Sohn Conference Foundation · · 15 min
Short-Selling Panel at Sohn Investment Conference 2026
Soren AandahlJoyce MengZachary DatikashJim Chanos
TL;DR
- Joyce Meng sees “fake AI” as a rich short theme—companies claiming exposure they do not have. Her screen of name changes that suddenly added “AI” surfaced Resolve AI and a Chinese home-decoration company turned AI-server maker that she said Photoshopped NVIDIA’s product catalog onto its website and claimed hires LinkedIn could not verify. Her short book is up this year, with positive short performance since fund inception, and is ballasted with lower-volatility secular decliners.
- Soren Aandahl’s best shorts are the boring ones—“avoiding the topless parties.” Breadth is so narrow “the modal stock is down,” which he sees as emblematic of 1999–2000. His agriculture ideas examine Straits of Hormuz closures, chemicals, fertilizer and sulfate prices, and their supply-chain and planting consequences in names “someone in their mom’s basement with a wife-beater on” is not blogging about on Twitter.
- Zachary Datikash sees short opportunities in the private-capital pile-in—private credit, PE sponsors and BDCs, as well as the businesses they crowd into. He said he thinks Brookfield has over $1 trillion of AUM with negative tangible equity, versus a $1.5 trillion U.S. high-yield market. Chanos separately cited Planet Fitness, where PE-owned gyms expanded despite terrible unit economics before cutting back; Zach used Mister Car Wash to show how three locations within a one-mile radius can destroy a good business.
- Meng’s fund is over 50% international, with exposure to the real AI supply chain against fake AI, and shorts terminal-disruption stories. She cites information asymmetry, poor reporting standards and business-quality inflation metrics in China and elsewhere. A 1% change in terminal growth can swing a DCF by 8–15%, making legacy software and commoditized BPO/contact-center services in India attractive, asymmetric shorts.
- Jim Chanos’s contrarian data point: the Internet barely moved the macro needle. He said U.S. GDP growth and S&P corporate-profit growth were both about 6% CAGR in 1986–1996 and 1997–2007, and that he did not penalize the comparison for the GFC. The opening up of China, he said, had a much bigger impact than the Internet, even as Soren’s railroad-and-Internet lens shows that transformative technologies can produce bubbles and bankrupt highly valued early firms.
- The panel’s forward short theme is that AI may commoditize software and white-collar services. Zach counted 145 software companies with $1B-plus market caps, at least at the time of his review, and said a competing product can add a desired feature literally overnight. Meng says software’s 80%-plus gross margins do not make code creation as strong a moat, especially for light-touch point solutions; she prefers less-covered data names such as Morningstar after FactSet has been crushed. AI may produce output better than the bottom 25% of knowledge work, which Meng says represents 45% of the U.S. population; she cited 60–70% of new compute as substituting for human capital, while stressing that the timing and duration remain unknown. She sees a rich long-short-spread opportunity in the K-shaped economy.
Digest · the substance, structured for research
1. Fake AI is a rich short pond in a frenzied tape
- Joyce Meng says shorts require variant perception plus a catalyst, but her book leans thematic because “a cresting tide takes down a lot of names in the same neighborhood.” Her favorite theme is “fake AI”: she described a Chinese home-decoration company turned AI-server firm that Photoshopped NVIDIA’s product catalog onto its website and claimed hires LinkedIn could not verify.
- The screen that found Resolve AI looked for corporate name changes that suddenly contained “AI.” Meng says the short book is up this year and has had positive short performance since fund inception; she balances volatile names with secular decliners that have less volatility.
2. The boring shorts are working—and this rhymes with 1999–2000
- Soren Aandahl’s read is that the rally’s breadth is so narrow that “the modal stock is down,” which he sees as emblematic of his study of 1999–2000. His answer is “avoiding the topless parties.”
- His agriculture shorts examine Straits of Hormuz closures, chemicals, and fertilizer and sulfate prices, tracing their consequences through supply chains and growers’ planting decisions into obscure companies with little retail or Twitter attention.
3. Private capital’s pile-in can ruin good businesses—and sponsors look stretched
- Zachary Datikash follows the capital cycle into private credit and private equity, seeing opportunities both in sponsors and BDCs and in the businesses where capital has piled in. He used Mister Car Wash as an example: car washes can be excellent businesses, but three locations within a one-mile radius can make the economics much worse.
- Jim Chanos added Planet Fitness: PE-owned gyms felt pressure to expand despite terrible unit economics; once expansion slowed, the growth story fell apart.
- On sponsors, Datikash said he thinks Brookfield has over $1 trillion of AUM with negative tangible equity, against a $1.5 trillion U.S. high-yield market. He cautioned that the figures “might not all be accurate” and that investors have become carried away with how well the business scales. Chanos interjected: “It’s a simple business. You take money off the top.”
4. Asia offers information asymmetry; terminal-value math makes services shorts large
- Meng says her fund is over 50% international, with significant exposure to China, greater China and India. It owns substantial AI-supply-chain exposure—“real AI against fake AI”—while exploiting information asymmetry, poor reporting standards and business-quality inflation metrics.
- She is focused on secular decliners as terminal value is repriced, especially in China’s fast AI cycle. A 1% change in terminal growth can swing a DCF by 8–15%, so terminal-disruption shorts in legacy software and commoditized business services—particularly BPO and contact centers in India—can offer substantial downside asymmetry. She says the fund has made a lot of money on large shorts there.
5. AI’s disruption is real—and historically so is the bubble
- Aandahl’s historical lens includes both the Internet and railroads in the 1850s. Revolutionary technologies can transform society and commerce while also producing massive bubbles and bankrupting highly valued firms and funds.
- Chanos’s supporting data: U.S. GDP growth and S&P corporate-profit growth were each about 6% CAGR in 1986–1996 and 1997–2007. He said, “We didn’t penalize it for the GFC,” and argued that the opening up of China had a much bigger impact than the Internet, despite the Internet’s profound changes.
- Datikash sees a new software dynamic. In a review of 145 global software companies with $1B-plus market caps—“at least back then”—he noted that a lower-priced sales-force-automation product can add a desired feature literally overnight, eliminating the reason to pay more for a competing product. That makes everyone compete with everyone; he extends the question to services automation and consulting.
- Meng says software’s 80%-plus gross margins do not mean code creation remains as strong a competitive advantage, especially for light-touch, front-office point solutions. Services could be massively deflationary in hours worked, creating difficult price-elasticity questions. In uncertain markets, she expects valuations to follow earnings cuts or out-year revisions; while bellwethers have already de-rated, she prefers less-covered, low-short-interest data names—“if people have crushed FactSet, why not Morningstar?”—where data is easy to scrape.
- Her broader K-shaped-economy theme includes a weak consumer backdrop, lower-end white-collar work and potentially deflationary AI labor. She says AI may produce output better than the bottom 25% of knowledge work, which represents 45% of the U.S. population; she has seen estimates that 60–70% of new compute substitutes for human capital and points to major technology layoffs. The change may take years, its duration is unknown, and she sees a rich environment for long-short spreads.