custom weights; as Instinct and Muse degrade under costly 24/7 frontier inference, its three-year supply-chain lead makes it “probably the perfect Valhalla investment.”
10. Personal downside tolerance and aligned incentives complete the strategy
Pujara grew up in a low-income household supported financially by his mother, moved seven or eight times, and always knew his parents’ bank balances. Repeatedly entering new cities without social context trained him to adapt, while early scarcity made earning money a practical concern rather than an abstraction.
Leveraged crypto trading later wiped him out multiple times. At college he reached $0, took a food-delivery job, and learned that embarrassment mattered less than eating; the durable lesson was “Don’t gamble,” alongside the realization that a satisfying life requires far less money than status-protection implies.
Work became his social world because it connected him with the first people he deeply admired. He describes founders and investors as friends, talks with portfolio founders nearly every day, and wants a life built around people who “energize you” and “ignite your ambition.”
Over five to ten years, he wants Valhalla to recruit similarly independent investors and pursue new private-market alpha. His diagnosis is that institutional LP structures reward AUM, management fees, and not looking bad more than performance; the antidote is personal capital at risk throughout the chain. Valhalla says it has funded at least 50% of the fund with its own money. “If you truly understand incentives, you can understand how the world works very quickly.”
Full transcript
This list on my cell phone is a list of people who have generated over $1 billion in DPI (Deep Packet Investment) at least once through personal investments. There are only 21 people on that list so far. In other words, it’s a surprisingly small number compared with the number of people who have invested in venture capital over the past 50 years. I think it indicates that many people are lucky only once in their lives.
San Francisco is the home of consensus. If the founders are in San Francisco, everyone will be desperate to meet them. There’s also a kind of intellectual bubble: everyone has the same views, everyone says the same words, everyone goes to the same parties, and everyone follows the same investors. All investors are following the same companies, and that’s far too much. I hate it. I really hate it. I hate being there, and I hate listening to people talk there.
Rohan, you joined Valhalla when you were 20 years old, right? In 2022, they launched the fund during one of the most difficult periods for fundraising, following the collapse of the tech stock bubble. To be honest, Fund 1 was a great success.
Earlier, I received the latest information on K2 Space, Jaza, and many other funds. I’ll explain in more detail now.
It was truly a wonderful achievement. Our partner, Davin, deserves a huge amount of credit for raising funds during that period. It was by no means easy, but he’s a very hardworking person.
What I like about you is that you’re a truly unique investor. You have a truly original perspective, which is quite rare these days. You also said that my podcast isn’t exciting enough, so I’m looking forward to hearing your inspiring stories on this podcast. I’m really glad you came.
I’m looking forward to it.
1. Why Rohan Avoids Consensus Investments
I’d like to begin with the method I’ve devised. Let’s take a look at some of Rohan’s ideas, which seem to differ from those of other investors, one by one.
First of all, I consider you to be a contrarian investor. In our initial meeting, you mentioned that you were tracking what YC and other major funds were looking for and using that as a list of companies to exclude. In other words, you wouldn’t have meetings with those companies. Why?
I want to make it clear that we do have meetings with all kinds of companies. The underlying idea is that there’s a dynamic in which a consensus is formed around certain areas, especially in venture capital and particularly at the early stages.
Let’s take Instinct as an example. Once a category is formed, all consensus investors immediately follow that category. If a group of investors misses out on it, they almost have to create their own competitors. Existing companies such as Meta, OpenAI, and Anthropic are also entering that field. Therefore, fierce competition unfolds from day 1, and early-stage startups end up wasting all their funding competing with one another from the very first day. That isn’t ideal, is it? It’s simply a way to waste a lot of money.
I believe the cause lies in the combination of incentive issues between LPs and GPs and this short-term focus. I also think that the dramatic change in the quality of talent in venture capital is one of the contributing factors.
In my opinion, while many people in the VC industry today are highly intelligent individuals with very high IQs, those who were doing this job 20 or 30 years ago were people who weren’t afraid of taking risks. As time went on, this job became a kind of status symbol, attracting people with top-tier qualifications from Harvard, MIT, and other prestigious institutions. They tend to be either status-conscious or simply risk-averse. They want to do the right thing with everyone else, and they’re afraid of being wrong when everyone else is right. In other words, the fear of making mistakes influences all decision-making and generates consensus.
That’s how I explain this problem very concisely. Basically, what I want to do is find areas where there’s no consensus. In other words, my company isn’t engaging in a knife fight with each other from day one.
However, don’t you think the areas that attract that capital are instinctively very large TAMs and that, to some extent, they’ll produce the greatest results? Just like with coding, that could have been said for any round of Cursor, right? Cursor itself has entered a somewhat congested area, hasn’t it? There were OpenAI and Anthropic, but ultimately it was clearly a massive achievement. How do you distinguish between them?
Well, you can’t agree at the initial stage, right? I don’t think Cursor, in its early stages, was even a coding agent yet. In other words, it was never a gamble on consensus.
Everything I’m saying is specific to the initial stages. Even at this stage, Series A is basically still in the growth phase, right? Series A and later are completely different games.
I agree with you on that point. In most cases, it’s better to pursue consensus, isn’t it? With growth investments, you don’t want to do anything too unconventional or try to go against the trend, right?
I think that would be very difficult, but the problem with doing this in the initial stages is that you would have to pay a very high entry price. Then there’s the issue of competition, isn’t there? That means everyone’s profit margin will be zero from day 1, right? When you’re competing with 10 other startups from day 1, it’s incredibly difficult to build a strong foothold.
One of the reasons Cursor was successful, if you look at Anthropic, is that Anthropic’s first funding round was extremely difficult. They had to raise funds from many wealthy individuals, which is great, but it wasn’t a consensus VC investment, was it? Thanks to that, they were actually able to succeed and grow significantly. In other words, I think it’s not a TAM issue but simply a matter of consensus.
Therefore, categories with extremely high consensus should be avoided. There aren’t that many categories like that, are there? What category do you think this is in right now?
Currently, agents, infrastructure model companies, neolabs, data center companies, data center-related companies such as energy and chips, and defense are all areas of complete consensus. However, I wouldn’t say I’m avoiding all AI companies, even when it comes to AI itself.
We’ve invested in companies on the periphery of the AI field that are doing something unique and original. I think it will take a few more months for that to become interesting, or even for a consensus to be reached. That’s what we’re looking for and want to invest in.
That makes sense. We should also mention that we’re LPs in your second fund. We’re very happy to be LPs, so we’re happy to have you. For disclosure, in what sense are those things a consensus? What is consensus in venture capital investment?
I spoke with Ian from Cantos last week, and what he said was interesting. The concepts of consensus and non-consensus originated in stock-market investing by people like Howard Marks, Warren Buffett, and Seth Klarman, but they don’t apply to venture capital investing.
In the stock market, you can get complete information, and most people have access to the same information. Therefore, in order to obtain alpha, there must be a lack of consensus by definition. However, in venture capital, only a very small number of companies have access to the information, so there is no consensus.
When you look at a particular transaction, it’s like arbitrage in venture capital, where, in a sense, the consensus of a few people becomes a larger consensus, just as it does during an IPO.
So why do you think it’s so important?
I think what’s really important is the availability of information to potential investor groups. Your competitors are all funds, regardless of the category they fall into, including Tier 1 VC funds and emerging managers.
Usually, in consensus-based transactions, information is shared almost completely among these groups, right? First, you go to Tier 1, then, as a backup, you go to all the emerging managers for the seed round, and then you watch the market settle down naturally. Information is shared quite efficiently in that market, isn’t it?
There are lots of group chats where people share information about alpha, all the deals and agreements happening, who’s starting which company in which category, and so on. However, once you leave that area, the situation changes completely.
For example, let’s take Jaza. Jaza is building an energy business in sub-Saharan Africa. Even if we introduced Jaza to every Tier 1 company we know today, most of them wouldn’t even bother to take a meeting because Jaza is far removed from what they’re interested in.
Even in that case, we can imagine opportunities that can’t be priced by private capital around the world, right? Shouldn’t we consider it in relation to time and opportunity, rather than focusing on how many people are paying attention?
Yes. In other words, I think there are a lot of TAMs. I don’t think a consensus can be formed solely through TAM (Terms of Matter). I don’t think consensus will be formed by people watching something work out. That’s pattern matching.
In retrospect, it was quite obvious that there was an agent inside the cell phone. That’s exactly what happens with LLMs. But someone had to do it.
Intuition was necessary for others to quickly follow suit. As soon as they successfully launched their product, existing companies like Meta entered the market almost instantly to compete with them. Then I saw this consensus being formed. Therefore, investing in Instinct when it was a seed company was a good gamble.
2. Has Venture Capital Lost Its Risk Takers?
You were also saying what I thought was right. That means there aren't many people willing to take risks in venture capital. This is strange because venture capital is literally a risk-taking business. Why do you think that is, and how do you think that will manifest itself?
I think this is something like the MBA-ification of venture capital, although it's not necessarily limited to MBA holders. Many people are preparing a career path to join venture capital firms.
You've been preparing to become a venture capitalist since your university days, right?
Wear a vest.
Yes, exactly. It looks like someone is wearing a vest for Halloween or something.
In other words, I think it's because venture capitalists became objects of admiration for legitimate reasons. I think it's a truly wonderful, enjoyable, and stimulating job. As you know, it became something that many people aspired to.
And as these companies' teams start to grow in size, their operating methods also expand. Since they receive all of these management fees, they have to use them, right?
The fund has grown so large that it's probably too big to generate the same returns as before. As you know, it's being driven by 2 and 20, but that has its drawbacks and problems. I can explain that later. I think that's mainly driven by LPs.
In any case, when these people come in and take on these roles, I don't think that, if they are risk-takers, they start preparing to be risk-takers from their first day of undergraduate studies. In other words, you need to have a strong desire to become an investor, right?
I don't really agree with the idea that venture capital is a special form of investment. I believe that investors should be able to operate in a variety of asset classes to some extent. Furthermore, I suspect that the people who have recently entered the market are not true investors. When you talk to them, you get the feeling that it's hard to imagine them operating in any other asset class.
What do you think defines a good investor, or defines the character of an investor? You're the complete opposite of someone who seeks the truth, thinks clearly, and experiences FOMO when it comes to investing, right? A good investor should never invest while feeling FOMO. That makes absolutely no sense, right?
I think the only true form of investing is, to some extent, value investing. What we're trying to do could be described as a kind of value investing.
Right. What we're trying to do could be described as a kind of value investing.
Not exactly, but yes, that's what I think.
3. Investing Skill vs. Getting Lucky
And when people start leaving, I think career-oriented people will start creating some kind of norms about how venture capital should be done.
In short, I think it was a lot of things that ultimately changed the way you did things.
That's right. Many great investors didn't enter venture capital through a straightforward route, did they? Michael Moritz, Vinod Khosla, Fred Wilson, and others. All of these people came from unusual backgrounds to get to where they are today.
I don't know if I've mentioned this list before, but I have a list on my phone of people who have made over $1 billion in DPI (Deep Packing Investment) at least once through personal investments. There are only 21 people on that list so far.
That's amazing, isn't it? That's a surprisingly small number compared with the number of people who have invested in venture capital over the past 50 years.
I think that shows that many people experience good luck at least once. Many people make good investments, because if you invest in enough companies, you're lucky enough to make a good investment. That's because these investments are very convex.
However, I think very few people would actually do that repeatedly.
That's because there are very few good investors. And that applies to almost every asset class, not just venture capital. In fact, we've seen that a lot with cryptocurrencies.
There are so many incredibly intelligent people, but they didn't make any money at all. Then there were those who weren't particularly intelligent but had an exceptional ability to take risks and were confident when it mattered.
I'm curious to know to what extent contrarian investing is acceptable. The best investors I've ever met are those who have made profits from extremely contrarian investments that no one else dares to touch. I've also been investing little by little with confidence in non-contrarian investments.
Among those we admire, there are some, like Corey from FoundersFund, who have executed this strategy quite successfully. I think he's managed to do both.
I think you definitely need a certain skill set to do that kind of thing. I need to know what I'm good at, and I'm not good at that. Furthermore, as I've said before, I don't consider myself very intelligent. I consider myself to be on the left side of a normal distribution.
I'm not that smart, so I don't overthink things. I try not to think that I know what will happen in the future. I think what we're really good at is finding people who are a little controversial, a little polarizing, or a little bit of a crowd-pleaser.
It's about understanding why they think that way and why they see the world from a different perspective. Then you have to determine whether you think this person has the necessary qualities and whether you think what they're saying is correct. If those 2 conditions are met, you'll make a seed investment, right?
That should be all there is to it. It's not my area of expertise to tell you which companies will rise in value from $2 billion to $10 billion, or even just reach $2 billion. I have absolutely no idea. I don't have that intuition.
I probably don't have enough social connections to fully understand this, but at least from what I can see, many of these companies that are rising in value like this are not truly good, sustainable businesses that you would want to hold onto 10 years from now. That's due to other issues, such as the excessive competition and fierce rivalry you mentioned earlier.
I think that's the core of Valhalla's investment approach: We choose every company we invest in that we would be extremely excited to own in 10 years, no matter what happens in the market.
My intuition tells me that very few investors can say that with such certainty about their own portfolios.
That's interesting.
4. Finding Founders Before They Raise
What are you looking for? I also find it interesting that, as you did, Fund I managed to bring about funding rounds for many of its top companies when they weren't raising capital. You said you really liked that.
Clearly, this is because there's a place where the most asymmetric information exists. Could you tell me a little about that? Could you also tell us how you find and implement those ideas, and discuss each of them individually?
The way we look for things is that we're always looking for people who are a little bit controversial. When we find people like that, we truly cherish those relationships and make them valuable. I spend a lot of time with them. Those kinds of people eventually become my friends, the kind of people I want to keep around me.
When they start a company or tell us, “Hey, check this out,” we go and see it, delve deeper, and get involved. Therefore, I'm not necessarily looking for things like, “Who is raising funds? Who is raising funds in the seed round?” That's not that important to me.
It's not always the case that we're the ones who bring about funding rounds or get ahead of the curve. Sometimes we happen to come across someone who is raising funds, and we're happy to invest in them. But in many cases, especially if you're working in a somewhat unconventional field, fundraising can be difficult, so many people aren't always open to accepting funding, right?
If someone believes in them, they're happy that there's someone who believes in them, because there's no consensus on their views. Another good thing is that you don't need to prove yourself, right? We have rarely found ourselves in a situation where we had to submit term sheets that competed with other companies.
If you saw other companies submitting term sheets today, you'd probably think, “Okay, this isn't for us.” And there is definitely something there.
5. What Makes an Exceptional Founder?
You mentioned controversial figures, but what specifically makes them particularly attractive to investors? Look at excellent entrepreneurs, entrepreneurs who have achieved great success. In the early stages of their businesses, it's highly likely that the entrepreneur, in some way, set themselves apart from their competitors, right?
Yes, that's what we're looking for. It doesn't necessarily have to be controversial. You don't need to say anything outrageous on Twitter, but you do need to stand out a little, right?
We're looking for standout figures, but I think people who invest in venture capital tend to be a little intimidated by them. I'm a little intimidated by people whose opinions differ drastically from those of my peers. When we call someone for a reference check, that type of person might say, “Oh, I don't like that person.” Or it could be something like, “I don't agree with them.”
In other words, we're looking for those kinds of signs.
There's no strict, clear definition of what makes someone stand out. It's more of an intuitive feeling. I believe that's one of the main characteristics we're looking for. I'm someone who thinks in terms of first principles.
You can see it throughout their lives. They think for themselves, and they haven't chosen the most common path. We don't need people like Alex Jones—people who say crazy things. Those are people from a different category, but we need people who aren't afraid to express their feelings.
There are so many people here today. I'm no longer talking about founders or VCs. Many people are very afraid to express their feelings, aren't they? That's because everything is public now, so they are very cautious and concerned about public perception and their status. Therefore, I think it's very rare to find a genuine person—someone who will express their feelings. That's why we're trying to find people like that.
That's wonderful. What other characteristics make up the founders of Valhalla?
What we immediately try to understand is that person's history of taking risks. It's about being brave, isn't it? In other words, it's about doing something when you feel that what other people are saying isn't necessarily right, or when you feel it's not right. It's a contrarian move. I think that's what we're looking for. That's the first one.
I think trust is another important factor. Especially in the current situation, I think there are many people who lie without hesitation. I don't know if it's always been like this, but especially now, it feels like a lot of people are lying throughout San Francisco's technology ecosystem. Therefore, I believe that building a relationship of trust is truly important when investing in someone.
You don't invest because you expect the stock price to skyrocket in 2 years and then you'll be able to sell it like magic. You should invest with a long-term perspective. You definitely wouldn't want to work with an entrepreneur you can't trust for 10 or 15 years, right? Therefore, I think trust and the courage to take risks are probably the 2 biggest factors.
The description of the founder mentioned that he was contrarian, risk-taking, and trustworthy, but there was no mention of intelligence—that is, speed of thought, or simply intelligence. How do you define that? That seems like the obvious thing to do.
That's necessary, isn't it? The degree varies depending on the business, but you always need agile people. In other words, they need to be good at getting along with people. I place far more importance on being good at navigating social situations.
That's why most of our founders don't even know which universities they went to. Literally, I couldn't tell you. Whether they went to university or what their grades were like has absolutely nothing to do with me. This might be a prejudice, but I had poor grades in school, so that's something I'm concerned about.
In other words, it comes down to whether those people are quick-witted, whether they can react quickly, and whether they are truly proactive. You can tell if someone is actively pursuing something. You can sense the energy. Whether or not you have that energy is far more important than your raw IQ.
However, it's clear that our founders generally have high IQs. Nobody thinks that's foolish.
So, how do you find this out? When you meet with them, do you talk about their past and their lives? Do you have questions you always ask, or do you just go with the flow?
I try to understand their background and then try to understand why they are doing what they are doing. That's probably the most important thing, isn't it?
What's interesting is what prompted you to start this. It's something like, “After graduating from university, AI was trending, so I decided to start an AI company.”
I think many people, especially recently, are opportunistically starting companies because it's an era where you can raise $5 million and launch any AI startup if you have an IQ as low as room temperature. In other words, I think many people are entering this industry opportunistically.
In fact, in many cases, especially if you're in Silicon Valley and part of the tech social circle, starting a company is actually quite low-risk. Even if it fails, you can always raise $5 million or $10 million, and if it fails, you might be able to raise even more. Then there's Y Combinator and many other talent pipelines, right?
Becoming a founder is now considered a career path. This is especially true if it belongs to a consensus category.
This reminds us of what happened to career paths in consulting, banking, and other fields. Those became career paths. Therefore, I think there are many categories that are being overlooked, but I believe they have the potential to become $50 billion companies.
Well, we have a lot of handguns for sale [?]. I don't think you even need to think about it, but there are plenty of such categories that I've found.
If I talk to the founder and investigate the background, how long will the process take? How quickly will it be decided?
I think investors do this in different ways. Some people rely on their intuition and can tell pretty quickly. Others, like Michael Dempsey, take more time and are more precise in their approach. I'm interested in how you process it and what you do to verify whether you truly understand it.
For me, as you move beyond the founder, the most important thing is that there are no competitors with what they have. In my opinion, truly good seed investments should have no competitors at the time of investment. Ideally, it should be quite difficult, and it should take time before any competitors emerge.
For example, take a look at K2 Space, a company in which we have invested. They took a very contrarian bet on large satellites while everyone else in the market was heading toward small satellites.
How did you know that team was great?
I spent time with them.
How long was it?
A few hours at their facility. These are good people. They were attracting top talent from SpaceX, and that was difficult. Then I thought about what they were doing.
How did you know they were the best talent from SpaceX?
I was just checking. In such an ecosystem, it's very easy to point a finger at a person and judge whether they are good or bad. So, I believe your question was, “How long?” That depends on my situation in that field.
At the time we invested, we knew that K2 had no competitors. In fact, even now, 3 and a half years later, it's safe to say that K2 has no real competitors. They are currently raising funds at a valuation of $6.8 billion. There are no competitors. Isn't that wonderful?
Those small satellite companies are becoming increasingly irrelevant these days. I think that's what we're concerned about. It really depends on the circumstances we face.
I like letters of recommendation. In short, I like letters of recommendation. Because I generally think that letters of recommendation, both positive and negative, are a good sign. Furthermore, even if you're talking to someone and you like them, there might be something suspicious about them, such as not paying attention to details or being a little easily excitable.
I will try to identify recommenders who might know about this. I think you can learn a lot from recommendation letters just by talking to people about the recommenders. For the type of person we're looking for, their reputation should be clear. You should leave a lasting impression on anyone who interacts with you. Whether that is the case or not should be very clear.
If, when I talk to that person, it's just lukewarm, then for me, that's extremely discouraging.
6. Concentrated Portfolios and Follow-On Investing
I agree that things like biographies aren't very helpful. It's almost like practicing with a checklist. But another approach that's completely different is concentrated investment, right?
In our initial fund, we adopted the common approach of investing in 20 to 30 companies, or 20 to 40 companies, per fund. It seems you're reconsidering that now. I believe Fund 2 is more aggressive and focuses its investments more intensely. What do you think about that? Does this simply mean that people aren't taking enough risks?
With Fund 1, we initially planned to invest in 30 to 40 companies, but we quickly iterated. The plan was to invest in 20 companies in the middle of the process and 10 companies at the end. So, we ended up with 20 companies, and we gradually became more focused on them.
The most important thing is that if we were an AI fund or a defense fund, we could easily find 20 or 30 companies in that field, because many companies have been established in those areas. I don't have the exact figures, but since the beginning of this year, $600 billion, or possibly even more, has been poured into AI companies. In other words, there are hundreds, perhaps even close to 1,000 companies. Therefore, you should be able to easily select around 20 or 30 of the best companies.
However, our strategy is to find companies that are highly differentiated, have very compelling founders, and can truly build a lasting moat. Therefore, the commonalities are very small, and the standards must be very high.
That's why I like the idea of finding the 2 to 4 best opportunities each year and investing in them at scale. By doing so, we could also lead a seed round with a fund of our size, right? Our fund is between $80 million and $100 million, but we can also write a $10 million check for the seed round.
Even after these companies become successful, we don't systematically make follow-up investments. As I already explained, because we don't have a follow-on reserve, we can make a pro rata allocation across all Series A and Series B rounds. What I want to do is wait for companies to say things like, “This is our K2 Space,” or, “This is our fund return.” I plan to allocate 20–30% of the fund to that investment. That's what we want to build, and we want to take our time to cultivate that strength. Therefore, I think it's important to focus and maintain a relatively small portfolio.
I completely agree. I think that's like forcing certainty. It's a kind of forceful-conviction function that you can't achieve by investing in 40 companies, for example. In the first fund, we invested 100% in 1 company. That was the beginning. Then I concentrated my investments at around 50–60%. That's right. Otherwise, it would have been absolutely impossible. Therefore, I think that's really important.
However, as a counterargument, there are people in venture capital who think like Dempsey. In his podcast, he says, “We did the calculations. We looked at the data and found that we need 30 checks to include 1 or 2 big hit stocks in any fund. And we know that if we meet that number, it's possible.” That's pretty close to a statistical approach, isn't it? I know I might make mistakes sometimes, but I know I can do it if I have 30. On the other hand, with 10 stocks, which is what most first-time fund managers do, do you know what happens if none of them are big hits? In other words, there is nothing to support my return.
Yeah. To be honest, it might not be entirely data-driven, but as far as I understand their strategy, I think it's more thesis-driven. If I were a paper-driven author, I would probably say, “Okay, this is my view of the world. If you know you're right, you're probably better off making several investments. You're probably better off making several different investments in the same direction.”
I don't know if this is how they do things, but it feels intuitively right. For example, if you have 10 theses in 1 fund, you might make 3 investments in each thesis. But I don't really understand the data-driven part. Generally speaking, I have a feeling that 1 in 12 or 1 in 15 times, our strategy will be fine. They don't have any reserves, do they? Because, at least for me, when you invest in a company, within the next 6 months you get a lot more data than you had before you invested, right? You can see every hair.
Yeah. It's like what they say happens in marriage, right? You get more information to make your next bet. What do you think about doing something like that? For example, putting 10% of the fund into a company, investing in it, and then investing further in later rounds. I know what you did with K2, but more than that, using SPVs.
Yes, I think there are several ways to do it. Certainly, as I said earlier, our intention is to do it selectively. But it really has to be selective. I don't want to say we have reserves, because that doesn't happen during the operating period of our funds—that is, in individual funds—and that's fine.
When you start investing across borders or start investing more frequently, the problem is that in such a convex market, there are a lot of short-term distortions. For example, something might be shooting up vertically at one moment, but that might not actually be the case. But even then, you're wrong, aren't you?
Right now in the market, and I think you'll probably agree, there are companies with market capitalizations of billions of dollars—or even more than $10 billion—that are going to zero in the private market or being sold as preferred stock. I think there are many such cases. So I don't think it's necessarily wise to jump on something that has momentum 6 months after investing.
In other words, the hurdles are very high. If I were to put Elon Musk into my portfolio next, okay, I'd invest 30% of the fund there. I think that would be obvious. All the best funds do that, so we should too.
7. San Francisco Groupthink vs. New York
Now, changing the subject a bit, another point about Rohanism: you said that New York is a better place to find founders than San Francisco and that you would never go to San Francisco. Is that the reason?
Yes. I think the problem with San Francisco is that it's a startup hub. Our company's philosophy is that there needs to be a little disagreement, but San Francisco is a place of agreement. When there are founders in San Francisco, everyone is desperate to meet them, and there's a kind of intellectual bubble. Everyone has the same views and says the same things. We're all the same people, we all go to the same parties, and we all chase the same investors. They're all just as weird. All the investors are chasing the same company. I'm tired of it. I really don't like being there, and I don't like listening to people talk there.
Besides, New York is better. All smart people go to New York at least once, right? I go about once every 2 months. It's a great place to meet people. I get to meet really interesting people. When I meet someone interesting in New York, they may live in San Francisco or somewhere else, but I can go and meet them. New York is a great hub, and I like it much more than San Francisco.
Like I said, a lot of our really good companies are actually in LA. I think there's an interesting group of people forming in LA who don't necessarily agree with the San Francisco consensus. They move to LA, stay on the West Coast, and still feel normal. We both know LA has its flaws, but considering the level and talent that companies like SpaceX, Anduril, and K2 Space have brought to this area, I think there's potential for something interesting to happen in the future.
There are also a lot of companies in El Segundo. There are definitely a lot of ambitious people with high IQs in various parts of LA. It might become an even more interesting place in the future.
8. TCI, Monopolies and Long-Term Moats
Interesting. Another thing I like about you is that you cited TCI as an example company. TCI is actually not very well known, considering its success. I think Geo likes TCI too. In fact, they manage 9 monopolies and $100 billion.
Well, it's a publicly traded company, so yes.
Usually, if you ask someone about their inspiration, they'll mention names like Sequoia or Founders Fund. Why TCI, and what did you learn from TCI in building Valhalla?
I'm very interested in TCI. Let's talk a little about TCI, because I think a lot of people don't know about it. TCI was founded by a man named Chris Horn. He's a hedge fund manager, and I think they currently manage nearly $100 billion. That means it's one of the world's largest hedge funds.
TCI stands for The Children's Investment Fund, and very few people have heard of it. It's an interesting group with a very talented team. I think it consists of Chris Horn and perhaps three or four other members. Analysts. Unbelievable, right? They manage nearly $100 billion.
I think the reason is that their strategy is very simple: it's to have a monopoly—a monopoly in the public markets. I think they always hold 9 to 12 stocks. All they spend their time doing is figuring out whether each monopoly is still viable or whether it's declining. Then they change the stocks in their portfolio accordingly.
You wouldn't overdo it, would you? I think this is a smart investment strategy, because companies that are truly growing and experiencing sustainable growth will never fall by 100%. Even if they go bankrupt, they will decline slowly. That's why you want to hold these assets long term. In most cases, they compound at 20–30% annually.
TCI has a fantastic track record of returns, which is why they have such enormous capital. I think a lot of people are trying to overcomplicate things. What we should be looking for is monopolies—future monopolies—in their early stages and supporting them with conviction.
If you hold a large amount of stock in a company with a very strong competitive advantage, that's the type of company that's best to hold for 10 years. You can hold it for more than 10 years. If you hold many companies for 30 years, you will get very good results. So I think the best long-term direction in investing is to have a competitive advantage.
Is there anything I haven't asked you yet about how to choose investments or investment strategies? Otherwise, I'll have to delve into each company individually.
No, I think you've covered all the points. It's about the moat: it has to be non-consensus in the early stages, and it has to be a great founder. So it's not that complicated.
Yes, it's not complicated, but it's difficult.
Yes. It's difficult to do consistently.
9. From Crypto to Deep Tech and Beyond
In fact, I'd like to talk about your company's website in 2021. When I looked it up on the Wayback Machine, it said things like digital assets, games in the metaverse, psychedelic therapy, and so on. Then, in 2023, I think I found an interview where you were talking about deep tech. You started deep tech pretty early. How did that change happen? I thought it was pretty much consensus at the time, so there was some kind of shift from consensus. I've been burned a few times by cryptocurrency. Is that what you mean?
No, no, no. That would have been a convenient story, but I was 20 when I met the people at Valhalla—the people with whom I eventually formed a partnership in 2020. They were much older than me, and I don't think I knew much at the time. I didn't have enough knowledge to express my opinion.
So, in the first year or 2 of working together, I started to think, “This physical technology seems much more interesting than other things like cryptocurrency or software.” So I said, “I think everyone should spend more time on this.” But I was only 21, right? So that wasn't that important.
Yes. Then, basically, I don't think I invested myself from 2020 until the latter half of 2022. I was observing, networking, finding people, and developing my eye for people. During that time, I thought, “Obviously, if you're trying to invest in deeply technical companies, there are better places to build a moat than the physical world, aren't there?” Even before AI—as you can see from our materials before ChatGPT came out around 2022—I thought software was going to be commoditized, not by AI, but by the fact that there were so many software engineers.
Everyone smart I knew had a degree in software engineering. I also minored in software engineering. It felt like everyone was learning it, right? That was the consensus. What wasn't part of the consensus was that students should study mechanical engineering, electrical engineering, or aerospace engineering. These were areas that people basically didn't talk about. They involved only one company, SpaceX, which was very popular and people were investing in, but there was nothing else besides SpaceX.
So I thought, “This is clearly an area I should be dedicating time to.” Finally, in the second half of 2022, we gained the trust of our partners and were able to focus our efforts on this area. Fortunately, the timing was just right for our Fund 1, and from mid-2020 to early 2023, we were doing nothing but that.
As time went on, what happened was that, if you had asked me in January 2023, I would have said that we were investing only in physical technology and nothing else. Then I learned something new. Over 2023, 2024, and part of 2025, we made what we think will be really great investments in physical technology. However, I think the consensus is very high now, so I'm looking for what comes next.
So, is it something like a transition?
Probably. Yes. What I realized is that you can't confine yourself to a box. LPs want you to confine yourself within a box. It's an incentive issue. The LP wants you to fit into the box, and I think the reason we created that kind of website in 2021 was because there was a tendency to feel the need to confine ourselves to a certain framework.
However, as the scope of our business grew larger and larger, we wondered whether we should confine ourselves to that framework to make fundraising easier or whether we should become a flexible, generalist fund.
And they actually make a profit, and in the long term, they profit from carry rather than management fees.
I wanted to make a profit with carry because I'm very competitive and want to get a higher return than anyone else. If we're confined to a box and have to chase defense deals in the hottest defense market in history, we can't do that.
Yes, that's right. We don't like funds that are clearly confined within a framework. I believe flexibility is essential when it comes to investing. For almost everything, there are actually very few rules. It's what allows you to break almost any rule.
We probably should. Rules shouldn't be necessary. In fact, I believe masters in all fields do so. Rules are like a walking stick for someone who is talented but not a master.
10. K2 Space: Betting Against Small Satellites
Let's take a look at each company individually. We were talking about K2 Space, weren't we? I find that very interesting. You researched that field and spoke with them, and even though they hadn't raised any funds, you managed to make this round happen. That will be one of the best investments in Fund 1, at least for now—an incredible investment.
So how was it? You met him, called him 6 hours later, and said that?
Yes. As far as I know, they weren't raising funds at the time, so I thought, “I really like this. I want to invest.” I called him and asked, “Can I invest?” After thinking about it for about a day, they were very kind and accepted our money. They're a great team.
I think the great thing about it is not trying to invest in what everyone else is investing in. It was incredibly difficult. Besides the team being fantastic, were there any other insights you gained? What were the key insights from that experience?
It's like what I've said before, what I've hinted at before: they were doing something that people didn't appreciate, so it was really hard for them to raise funds in the seed round. They foresaw a future where starships would be put into practical use within a few years. However, the power and mass being launched into orbit were, or were about to be, increasing rapidly.
Yet everyone was developing products with increasingly stringent limitations. In other words, those two things did not match. The reason why everyone was making them smaller and smaller was pattern matching. People had seen both Planet Labs and Starlink build successful businesses with small satellites.
Starlink was clearly a great bet, but that was because it was vertically integrated. SpaceX was able to make it a success. People were watching Starlink succeed, and everyone was pursuing things that would make the value of what small satellites could do increasingly diminish.
Elon achieved great success by realizing what was most valuable with small satellites: Starlink. That would be a monopoly for many years, and everyone else was trying to imitate it. Since SpaceX owns the launch platform, there's really nothing else they can do. Because there is no cost advantage in launches, it isn't actually possible to do anything that is economically valuable. In other words, it was actually meaningless. Therefore, none of those companies were appealing to me.
With K2 Space, we were going to build a truly massive satellite. We were building it for the future, which will likely be realized in the next 5 to 10 years. The Starship will be launched and become a rocket. And launch capacity will continue to increase. Therefore, let's build something bigger and more powerful.
11. Jaza: Bringing Electricity to Africa
We found that satellite customers were looking for more power and mass. That made perfect sense, and you've seen it too. K2 Space was an astonishing success. They signed a deal worth more than $1 billion. They were very successful in the commercial market and proved their capabilities, and they're now rapidly proving their capabilities in the government market. I believe it will become a great company for a long time to come.
That's wonderful. Then there's Jaza Energy, which I think is really interesting. It's an energy service in Africa, right? It seems all the VCs have passed on it. It was a graveyard for African companies. I'm interested—what insight led you to say yes? The 9-month process is also interesting.
Yes. Corey from Fundomo introduced me. I believe he was an angel investor.
How did you meet them? That's interesting.
Actually, I don't know this myself, but Corey is a genius. Corey really finds interesting things. I think Corey sent me Jaza because it had an unconventional feel to it, and that was really interesting to me.
First and foremost, Jeff was an incredible entrepreneur. I don't know if you've met him before, but he's one of the most energetic and serious business leaders I've ever met.
I met him. He is a beast.
Yes. He's not someone you want to compete with.
He's ruthless, isn't he?
That gave me a certain understanding of his journey. An American VC sitting in New York City would find the idea of an off-grid business terrifying. I understood his journey from where he started. First of all, his life was crazy, wasn't it? He grew up in northern Canada, in a far-from-affluent environment. He had to forge his own path in life.
Then he had a crazy trip in Africa. He ran a business there, planting trees, and realized that in sub-Saharan Africa there was clearly a problem: more than 1 billion people did not have electricity in their homes. He wondered, “Why is it like that?” and immediately began to solve the problem.
I think he started with a motorcycle battery. Ultimately, we repeated this process for many years, from 2017 to 2023. When I met him in mid-2023, he had just found a formula that worked well. It was small, but it worked.
It's incredibly difficult to get anything to work in sub-Saharan Africa, isn't it? We have to deal with all sorts of problems: terrorist groups, government opposition, local politics, a lot of violence, and the inability to communicate. These are all kinds of problems, and it worked.
You've grown, haven't you?
So we saw it, understood the journey, and grasped all the lessons. We spoke with many people who had invested in companies that had failed in that region. I was doing something very similar, and I thought, “This is actually working. I'm not overlooking anything.”
The reason nobody liked it was that people had lost a lot of money investing in Africa. They didn't really understand the market, and they didn't understand why this was valuable. We understood that, agreed with Jeff's worldview, and decided to support him.
He's been doing very well ever since. He has raised $100 million and is growing at an incredible pace. I think he's grown 10-fold compared with last year.
That's crazy. Yes, that's right. That was a case where you knew right away that this was what you wanted to do.
At the very least, we knew the founder was a special person.
Yes, I knew that. We all loved Jeff, didn’t we? He was truly wonderful.
That became the driving force behind all the work that followed. We had to make sure we didn’t invest in obviously bad areas—areas where we would obviously lose money. I had to make sure I wasn’t stupid.
I understand. So that’s what you were going to do. Got it.
12. Investment Mistakes and Overthinking
That’s one of the interesting things. I think you mentioned it, but I might be able to tell you a bit about the big failures with Fund 1. Even when you have a brilliant founder, you might overanalyze the business model. Today, it might have taken 3 months or even 1 month instead of 9 months.
I think some of my biggest mistakes with Fund 1 were when I really liked the field, what a company was doing made a lot of sense, and all the conditions were met, but the founder was only a 7 or 8 out of 10. Those investments all either went to zero, returned my money, or produced only a small return. Those are by no means good results.
So, how has my thinking changed over time? First, if a company is in a field I’ve never heard of before, seems surprising, or seems differentiated, then I’ll have a meeting with that company. That’s the criterion for having a meeting. The next criterion is the founder.
The founder must be highly competent. If the founder is competent, you do the work, but you shouldn’t take on too much work. In other words, I don’t want to tell myself that things won’t work out. One of the reasons I dislike sector-specific funds is that they have too much context related to their respective fields, which prevents them from investing in disruptive innovations.
We need to find a balance, but you also need to know what you’re doing. I’m the kind of person who cares about things like businesses and moats, so I can’t ignore them. We cannot ignore the business.
But among the businesses you considered quite early on, there were some that you passed on because you didn’t believe in the business model, but that ended up being very successful, right? Did you learn anything from that experience, or is there something else?
Because we have to do the whole thing. You need to think, “Okay, I should have done these things, but I should have done many other things as well.” In other words, one of them worked so well that I don’t think any mistakes or trades I make can compensate for the opportunity I missed out on.
I definitely overdeveloped some of the ideas, and I certainly regret those. But I think that’s part of the learning process, and I wouldn’t have done that now. That investment was particularly foolish. Because the price was so low, it wasn’t a problem. The risk adjustment worked well because the price was very low.
I think you need to adjust your strategy depending on whether you’re investing at a $10 million post-money valuation or a $100 million post-money valuation. There’s a big difference between $10 million and $100 million when it comes to owning a business for 10 years, but the difference between $10 million and $30 million is also significant.
With $10 million, you have a little more leeway and can bet on other people. For example, if someone is truly, truly, truly excellent, I think it’s worth betting on them with a perfect score of 10. No, I don’t think I’m betting on a perfect score of 100, but that’s just my personal preference. I haven’t done any calculations or looked at any data; that’s just how I feel.
So, what did you learn from that? Have you stopped overthinking your business model?
I try not to overthink things. For example, if someone scores 20 out of 10, meaning they’re incredibly talented, then we don’t worry too much about that area. If they’re either a 9 or a 10 out of 10, I’ll take a little more time, but ultimately, you just have to get used to it, right?
That’s the only way to get there. It’s not like a process. You have to get to the point where you say, “Yes, yes, yes, I understand. That’s it.” And I think it’s probably one of those things where, when you look at a certain sector, you’ll find that everything is difficult, and there are even some heterodox approaches.
13. Backing Technical AI Founders
I think Aftersort[?] is an interesting example because it’s clearly software and AI. I love Joe. He and Jeff have never met the founder of K2, but they think he must be an incredibly impressive founder. However, this is like investing in Fund 1 and then doubling your investment, and from my perspective, I’m worried that there might be something I’m unaware of.
This is highly advanced, technically sophisticated AI. You’re not an engineer, and you don’t live near the research lab, so you can’t know everything that’s happening there. Those concerns were actually one of the reasons why we decided against investing in Oliks[?].
These are the things I’m worried about. Even if I’m introduced to different tipping companies with different approaches, they all look like call options to me. You can say, “This founder is truly amazing. It’s a new approach.” But they all look pretty much the same. So the question becomes, “How many of these companies can I actually invest in?”
Ultimately, it turned out that whether the founder was truly talented or not was irrelevant.
How do you overcome that? You’re simply worried about being negatively selected, right? Why can’t you raise funds from VCs that invest in AI?
To begin with, regarding adverse selection, I think that if someone contacts me, that’s adverse selection. So if someone contacts Rohan at Valhalla Ventures, that’s a bad sign, right? In other words, they did a lot of research and just let a lot of things slide.
It’s not necessarily a bad thing, but we’re going to look for someone. We spoke with the founder because we had done a lot of research to find them. We know that this person is the type of person we want to meet, and this is an interesting idea for us.
In the case of Aftersort[?], I understood Joe’s view on consensus. “What’s wrong with LLMs? What’s the fun about LLMs? What are the fundamental shortcomings? LLMs are obviously great, very capable, and will create enormous value. But what’s missing? What’s the gap?”
There was a gap, so I agreed with Joe that there was a gap. He had something like a paper on how to bridge that gap. So I said, “Okay, this guy is really smart. He’s identified the real problem and has a rational, principled solution to it. It might take a little time, but it’s a good bet for me.”
I don’t care what investment decisions other people are making. Just because everyone else passed doesn’t mean it was a bad investment.
I think we try to avoid adverse selection to some extent, but the point of disagreement is that you’re not opposed to investing in things that other people don’t want to do, right?
In terms of technical understanding, we have invested heavily in highly technical areas. There’s a level at which you can judge whether something is within the realm of possibility, or whether it’s simply a matter of endlessly repeating experiments to see whether something works.
It’s like the difference between science and engineering. This was probably one of those things that was close to science, but I still felt it was a manageable problem. I believe they had sufficient technical understanding to make that judgment. You don’t need to dig 10 layers deep to understand the exact nuances of what they’re doing.
14. Truffle and the Future of Personal AI
Would you like to talk about Truffle? I think that’s interesting too. The founder has also been a source of considerable controversy. I love truffles.
What did you see there?
I think your investment was also a very contrarian bet.
It’s like making inferences at the edge of the spectrum. I think it’s still somewhat contrarian because people aren’t doing it enough or talking about it enough.
A lot has changed since we invested. When we met Srikanth, he was clearly a controversial character. If you look at his Twitter, you’ll see that he’s a controversial figure. That was cool, but what impressed me most when I met him was that he was a very sincere entrepreneur. He truly believed in what he was doing.
I understood the entire trajectory of his ideas, from starting the company to iterating over time. He had a very strong hypothesis that the demand for the token would increase exponentially over time. In other words, it increases vertically. It’s already vertical, and it will remain vertical.
Furthermore, it’s impossible to meet the demand for those tokens through investments in data-center infrastructure as time passes. It’s not clear yet, as we’re still able to meet the requirements, but it should become clear by 2028. I think it will become clear at the beginning of next year, at the beginning of 2027, but in the worst-case scenario, it could be 2028.
Current consensus forecasts suggest that AI capital investment in 2028 will be around $4 trillion to $5 trillion. In other words, there are limitations. There are all sorts of restrictions, including power limits and supply-chain limitations.
Srikanth’s thesis was simple: “The demand for tokens will eventually exceed investment in AI equipment. Although the exact cause of the bottleneck is still unknown, some bottleneck will form. When that happens, who will meet the demand for that token? It will be computers.”
The same thing happened during the transition from mainframes to personal computers. There were mainframe computers; all the computers were in one room, and people accessed them using terminals.
And eventually, people got personal computers. I think the same thing will happen this time. I think everyone will have their own personal computer, whether it's a Truffle or whatever. Apple might even enter this field eventually. Certainly, the Mac mini has far less GPU and memory than a Truffle, but I think a dedicated device with a dedicated operating system is necessary.
So, what would happen if Apple said today, “We're betting on a very powerful GPU”?
I think it would be very interesting if a device could be equipped with a large amount of memory, an operating system built around it, and a continuous learning system built on that—that is, if it were possible to do that on a local device.
But Apple is slow to act, isn't it? Apple has yet to find a way to make Siri work properly. That seems like a pretty simple thing.
Well, I think that's why I liked Truffle last year. We made an investment, and since then, Claude Bot has emerged. Everyone uses agents now, right? Agents are mainstream. Personal agents are mainstream.
Now, we're starting to see the limitations of personal agents, aren't we? Instinct is getting worse. Well, Muse is getting worse, right? Providing inference is very costly. Providing 24/7 inference from a data center using a frontier model is extremely costly, isn't it? Therefore, over time, we had to offer lower-quality products. It's a model that gets progressively worse in performance, isn't it?
What is Truffle? Truffle has its own unique device and its own custom weights, and I think people are really starting to understand its benefits.
The weights part is really the difficult part, isn't it?
In short, there are many truly difficult parts, but the custom-weighting part is particularly challenging. Everything is really difficult. Truffle is an incredibly difficult company to build because its supply chain is extremely complex.
Our competitor is Apple. How do they manufacture so many devices? The GPU and memory, 2 of the main components of a device, are currently among the areas where the world's largest supply chain constraints are occurring.
In short, it's extremely difficult, but Truffle has been working on it for 3 years. That's the advantage of early entry. If you enter the market early enough, you have time to resolve all of these issues.
Entering that market in just 6 months presents a lot of challenges, doesn't it?
Even if there isn't a consensus yet, there will be in a few months. Even if Muse and Instinct deteriorate slightly, people will still seek their own personal agents. Once Truffle launches, it will become clear that it's a better product for everyone—for businesses and individuals alike.
Truffle has already solved many of those problems. To be honest, Truffle is probably the perfect Valhalla investment. It's a perfect match from beginning to end.
All the conditions are met, right?
Yes, yes, I think it's a typical Valhalla investment.
There may be other typical Valhalla investments that I can't talk about, but I agree that Truffle is on the rise.
15. Growing Up Poor and Learning to Take Risks
Ah, yes. Now, let's move on to a personal story. In this interview, you also said that you don't think you're particularly intelligent, right? You did say something about how your poor grades and the fact that you had nothing to lose were your strengths, right? Could you tell me a little more about that? Where do you think your strengths come from? I think we've talked about it a little, but please tell me more.
Yes. As you know, when I was little, I grew up in a fairly poor family. Neither of my parents earned much money. My mother was the one who supported the family financially. As you know, we experienced many ups and downs while I was growing up.
I've moved a lot; I've probably moved 7 or 8 times. So, every 2 years, I felt like I was airdropped into a new city where I didn't know anyone. I couldn't understand the culture, so I had to figure things out on my own. I think that was part of my formative process.
Since then, money has always been at the back of my mind. I always knew exactly how much money was in my parents' bank accounts. It was a part of everyday life. Therefore, I always had to be careful not to overspend. That's why I think I started thinking about ways to make money from a young age. That has always been a major challenge.
Then, in my late teens, I learned about cryptocurrency and lost all my money multiple times. So, I got used to living with little to no money. In other words, I understood that I didn't need a huge amount of money to live. You can live a perfectly happy life even with a small or limited amount of money.
Therefore, I don't think that's what gives you the ability to take risks. Risk-averse thinking—that is, “I have to protect what I have,” “I have to protect my position,” “I have to protect my job,” and so on—is irrelevant.
I simply want to become a truly excellent investor. I think I discovered investing when I was probably around 20 years old. Then I started learning more about investing, and that's how I became obsessed with it. I want to become better than anyone else—not just at early-stage VC investments. Overall, I want to become a truly excellent investor. That's why I want to take my time and become proficient in this field.
What do you think is the right way to invest in a company during its growth phase?
I have many interests. Early stage is something I truly love, and no matter what happens, I want to keep doing this forever.
Yes, so I think all of those things are influencing it. I love it. Also, when we spoke before, you mentioned that you don't have any personal friends outside of work, right? I'm curious whether that was intentional or whether it happened naturally.
It just happened naturally. That's because I don't have much going on in my personal life. However, I do have something I truly love, and I dedicate all my time to it. I have a family. My family is like my personal friends.
Aside from that, although it might be a flaw of mine, I don't really think so. I'm not really interested in meeting people in a social setting. I don't like going to clubs, bars, or going out. I would never do anything like that. Sometimes I have no choice because of the circumstances, but most of the time I don't like doing that.
Therefore, in order to truly understand someone, you must first find that what they're doing aligns with what you're doing. What I'm doing is becoming a truly excellent investor. That's why I get really excited when I meet other great investors, founders, and people in the technology industry and related fields.
“Okay, I want to get to know this person better,” I think. Naturally, those are the people who become my friends, right? People I really like, people I really trust. I'd say I'm pretty close friends with most of our founders. I talk to them every day. I might be bothering them with my frequent phone calls, but it's really fun.
I think the best way to build a life is to have people who energize you, people who ignite your ambition. Everything is great, and I don't want anything more. It's really great.
By the way, it's interesting, considering that you weren't afraid of taking risks and struggled with money from a young age. You were selling things on eBay at 11 and working very early. So, it's interesting that you went down that path, because I think it could have turned out the other way around.
Yeah, something like confidence or a sense of security. I know I've done it before, and I've even had meals when I only had $0 in my bank account. So, you know you can do it, right? You can hang out with friends or do something else, like eat a meal.
I think there was a time in college when I lost all my money and had only $0 in my bank account. So, I found a delivery job. I think it wasn't DoorDash itself, but rather an early competitor to DoorDash. That's how I started driving and delivering food on campus.
I was a little embarrassed. You open the window, and a DoorDash delivery driver appears. It's like they don't care at all. I had to put food on the table.
So, do you care?
No, I didn't mind. Basically, I don't really care what other people think of me.
Has it always been like that? Did you have any experience with that?
Yes, I think so. It's true, I was a little shy as a child, but that disappeared soon after I started going out into the world.
The strange thing about my childhood, looking back, is that I never met anyone I truly admired until I was 20 years old. I lived in a place where most people lacked initiative. It was really boring, and people really didn't care. In other words, I was truly not connected to anyone.
I think I was about 15 or 16 when I made my first friend. He's also an entrepreneur, right? Of all the people I've met so far, there have been very few with whom I wanted to build a long-term relationship.
When I found something I was truly interested in and passionate about, I started building long-term relationships. That's when those relationships began to take shape. Although I now have friends I've known for a long time, it took a while to get to that point.
I think it's interesting how my life truly changed after I started meeting people I admired. That inspired me and fueled my ambition.
Having achieved great success with cryptocurrency, what lessons have you learned?
Don't gamble.
In short, don't gamble.
But were you investing at any point, or did you think you were investing?
No—that is to say, it is true that I bought Bitcoin and such, but that's a different matter. The reason I lost all my assets was leverage. Without a doubt, you must stop gambling because they tend to be more prone to gambling addiction. Therefore, you must make every effort to avoid gambling.
Have you not gambled since then?
16. Valhalla’s Future and Aligning Incentives
Yes, of course. I've never bet my entire fortune. But, yes, I think that's one of my bad habits.
So, what do you think Valhalla will look like in 10 years? That's one of the questions I wanted to ask you. As an investor, I imagine you want to find alpha wherever you are, and while venture capital has been where you've achieved that in the past, I think you'll also be opening up to private equity and other things. I'm curious to know what Valhalla actually looks like. Ten years from now might seem too far away, but what about 5 years from now?
Yes. As time goes on, besides meeting truly amazing founders, some of the moments I find most enjoyable are meeting amazing investors whom I truly respect. That usually happens when they have different views or perspectives on the world, when they're doing something together.
For example, there's James, an architect. When I met James, I was truly inspired by the difference in his perspective on the world. Corey is the same. Meeting people like this is absolutely wonderful. So, as time goes on, I hope to increase the number of people like that in Valhalla, among my comrades. I would like to partner with such people and work together. We want to find new frontiers for alpha in the private market.
We probably don't know if we'll enter the public market. I think it's something like 2 and 20, frankly. However, in the private market, I believe I have truly come to understand the incentive issues in the private market over many years.
Private-market fund managers and money managers are primarily looking for institutional investors, or LPs. Because they are looking for institutional-investor LPs, they optimize their strategies to serve institutional-investor LPs. However, due to regulations and other issues, institutional-investor LPs are increasingly less incentivized by the performance of fund companies. Therefore, other strange incentives exist that are unrelated to performance.
These negative incentives drive private-asset-management firms to do a lot of foolish things. They aim to optimize AUM to 100% and earn more management fees. Therefore, I believe there are many problems with private markets. The solution will take a long time.
If you can find a way to successfully overcome the incentive problem, I think you can earn a lot. I believe we can build a truly excellent investment company that will be able to respond very well to the next market trends. We have been in this kind of bubble cycle for a long time, ever since the financial crisis.
Therefore, we want to build something that is very sustainable and highly adaptable, capable of handling a variety of market conditions. And I think the problem with this incentive is how we find it. The LP market is terrifying.
I think it's broken. That's terrifying. When we started investing in funds, I went to talk to many fund-of-funds and similar organizations. Some things make absolutely no sense to me. It's very bad.
Even when it comes to concentration, many of them dislike concentrated investment. I just wonder why they don't increase their funding. You can replicate diversified investing simply by increasing your capital. But as you go down the list, the only explanation is, “That's not your money.”
No, it's not their money. The only thing that can explain that is that it's not your money, and your incentives are, “I don't want to look bad” and “I don't want to fail too much.” I think there are almost no exceptions. I believe all of our LPs are investing their own money.
Yeah. I don't think there are any exceptions. That's really interesting.
All of our LPs are like that. That's probably a pretty good incentive signal.
Generally speaking, when you look at LPs, it's not even a case where someone is working for someone else and managing that person's money. They are literally investing their own money. Even if it's a family office with a CIO, you'd think they're investing their own money personally, right?
That's why I think it's really, really important. LPs should invest their own money, or, if not, they should be strongly incentivized to make money. Alternatively, you should invest in people who are investing their own money, or at least people who have a significant portion of their money in funds.
Yes, there's another one. Many LPs said they didn't like it when managers put too much of their own money into the fund.
What? Because they don't take many risks, or something like that? I don't think so. We've always funded at least 50% of the fund with our own money, but I don't think that's what stopped us.
No, it's really that simple, isn't it? In other words, incentives are very important. One of our strengths is that we take incentives seriously at every level, from the LP level to our own level and the founder level. If you truly understand incentives, you can understand how the world works very quickly.
100%. I think that's the biggest lesson we've learned from cryptocurrency. Incentives rule the world.
Yeah. That's great. This was wonderful.
Yes, it was really fun. Thank you for the invitation.
Yes, thank you so much for your time. And I'll do the same next time.