Excess Returns · · 1h 1m
From Poker Pro to Elite Hedge Fund Manager: David Orr on Investing, Risk and AI
With David Orr
ChorusBring the market's conversation to your AI. Learn more
Summary
David Orr, a former poker professional turned hedge fund manager, discusses his investing approach across AI, Japan, defense and market structure. His central AI argument is that compute is becoming a commodity, with supply already pre-bought by software customers and new capacity on the way, so oversupply could hit AI stocks hard, though he is unsure of timing. He is bullish on Japanese equities, citing corporate reform, buybacks and cheap stocks such as Hikari Tsushin, while warning that family-controlled cheap names may be value traps. He is bearish on Seagate and Xerox, arguing that consensus AI tools cannot supply the differentiated views needed for strong risk-reward, and he sees crowded factor trades and levered positioning as a source of market fragility. He expects AI to displace white-collar work and eventually human portfolio managers over time.
Key takeaways
- David Orr on International equities: Japanese stocks are cheap and he is heavily long Japan; many trade near nine times earnings.
- David Orr (bearish on Seagate Technology Holdings): Seagate is a major short; AI consensus sees it as safe demand, but he thinks that is wrong.
- David Orr on AI infrastructure & data centers: Compute is a commodity, so a boom-bust oversupply is likely and could already be close.
- David Orr (bearish on Xerox Holdings): Xerox is a long-term short; its business probably won't exist in ten years.
- David Orr (bullish on Hikari Tsushin): Hikari Tsushin is a very cheap, well-run Japanese stock and one of his largest positions.
What was said, by company and theme
17 views from 1 speaker. Every quote is checked against the episode's transcript (how).
Seagate Technology Holdings
Across the shows: 5 speakers in other episodes, 5 bullish and 0 bearish calls, past 30 days
Where everyone stands on Seagate Technology Holdings
David Orr · Militia CapitalBearish
Seagate is a major short; AI consensus sees it as safe demand, but he thinks that is wrong.
Seagate is one of his big shorts. When he asked an AI system about the Seagate investment thesis, it called the stock high-demand and one of the lower-risk names. He says the AI is just repeating the consensus view. He says he may turn out wrong on this example. His view is that a human needs to read through the consensus and override it.
“one of my big shorts with this, Seagate, and maybe I'll end up being wrong on this example”at 20:40
Xerox Holdings
David Orr · Militia CapitalBearish
Xerox is a long-term short; its business probably won't exist in ten years.
Orr's short thesis rests on business durability. He asks whether the business will still exist in ten years and answers that it probably won't, because the product is obsolete: 'we don't need photocopiers anymore.' He calls it a long-term short he has held since the company was worth billions. Its market cap is now around $100 million. He says the thesis is not hard and needs no deep research. He has held the short for years and has discussed it publicly before. He sizes it as a small position, since a squeeze would cost only a few percent of the fund.
“Is this business even gonna exist in ten years, you know? Probably not. You know, Xerox or something was like this. So, we don't need photocopiers anymore. It's just nonsense.”at 38:51
Hikari Tsushin
David Orr · Militia CapitalBullish
Hikari Tsushin is a very cheap, well-run Japanese stock and one of his largest positions.
Orr calls Hikari Tsushin the most obvious long he has seen. He cites four points. The stock trades at about 10 times earnings. Its operating business is growing well. Management is reinvesting its cash extremely well. The CEO is buying a large amount of shares personally. He adds that the holding company owns a diversified portfolio of stocks, which protects the downside. Because the thesis is so simple and the stars have aligned, he sizes this position much bigger than his usual small bets. He says he is not betting the whole farm on it. He has posted about it repeatedly.
“Like this stock [Hikari Tsushin], I've posted about it a bunch. This stock is so cheap... I'm not betting the whole farm on it but I'm betting pretty big on it.”at 44:44
Market sentiment, positioning & volatility
Where everyone stands on Market sentiment, positioning & volatility
David Orr · Militia Capital
Recent selloff was levered funds blowing up on crowded trades, not a systemic threat to ordinary people.
He says the recent sell-off was simply levered investors blowing up. Multi-strategy shops run around 8x leverage with pairs such as long AI and short something else, or long Nikkei against short Japanese small caps. On the day, Japanese large caps fell 5% while small caps rose 1%, which he says shows how extreme the positioning was. He says the August 2024 Japan move was a warning, and that as far as systemic risk goes, regular people will not be hurt much. Stocks may reprice somewhat cheaper, which he doesn't think hurts people long term.
“all that happened is all the levered investors are blowing up. ... Maybe stocks will reprice, you know, a bit cheaper. I don't think that hurts people long term either”at 8:54
David Orr · Militia Capital
Traditional factors like value and low volatility are crowded and no longer reliable; AI-theme factors can work.
Low-volatility stocks got crowded after the betting-against-beta paper, pushing their multiples too high. Value has been crowded since the 2000s as people copied Graham, Dodd and early Buffett, so the cheapness that remains is often a trap. He cites a declining retailer on eight times earnings with a big lease liability as not real value. He also points to Terry Smith's quality bubble, where stocks like Nike fell about 80%. So simple factor investing doesn't work. Newer thematic factors, such as AI and drone warfare, can work if you get in early and are right.
“value's been overcrowded basically since the two thousands”at 14:29
David Orr · Militia Capital
Crowded shorts move together as one bet, so keep them to a small capped budget.
Orr avoids crowded shorts because they move together and are effectively one bet. Investors who think they hold 50 different shorts are often exposed to the same thing. He says a differentiated fundamental reason does not change this, since the trade still behaves like every other crowded short. He suggests a small budget for them, perhaps 3 to 5% of the book. Adding another 20 crowded shorts would not work. He agrees with the host that baskets of stocks now move together regardless of fundamentals. He says investors must accept this as the market's current structure instead of fighting it. Shorts that are fundamentally right can still lose money if they move with the crowd.
“One thing I like to avoid are these kinda crowded shorts because they all move together. It's all one bet.”at 40:17
International equities
Where everyone stands on International equities
David Orr · Militia Capital
Japanese stocks are cheap and he is heavily long Japan; many trade near nine times earnings.
He moved to Osaka to find cheap Japanese stocks. Walking around and seeing businesses, he looks them up and finds some trading around nine times earnings, despite growing and reinvesting well. Almost no one else is actively investing in Japan from where he lives, so he thinks he has an edge. He is heavily long Japan, and the August 2024 crash cost him only a few percent.
“because I'm so long in Japan. You know, I moved here to find cheap Japanese stocks.”at 10:29
David Orr · Militia Capital
Japan's push for better governance is already working, with buybacks and balance sheet gains.
Orr says the Japan thesis is not hypothetical, because companies are already acting. Firms are improving their balance sheets, returning cash, raising return on equity, and dropping bad business lines. He uses an AI tool that summarizes Japanese filings and ranks them by importance. Each day it shows regular, lumpy announcements from companies buying back far more shares. Around 30 companies are doing this, and he checked that many did not do it five years ago. He sees this as a broad change across many companies, not a few isolated cases. He presents it as a thesis that is already paying off.
“That thesis has already been working too. Like, it's not like it's a hypothetical... they're already way, way improving their, balance sheets and returning the cash and improving return on equities and dropping bad business lines.”at 45:54
David Orr · Militia Capital
Avoid the cheapest Japanese stocks controlled by families that ignore government governance reform.
Orr says cheapness alone is not enough in Japan. Some very cheap companies are under firm family control, and those families have said they will ignore the government's reform guidance. He says investors buying these names may get lucky in a few cases. A possible catalyst is a founder's age, since a founder who is 93 may soon leave the picture. But if the founder is 76, he may live another 15 years while refusing to change anything, and the return will be poor. He says he would avoid these situations where the family has already shown it is stubborn. The rest of the reform-driven Japan opportunity, he says, should do fine.
“You know, you don't wanna, you don't wanna, like, buy those ones because, like, the family has very firm control, and they've basically said they're gonna ignore this whole guidance from the Japanese government.”at 47:02
AI: overall outlook
Where everyone stands on AI: overall outlook
David Orr · Militia Capital
Current AI gives consensus views and misses differentiated theses, so humans must override it.
When asked about Seagate, the AI system said it was high-demand and low risk, which is the consensus view. He says consensus is right 70-80% of the time, but a good risk-reward usually needs a differentiated view, which AI doesn't have. So when he is right that the consensus is wrong, the AI is simply wrong. AI has only recently become willing to change its view when pushed with deeper questions, and he is not sure it will ask those questions itself in a few years.
“So it just sort of is going with the consensus view still. It's kind of victim to group think.”at 20:57
David Orr · Militia Capital
Picking AI winners is nearly impossible; value may spread across many firms, as autos did.
Buffett's 1999 speech said most car companies went bankrupt, but automakers are still a large industry. The combined market cap of car manufacturers is around 1-2 trillion dollars, about 1-2% of global equities. So he argues against guessing which layer of AI wins, whether frontier models, services, memory makers or hyperscalers. He says no one can know that.
“So, the way I see it is you shouldn't try to guess. Is it the frontier model? Is it the services company? ... I don't think anybody can know.”at 31:04
AI & jobs
Where everyone stands on AI & jobs
David Orr · Militia Capital
Hybrid AI-human investing wins for 5-10 years; humans may become irrelevant after that.
He expects the best results for at least five years, maybe ten, to come from a custom AI system combined with a strong human portfolio manager, as chess engines did. Over longer horizons he worries humans become irrelevant, perhaps in seven to ten years. AI can know every business, competitor and country and has no bias, so it could ultimately be a superior investor. He says the edge could go to zero, which would be fine for the world, as it did for poker players.
“for at least five years, maybe ten, it'll be the... If you combine a really good custom AI system with a really good human portfolio manager, that will be the best thing.”at 23:10
David Orr · Militia Capital
AI will displace white-collar lawyers and accountants over 10-20 years, which is good for society.
He thinks AI is good for society, making things more efficient. He argues white-collar workers are not the most important to the economy. He would like highly paid corporate lawyers and accountants to be displaced over ten to twenty years, through a transition period. Older workers may retire earlier, middle-aged workers may retrain, and young people will avoid those careers. He thinks most people he talked to have accepted this.
“hopefully, all of the super well paid corporate law people, hopefully, they have to go find something else to do. ... I think it'd be good for the world if, you know, our accounting or all of those kind of jobs. If that can go away”at 27:59
AI infrastructure & data centers
Where everyone stands on AI infrastructure & data centers
David Orr · Militia Capital
Compute is a commodity, so a boom-bust oversupply is likely and could already be close.
He argues compute is a commodity because several competing models are all at roughly the same level, and frontier model makers look like a poor long-term business. The main buyers are software companies, which need far less compute than hyperscalers are building for. Those companies have already pre-bought about two years of capacity in advance, and CapEx plans add two more large years of supply. Once customers learn to use the compute they already bought, supply could already be close to oversupplied in the short term. Compute processing power is also improving fast. He is not confident on timing: if compute stays a bottleneck for about seven years, prices could stay high for five years and AI stocks would keep rising. But if oversupply is close now, AI stocks could fall around 80%.
“It's just a commodity boom bust cycle. Compute's a commodity. ... it could already be the case that we're close to oversupplied short term now, you know, once these companies learn to use the compute.”at 31:43
David Orr · Militia Capital
AI compute is heading toward oversupply as inference gets faster and cheaper to sell.
Orr compares AI model output over about 18 months. A year and a half ago, a top reasoning model in deep-research mode took 20 to 30 minutes to produce a good answer. Now it produces an equally good answer in about 90 seconds. He extrapolates that in another year and a half the same answer could take about five seconds. If answers arrive that fast, he argues providers will simply charge for the output, so the price of compute falls. He says the oversupply is 'super obvious' on that path. He supports this with his own firm's experience: an employee who is very good at vibe coding builds large feature changes in about 30 minutes, and the code works nearly perfectly. He admits his research questions may be less complicated than building a large software system. He also says the same reasoning applies over five years, and that he cannot see the improvement rate slowing for what his firm does. He is less sure this holds for a large company.
“In another year and a half, is that gonna be like five seconds? And if it's five seconds, you're just gonna charge money for that. Like, this is, the price of this, the oversupply is just super obvious then, isn't it?”at 35:41
Defense
Where everyone stands on Defense
David Orr · Militia Capital
Drone warfare is a real new-defense theme, but the trade has been overdone with nutty valuations.
He sees drone warfare taking over, with a set of modern companies building for it. He calls it a clever thematic idea that separates new defense from old defense. He says the trade has probably been overdone and the valuations have become nutty. If you can identify the theme early and get long before it is priced in, it is a good way to invest.
“Because now we have these drone warfares kinda taking over... Probably the trade's been way overdone. The valuation's gotten nutty and everything.”at 15:54
Enterprise AI adoption & AI software
Where everyone stands on Enterprise AI adoption & AI software
David Orr · Militia Capital
AI now handles his filing translation and news triage, saving a lot of time for his investment process.
His risk system was rebuilt with AI and is now automated. It is somewhat more precise, but the main benefit is time saved. AI translates filings for global investing, and it sorts earnings releases and other news by importance, so he skips routine releases. He says he no longer does any other research, and that everyone will have these tools cheaply in a few years.
“it's automated, so it saves me time. I don't think the extra few percent of precision really even matters, to be honest, but the time save is enormous.”at 19:10
Across the shows
What other episodes Chorus covered said about the same companies in the 30 days up to this one.
- Seagate Technology Holdings5 speakers · 5 bullish, 0 bearish calls
Topics
- AI for investment research workflows
- AI in investment research
- AI replacing human portfolio managers
- AI storage demand
- AI value capture
- Japan cheap family-controlled stocks
- Japan corporate reform and buybacks
- Japanese equities
- Japanese value stocks
- business durability
- compute oversupply
- crowded shorts
- drone warfare
- factor crowding
- leverage and crowded positioning
- white-collar job displacement
More from Excess Returns
- Show Us Your Portfolio: Meb Faber · November 15, 2026
- Fidelity’s Jurrien Timmer on AI Spending, Bond Yields and the Next Phase of the Bull Market · November 13, 2026
- Stock Investors Fear an AI Bubble. Bond Investors Fear Inflation. How Can You Diversify Without Selling Them? · November 12, 2026
- Tech Is the Cheapest Since 2022. Walmart Is 46 Times Earnings. Where Is the Bubble? · November 11, 2026
- This $1.5 Trillion Manager Expects More Fed Hikes. So Why Is He Still Bullish? · November 10, 2026
Ask your AI about this episode
Chorus gives Claude, ChatGPT and other assistants every view from this episode and every other show Chorus follows. Ask who agrees, who changed their mind, and what else was said about the same companies.
Chorus reports what speakers said. It isn't investment advice. How Chorus works · About Chorus