Top quantitative strategists at Goldman Sachs believe the time is right to go long on semiconductor and momentum stocks.
In a recent report, Shawn Tuteja, head of ETF and custom basket volatility trading for Goldman Sachs Global Banking & Markets, noted that the nearly 20% momentum factor pullback in the current semiconductor and AI sectors is primarily driven by technical and structural factors.
He argues that with lower systemic leverage and more balanced institutional positioning, investors have a tactical opportunity for "tentative positioning" on the long side of semiconductors and momentum stocks.
The consensus forecast for the S&P 500 year-end target above 8000 points remains largely undisputed.
Goldman Sachs's 2027 earnings per share forecast is $385, while the market consensus is $398. Taking the midpoint of these and applying a 20.5x P/E valuation still results in a figure exceeding 8000 points, less than 7% above the current closing price.
Technical Drivers Behind Momentum Factor Pullback
Tuteja posits that this momentum factor pullback of over 20% fundamentally reflects improved market pricing efficiency, not a collapse in AI trading logic.
He points out that momentum strategies have historically performed weakly in the latter half of July. Having fully recognized this pattern, "rotation trades" did not wait until July 18th to initiate but started as early as July 1st.
Judging by the volatility-adjusted magnitude of the decline, this drop closely aligns with the historical sell-off trajectories of the momentum factor under similar realized volatility environments; the market simply reached this level faster.
Client feedback from Goldman's cash trading desk corroborates this view.
According to the firm's TMT trading desk, current client sentiment sits around 7.5 to 8 out of 10, cooling from 9.5 to 10 a month ago. However, clients widely agree the core driver of this pullback is technical.
Several clients noted that the SOX semiconductor index rose approximately 100% cumulatively from April to May, experiencing only two ~5% corrections during that period, making a subsequent consolidation phase reasonable.
They also observed that if the market were truly "declaring" the AI trade over, a broader decline across the AI sector would be expected. However, stocks like Dell and CRDO remain at elevated levels, and large-cap SaaS and IT services stocks have not shown significant rebounds, which is inconsistent with a "top" signal.
Leverage Recedes as Structural Pressure is Absorbed
The size of the US leveraged semiconductor ETF complex plummeted from a mid-June peak of roughly $157 billion to about $104 billion by July 8th, a cumulative reduction of approximately $53 billion. This deleveraging process represents not just fund outflows but also directly alters market microstructure.
At the peak, leveraged semiconductor ETFs alone generated a daily short Gamma exposure of about $2.8 billion, implying that a 3% single-day rise in the semiconductor sector would require ~$8.5 billion in stock purchases for leverage rebalancing. This daily Gamma exposure has now compressed to around $1.9 billion.
Notably, if considering only implied size compression from spot price declines, the complex's size should have fallen to about $89 billion. The actual $104 billion size suggests multiple investor groups bought an additional ~$15 billion through leveraged semiconductor ETFs during the decline, indicating "buying the dip" behavior persists.
Concurrently, hedge fund positioning data also points to a deepening deleveraging process.
Goldman Sachs Prime Services data shows that over recent weeks, fundamental long/short funds have significantly reduced AI and momentum exposures, with total leverage now at its lowest decile in nearly a year. The fundamental long/short strategy manager is down 2.2% since June 22nd but remains up about 15.5% year-to-date.
Furthermore, financing costs for institutional clients have risen markedly. Last week, the one-month financing rate for popular names like SK Hynix and Samsung reportedly reached as high as Fed Funds Rate + 12%, further diminishing the appeal of holding positions in a sideways market.
Single-Stock Volatility Premium at Historic Highs, Compression Window Opening
On the volatility front, the current premium of single-stock implied volatility over index implied volatility has risen to its highest level in the past 20 years.
Goldman Sachs data shows the weighted average implied volatility (3-month term) of the S&P 500's top 50 constituents is about 26 volatility points higher than index implied volatility. This indicates extremely low index implied correlation, meaning a decline in the semiconductor sector necessarily implies gains in sectors like healthcare and financials.
Tuteja notes that a one-month straddle strategy (weekly roll, daily Delta hedge) on a semiconductor ETF has generated consistently positive returns since the AI trade began, while the same strategy on the S&P 500 has yielded consistently negative returns—a stark contrast in performance.
He believes that with lower systemic leverage, reduced daily short Gamma exposure, and a return to more stable positioning structures, implied and realized volatility in the semiconductor sector should begin to compress.
The current skew favoring single-stock call options over puts remains attractive, suitable for investors wishing to maintain long exposure while locking in upside and hedging downside via collar or put spread collar strategies.
Upside Potential Converges, Focus Shifts to Earnings Season Fundamentals
Compared to a month ago when some AI stocks doubled or tripled in short order, current market expectations for upside magnitude have rationalized.
The collar strategy—selling out-of-the-money calls to fund put protection—heavily promoted by Goldman's research team saw active inquiries but few transactions in April-May, as investors widely feared missing further upside by selling calls.
Tuteja notes that the ~$100 billion upward revision in 2026 capital expenditure expectations for hyperscale cloud providers in Q2 was a key catalyst for the April-May semiconductor rally. However, the research team does not anticipate a similar scale of capex expectation revisions during the upcoming earnings season.
Entering the July earnings season, client focus will center on AI investment returns, the development trajectory of open-source models, and token usage trends.
Tuteja also highlights that the most significant variable affecting the medium-term market landscape will be the direction of the Federal Reserve's potential interest rate hike cycle.
The base case for equity accounts is 0 hikes, current market pricing suggests 2 hikes, while several macro accounts are positioning for a scenario of 4 or more hikes. The impact of these three paths on market leadership would be fundamentally different.
Comments