The New Market Structure: AI Concentration and the Largest Hedge Fund Degrossing in 10 Years

Market and Positioning

• The S&P 500 fell 0.6% for the week and 0.93% in October, ending about 4% below its record high. Despite the negative index return, market breadth improved: 9 of 11 sectors rose, 327 of 500 S&P 500 stocks gained at least 2%, and the share of constituents above their 50-day moving average climbed from 20% to 61%.
• The presenter characterized the decline as a rotation and hedge-fund deleveraging, not clear evidence of a market top. Morgan Stanley prime-brokerage data was described as showing its largest degrossing on record. Charlie McElligott’s analysis cited 124 long-short pair trades under stress, with 47 moving more than three standard deviations.
• Volatility and credit markets appeared less panicked than equities: the VIX briefly moved above 20 before retreating, while investment-grade credit spreads were not at prior stress-period highs.
• The presenter acknowledged bearish technical signals in the Nasdaq, including a possible head-and-shoulders pattern and an “island top.” He said a recovery above the 50-day moving average and a return of the VIX toward 16 would be constructive, while warning that volatility and deleveraging could persist.

Market Structure and Investment Approach

• The presenter argued that returns are increasingly concentrated in brief bursts: “strikeout, strikeout, strikeout, home run.” As an example, he said Nvidia’s 87% total return over 30 months—about 30% annualized—was generated largely in a four-month period.
• His caution was that exiting equities because of bubble fears can mean missing a sharp rally and buying back at a worse price. He described the current market as one requiring rotation and active risk management, rather than assuming past market patterns will repeat.
• He also warned that hedge-fund positioning had become concentrated in AI-related winners, increasing correlations among hyperscalers, semiconductors, memory, and other AI-linked stocks. Deleveraging in one group can therefore pressure the wider trade.

AI, Earnings, and Corporate Funding

• The presenter saw fundamentals as broadly supportive of AI investment:
  • GPU rental prices were described as moving into contango, a sign that near-term computing capacity was becoming more expensive.
  • Corning reported accelerating AI-related demand and said optical connectivity needs could rise as much as 10 times per GPU, with capacity constrained.
  • SK Hynix beat expectations; the speaker also pointed to continued demand for memory and supply commitments extending toward 2030.
  • He cited potential new memory demand from robotics and “embodied AI,” including Elon Musk’s view that robots could become major consumers of RAM.
• Anthropic’s S-1 was presented as evidence of rapid growth and available funding. The speaker cited $132 billion in cash and adoption by about half of Fortune 500 companies, while saying customer spending remained below 0.1% of potential customer budgets. He argued that enterprise adoption and agent workflows could still expand substantially. An Anthropic IPO was described as likely within weeks, though the valuation figures discussed in the subtitles are unclear and should not be treated as a reliable estimate.
• On broader AI financing, the speaker cited more than $1 trillion raised for AI-related investment during the year and said financing needs may peak around 2027. He emphasized that the funding story extends beyond hyperscalers to suppliers, sovereign AI projects, and other companies building infrastructure.
• Oracle said it expected to become free-cash-flow positive in the second half of 2028, with free cash flow reaching $30 billion in 2029, and said it would limit additional funding to $20 billion through 2030. Its credit-default swaps reportedly had their sharpest weekly tightening since the prior September. The speaker also cited Gavin Baker’s view that external AI funding may be nearing a peak, while noting that the timing remains uncertain.
• S&P 500 earnings growth was cited at around 30%, versus expectations closer to 25%, with revenue estimates also running ahead of earlier forecasts. The presenter said European profit growth was, for the first time in years, approaching U.S. growth. He also warned that pressure on returns on invested capital and valuation multiples among some “Magnificent Seven” companies could make their large market moves consequential for the broader index.

Rates, Credit, and Macro Risks

• The speaker said long-term rates were likely to remain relatively high, with the 10-year Treasury yield above 4.5% as a key threshold. He suggested that a drop toward 4.5% was possible, while describing 5.5% as a level that would make borrowing more expensive.
• He highlighted the U.S. Treasury’s November 4 quarterly refunding announcement as a near-term event to watch, particularly for any change in long-term bond issuance.
• He said CCC-rated debt yields were around 16%, higher than in 2022 and at levels generally associated with crises, and cautioned that persistent high yields could lead to more corporate bankruptcies. Mortgage rates were said to have risen about 100 basis points since April.
• AI-related borrowing and the concentration of large technology companies in debt markets were identified as credit-market risks. The speaker said insurers’ demand for long-term corporate debt had fallen by roughly 50%, partly because of existing exposure to a small group of large issuers, potentially shifting more financing toward private markets and special-purpose vehicles.
• Other risks mentioned included Iran-related geopolitical threats, upcoming midterm elections, and the possibility that tighter financial conditions could affect AI funding.

Portfolio Views and Instruments Mentioned

The presenter shared his own positioning and broader portfolio preferences:

• Reduce hyperscaler concentration and increase exposure to agentic computing, memory, and optical infrastructure.
• Hold exposure to longevity and healthcare, particularly Eli Lilly (LLY), and include energy stocks.
• He said he had added to Micron (MU) and some optical names, while continuing to add Nvidia (NVDA) in smaller increments. He said he might reduce Marvell (MRVL) if a reported deal with Lumentum (LITE) proceeds, though he had not yet sold the position.
• He said silver remained part of his portfolio, while acknowledging its poor recent price action and warning against betting on a quick recovery. He also discussed gold and central-bank buying.
• He described EWZ (Brazil equity ETF) as having suffered a sharp decline, and said SMH (semiconductor ETF) had reached and held around its 200-day moving average.
• He was bullish on Eli Lilly, citing quarterly revenue growth of 39% year over year to $24.4 billion and room for GLP-1 drug penetration to grow. He noted U.S. obesity/overweight prevalence of about 110 million people, with penetration around 10% in the U.S. and 2% internationally for injectable drugs.
• For crypto, he said he favored Bitcoin versus the Nasdaq-100 (NDX) as a relative-value trade, including buying Bitcoin and selling NDX through perpetual futures. This is a leveraged instrument and carries substantial risk. He also discussed Bitcoin, Solana, and real-world-asset/tokenization projects Ondo, Centrifuge, Figure, and Providence.

Frameworks and Technical Methods

• AI-stock technical screen: Identify names classified as “advanced” or “pullback,” with a rising MACD histogram and price above a rising 200-day moving average. The screen produced five names.
• Thematic technical review: Track daily and weekly RSI, Williams indicators, MACD, and ATR across a 100-stock agent-technology universe. The presenter said average 14-day RSI was near its lowest level in a year.
• Bitcoin technical view: Apply Elliott Wave analysis; the presenter described a five-wave advance followed by a roughly 38% retracement within the fourth wave, which he viewed as a potentially standard correction.
• Portfolio construction principle: Avoid excessive concentration in hyperscalers; rotate toward agentic compute, memory, optics, healthcare/longevity, and energy, while monitoring rates, credit conditions, and crowded positioning.

Disclosures and Cautions

The speaker referred to providing a disclosure, but the supplied subtitles do not include its wording, and they contain no explicit “not financial advice” disclaimer. The portfolio comments above reflect the presenter’s views. The subtitles also contain some apparent transcription errors, so figures or valuation references that are unclear—particularly Anthropic valuation estimates—should be treated cautiously.

Presenter and sources mentioned: Jordi Visser; Morgan Stanley prime brokerage; Charlie McElligott; Goldman Sachs; Gavin Baker (on TBPN); Anthropic’s S-1; SK Hynix; Corning; Elon Musk; Oracle; Microsoft; Scott Bessent; and market commentary attributed to Mike Wilson.