Goldman Sachs Analyst: Now Is the Time to Buy Semiconductor and Momentum Stocks — Leverage Is Lower, Positions Are Lighter

Top quantitative strategists at Goldman Sachs believe now is the ideal time to position for semiconductor and momentum stocks. According to 'Wall Street Insights,' Shawn Tuteja, Head of ETF Trading at Goldman Sachs Global Banking & Markets, noted in a recent report that momentum factors in semiconductor and AI-related equities have pulled back nearly 20%, primarily due to technical and market structural factors.

Tuteja believes that with overall market leverage declining and institutional investors holding more balanced positions, this presents a tactical opportunity for investors to initiate 'exploratory long positions' in semiconductor and momentum stocks.

The consensus expectation for the S&P 500 to surpass 8,000 points by year-end remains dominant, with little debate.

Goldman forecasts S&P 500 earnings per share (EPS) for 2027 at $385, compared to the market consensus of $398. Taking a midpoint and applying a 20.5x P/E multiple, the implied level still exceeds 8,000 points—less than 7% above current closing levels.

Tuteja argues that this over-20% pullback in momentum factors reflects improved market pricing efficiency, not a breakdown in the AI investment thesis.

He notes that historically, momentum strategies tend to underperform in late July. As markets now anticipate this pattern, 'rotation trades' are starting earlier—around July 1 instead of July 18.

On a volatility-adjusted basis, the current correction aligns with past pullbacks under similar realized volatility conditions, though the pace of decline has been faster.

Client feedback received by Goldman’s cash trading desk supports this view.

Data from Goldman’s TMT trading team shows client sentiment currently at 7.5–8 out of 10, down from 9.5–10 a month ago. Most clients still believe the correction is driven primarily by technical factors.

Many clients point out that the Philadelphia Semiconductor Index rose nearly 100% from April to May, with only two brief 5% pullbacks, making a subsequent consolidation phase reasonable.

They also argue that if the market were truly 'calling the top' on AI investments, AI-related stocks would show broad weakness. Yet names like Dell and Credo Technology remain strong, and large SaaS and IT services stocks haven’t shown clear rebounds—contrary to typical 'topping' signals.

Leverage has receded, and structural market pressures are gradually being absorbed.

The total size of leveraged semiconductor ETFs in the U.S. has dropped sharply from a mid-June peak of ~$157 billion to ~$104 billion by July 8, a reduction of ~$53 billion.

This deleveraging process not only reflects capital outflows but also directly alters market microstructure.

At its peak, leveraged semiconductor ETFs generated ~$2.8 billion in daily short gamma exposure, meaning a 3% daily rise in semiconductors would trigger ~$8.5 billion in ETF rebalancing buys.

Daily gamma exposure has now fallen to ~$1.9 billion. Notably, if only price-driven shrinkage were at play, ETF assets should have dropped to ~$89 billion. Yet they remain at ~$104 billion, implying investors added ~$15 billion in new leveraged ETF purchases during the correction—evidence of ongoing 'buy-the-dip' behavior.

Meanwhile, hedge fund positioning data also shows deleveraging continues. Goldman Prime Services data indicates fundamental long/short funds have sharply reduced AI and momentum factor exposure, with total leverage now in the lowest decile of the past year.

Though these funds are down ~2.2% since June 22, they still hold ~15.5% YTD returns.

Institutional financing costs have also risen. Reports indicate one-month borrowing rates for popular names like SK Hynix (000660-KR) and Samsung Electronics (005930-KR) briefly reached Fed funds rate +12%, reducing the appeal of holding positions during sideways markets.

The implied volatility premium of individual stocks over index volatility has reached a near 20-year high.

Goldman data shows the 3-month weighted average implied volatility of the S&P 500’s top 50 constituents exceeds the index by ~26 volatility points.

This implies extremely low implied correlation—when semiconductors fall, sectors like healthcare and financials often rise simultaneously.

Tuteja notes that a monthly straddle strategy on semiconductor ETFs (weekly roll, daily delta hedge) has delivered positive returns since the AI boom began, while the same strategy on the S&P 500 has consistently lost money—a stark contrast.

He expects implied and realized volatility in semiconductors to gradually decline as leverage falls, short gamma shrinks, and positioning stabilizes.

With call-to-put skew still attractive, investors seeking long exposure while capping upside and reducing downside risk may consider collar or put-spread collar strategies.

Market focus is now shifting to earnings season.

Compared to a month ago, when some AI stocks doubled or tripled rapidly, expectations for further upside have become more rational.

Goldman’s recommended 'sell out-of-the-money calls to fund put hedges' collar strategy saw high inquiry in April–May but limited execution, as investors feared missing further rallies.

Tuteja notes that a ~$100 billion upward revision in 2026 capex forecasts for hyperscalers in Q2 was a key catalyst for the April–May semiconductor rally.

However, Goldman Research expects no similarly large capex revisions this earnings season.

With July earnings season underway, focus will shift to AI investment ROI, open-source model trends, and token usage metrics.

Tuteja concludes that the biggest variable for medium-term market direction remains the path of the Fed’s potential rate hike cycle.

Equity markets broadly price in 'no hikes,' rates markets imply ~2 hikes, and some macro funds bet on 4+ hikes.

He believes these three divergent paths will lead to vastly different leadership groups in the market going forward.