How Should Investors Short the AI Market?

TL;DR

Shorting Nvidia or other leading technology companies while their prices keep rising is too dangerous, according to Carson Block. More promising targets may eventually emerge among AI pretenders, but he would wait until an oversupply of speculative names overwhelms demand. He also argues that passive investing weakens price discovery, large language models threaten legal services, and Snowline could become an acquisition target.

Transcript

We're seeing green on the screen today. I think there was a lot of jitters earlier this week about whether or not we would get to this point coming after those NVIDIA earnings. There's a lot of people, Guy Johnson included, who are looking for the moment that the bubble pops, if you're one of those people. How easy is it to short the market right n... Read More

Key Insights

• Shorting Nvidia or other leading technology companies is exceptionally dangerous while their prices continue moving upward. Block says he would rather hold long positions than short positions in the current environment, even though Muddy Waters continues to research companies that may eventually become attractive short candidates.
• AI pretenders are more plausible short candidates than established market leaders. Muddy Waters has identified companies it considers suspect, but Block believes initiating those trades before enthusiasm fades among leading AI stocks would expose the firm to excessive risk from continued speculative momentum.
• An oversupply of speculative assets can overwhelm investor demand and end a market cycle. Block expects Wall Street to produce too many aspiring AI and technology companies, much as the proliferation of SPAC transactions helped collapse speculative demand after the extreme trading environment around 2020 and 2021.
• Large language models are useful for parsing extensive collections of documents and producing summaries. Muddy Waters uses them to accelerate document review and internal information sharing, but Block says they have not yet helped the firm identify short ideas or perform the pattern recognition required by its investment process.
• The legal services industry is particularly exposed to improvements in large language models. Much legal work involves reviewing, distilling, and synthesizing documents, although current systems remain imperfect and have caused problems for lawyers who relied on inaccurate or fabricated case references.
• Passive investing greatly diminishes price discovery because index funds buy included stocks whenever money flows in, regardless of valuation. Block cites the Inelastic Market Hypothesis, which estimates that each net dollar entering the US market increases aggregate market capitalization by $5.
• Index concentration is an offshoot of passive investment flows. Block says roughly two thirds of companies in an index can underperform the index's aggregate return in a given year because a small number of heavily influential companies drive most of the reported performance.
• Snowline is attractive because its deposit could materially benefit a mid-sized or major gold miner. Block argues that major miners have reduced greenfield exploration, concentrated spending around existing deposits, and depleted reserves through production, creating pressure to acquire companies holding significant deposits.

Questions & Answers

Q: Why does Carson Block say investors should not short Nvidia?

Carson Block says shorting Nvidia or other leading technology companies while they continue rising is a dangerous trade that could quickly force a short seller out of business. Although Muddy Waters has identified suspect companies connected to the AI theme, he prefers to wait until enthusiasm and momentum weaken among the market leaders before pursuing weaker AI-related companies as short positions.

Q: What AI companies does Muddy Waters consider better short candidates?

Muddy Waters views AI pretenders, meaning companies presenting themselves as significant AI participants without the strength of established leaders, as more suitable potential short candidates. The firm has a list of names it considers suspect, but Block does not identify them in the discussion. He says the timing remains unfavorable while Nvidia and other leaders continue moving upward.

Q: How could the speculative AI market begin to decline?

Block believes the decline could begin when the supply of speculative AI and technology names becomes greater than investor demand. Wall Street may continue producing companies and financial products tied to the popular theme until buyers are overwhelmed. Weaker aspiring AI companies could fall first, creating pressure that may eventually pull leading companies lower as speculative demand contracts.

Q: How does Muddy Waters use large language models?

Muddy Waters uses large language models primarily to process large collections of documents and create summaries that can be circulated internally. Block says the tools are helpful for parsing and condensing information, particularly when the underlying material is not exceptionally important. However, the firm does not use them to identify short candidates, and they have not improved its pattern recognition process.

Q: Why could large language models disrupt the legal industry?

Large language models could disrupt legal services because much legal work consists of reviewing documents, extracting relevant information, and synthesizing findings. Block acknowledges that the systems are not completely accurate and that lawyers have faced trouble after relying on fabricated case references. Still, he expects the technology to improve, making the legal profession particularly vulnerable to automation.

Q: How does passive investing weaken market price discovery?

Passive funds purchase stocks included in their tracked indices whenever they receive net inflows, without deciding whether individual companies are too expensive. Block argues that this behavior reduces price discovery because buying continues at prevailing prices until funds experience outflows. He cites an academic hypothesis estimating that every net dollar entering the US market adds $5 to aggregate market capitalization.

Q: Why is index performance concentrated in a few companies?

Block connects index concentration to passive investment flows that direct money toward index constituents without regard to their individual valuations. He says roughly two thirds of companies in an index may underperform the index's aggregate result during a given year. A small group of companies can therefore drive overall performance even while most constituents deliver weaker returns.

Q: Why does Muddy Waters hold a long position in Snowline?

Muddy Waters believes Snowline holds a rare deposit capable of materially benefiting a mid-sized or major gold miner. Block argues that major miners have reduced greenfield exploration and focused exploration budgets on deposits near existing operations. Because production continually depletes their reserves and terminal value, those companies must acquire deposits such as Snowline's to replenish future production opportunities.

Summary & Key Takeaways

• Carson Block says he would rather be long than short in the current market. Shorting Nvidia or other leading technology companies could quickly put a short seller out of business. Muddy Waters is watching suspected AI pretenders, but it plans to wait until enthusiasm and upward momentum among market leaders begin to weaken.

• Block expects speculative excess to end when Wall Street creates more AI and technology investment products than buyers can absorb. He compares the mechanism with the proliferation of SPAC transactions around 2020 and 2021. An oversupply of speculative names could first hurt weaker companies and then pull down leading stocks through broader market pressure.

• Muddy Waters uses large language models to summarize documents, but not to identify short candidates or recognize investment patterns. Block also argues that passive funds have weakened price discovery and concentrated index performance. His mining thesis centers on scarce expertise, limited greenfield exploration, declining reserves, and Snowline's potential value to larger gold miners.