Hedge Fund AI Bets Unwind: Goldman Partner Warns Junk-Rated AI Assets Must Stabilize First

Hedge Fund AI Bets Unwind: Goldman Partner Warns Junk-Rated AI Assets Must Stabilize First

TREE NEWS reports: A sharp reversal in AI-related trades has wiped out a large chunk of hedge fund gains this year, with losses concentrated in recent weeks. Goldman Sachs Prime data shows that Asia fundamental long-short funds are down 7% for the month, with year-to-date returns collapsing from a 40% peak in June to just 16%—erasing more than half a year’s gains in four months. Global AI stocks have been net sold for six consecutive sessions, with the cumulative six-day flow Z-score falling below -3, the most extreme position reduction since ChatGPT’s launch.

Goldman analyst Mark Wilson argues that sentiment, pricing and positioning momentum have all turned sharply negative, and the pendulum may now have swung too far in the opposite direction. Yet he cautions that any AI rebound before junk-rated AI assets stabilize would be merely a positioning repair, not a fundamental repricing.

Loss Map: A Rare Speed of Giveback

The drawdown is broad. TMT long-short funds are down 5% for the month, with YTD returns narrowing from 24% in June to 12%. Global fundamental long-short funds are down 2% for the month, with YTD returns falling from 17% to 10%. European long-short funds are flat for the month but still up 8% YTD.

Positioning stress is equally evident. The six-day net selling in global AI stocks (GSXGGLAI) has surpassed the previous week’s record for the second-fastest sell-off this year, coinciding with the worst single-day drawdown for quant strategies since August. Wilson highlights a three-and-a-half-year trade—long AI winners, short AI risk assets—that had returned over 90% this year but has nearly halved since late June and is now retesting lows.

Dual Catalysts: Cracks in the Financing Narrative

Wilson attributes this week’s sharpest moves to two triggers. First, CoreWeave co-founder McBee said in an interview that life insurers provide “unlimited” capital for AI—”all the money in the world”—prompting a rethink of private credit pricing and exposing the weakest link in the AI financing chain. Second, debate over the business model and economics of frontier AI models intensified, including OpenAI CFO Sarah Friar’s public warning for enterprises to be cautious with cheap open-source models. Wilson expects this debate to become clearer in coming weeks, directly affecting views on capex scale, credit quality and AI value capture.

Both threads point to the same question: who pays for the high financing costs in AI’s capital stack? Goldman credit strategists identify bonds of emerging cloud providers and frontier-lab-related AI companies as the epicenter of the financing-cost debate.

Bull Case: No Evidence of Demand Slowdown

Wilson does not subscribe to a fully bearish view. He acknowledges that the second derivative of capex and earnings growth will likely slow from current levels, but stresses that prices already reflect much of this. He cites Friar’s comments at a Goldman conference about a month ago: “Criticized for overinvesting in compute for two years, but this year it’s fully worth it. Compute I bought a year ago can sell for 3-5x today. The only bad news is we still don’t have enough—we should have bought more.”

Wilson reiterates his core call: market leadership will shift from AI spending beneficiaries to the largest AI investors—mega-cap tech companies delivering returns through Q3 earnings. These firms can largely self-fund capex, are relatively insulated from value-capture disputes, and their shares are already outperforming the broader AI winners basket.

Bond Signals and Supply Outlook

Wilson notes a notable change in bonds: despite drivers—growth slowdown risks and financing surprises—that equity investors would not welcome, 10-year Treasury yields failed to rise further after CPI and PPI data and fell notably on Thursday, breaking a steep uptrend line. He sees this as “some kind of fatigue that hasn’t appeared before.” However, he warns that energy price pressures have not eased similarly, with December WTI futures near $100, so rate risk should not be underestimated.

On supply, Goldman’s Ben Snider estimates that net US public equity supply, including lock-up expiries, will see the largest increase since 2000 next year. Wilson is relatively relaxed, arguing history shows higher supply alone is not an obstacle to index gains as long as the market believes equity financing is used for productive investment—the key is whether the market believes it.

Key Takeaways for Investors

• Positioning is washed out, but junk-rated AI credit is the swing factor. Watch CoreWeave’s July low as a key technical level; until it holds, AI rallies are likely position repairs.
• The trade is shifting from AI spenders to AI investors. Mega-cap tech with self-funded capex may lead the next leg.
• Bond market fatigue is a warning. Falling yields despite growth risks suggest markets are pricing something new—monitor credit spreads and energy prices.
• Supply is rising but not necessarily bearish. The market’s belief in productive use of proceeds matters more than the absolute supply number.