Research Digest | Is Now the Time to Buy the Dip? Goldman Maps Out What Comes Next
PowerShares QQQ Trust,Series 1 QQQ | 0.00 | |
VanEck Vectors Semiconductor ETF SMH | 0.00 | |
iShares Expanded Tech-Software Sector ETF IGV | 0.00 | |
NVIDIA Corporation NVDA | 0.00 | |
Amazon.com, Inc. AMZN | 0.00 |
Core Judgement from Goldman Sachs Trading Desk: "Choppy but Upward," but Much More Confidence in "Choppy" than "Upward"
Goldman Sachs US Equity Weekly Review stated clearly:
The base case remains "choppy but upward," but our confidence in the "choppy" part is much higher than in the "upward" part. Currently, we are focused on high-quality risk-reward opportunities.
Key Positioning Data Supporting This View:
| Indicator | Current Value | Signal Strength |
|---|---|---|
| Hedge fund net leverage | 48.3% | 1-year low (down 3.0 pp WoW, largest in 5 mos) |
| Long/short ratio (MCap-w) | 1.617 | 1-year percentile: 4% |
| Total leverage | 205.0% | 1-year percentile: 11% |
| ETF % in HF positions | 5.6% | Highest since financial crisis |
| S&P 500 median SI | Highest in 15+ years | Significant defensive sentiment |
Two Scenarios: Choppiness Is Consensus, Direction Yet to Confirm
- Scenario One: Tech stocks continue to trade sideways, outpaced by the surge in hard assets.
- Scenario Two: As deleveraging ends and volatility resets, buyers gradually return, and tech stocks see a slow but persistent recovery—modest gains, modest pullbacks, making steady progress in a calm manner.
Conclusion: "The most likely scenario is a trading range, with the market led by rotations."
Sharp Divergence Within Tech: Semis Under Pressure, Software/Cloud On The Rise
Semiconductor Sector: Valuation Absorbed Through Earnings, Yet Short-term Pressures Persist
Price/Earnings to Growth Ratio Down to 1.2x, Earnings Growth Digesting Valuation:Over the past year, rolling PEG for semis has dropped from 2.6x to 1.2x. Share prices are up nearly 60%, but earnings grew even faster. Semiconductor Q2 earnings rose about 133% YoY, contributing roughly 44% of S&P 500 earnings growth. The low PEG supports the view that "semi rally is primarily earnings-driven, not just multiple expansion."
Tech ETF Flows: Record QQQ Inflow, But Rotation Underway
| ETF | Aug-to-Date Flows | Interpretation |
|---|---|---|
| PowerShares QQQ Trust,Series 1(QQQ.US) | +$1.09B | Poised for record inflow, more than double July (+$490M) |
| VanEck Vectors Semiconductor ETF(SMH.US) | -$280M | Poised for record monthly outflow |
| iShares Expanded Tech-Software Sector ETF(IGV.US) | -$610M | Likely third consecutive month of outflows |
Investors are rotating within tech—from semis and software to broader tech exposure, not exiting tech altogether. Confidence in the sector persists, but positioning is undergoing significant adjustment.
Structural Shifts Beneath the AI Theme
AI Revenue Still Accelerating, But Cost Pressures Are Now the Main Story
Compute Cost Rising: In July, CoreWeave hiked prices for various compute services by around 25%, citing improved demand and customer ROI. Higher prices confirm shortage of high-end GPUs, but risk is—if per-inference cost doesn't fall as quickly as rental rates rise, demand may shift to cheaper new cloud providers or in-house clusters.
End-to-End Cost Pressures:
| Link/Stage | Cost Change | Outlook |
|---|---|---|
| GPU/Rack | Rubin NVL72 ~$8M, +17% price | Cloud vendors may pass costs to clients |
| Memory | Tight supply may last till 2029–2030 | Top 3 makers’ CAPEX to rise ~340% (2024–2027) |
| ABF substrate | Shortage: 8% in 2026 → 34% 2027 → 51% 2028 | AI servers to reach 45% share by 2028 |
| Power | TVA data center up +10%, capacity fee $1.5M/MW | Power is becoming a capex bottleneck |
Key Question: Can cloud vendors pass along higher chip, memory, packaging, and power costs to clients? Will cost-per-inference fall faster than full-stack cost increases? GPU shipments remain the key demand metric.
AI Positive Feedback Loop: Strong but Fragile
Current Loop: AI CAPEX → Tech earnings expectations → Stock market rise → Wealth effect for high earners → Premium/luxury consumption
Key Risk: If AI CAPEX growth slows from +50% to +15%, the feedback loop could reverse. By H1 2027 at the latest, the market should recognize declining consumption, with AI CAPEX and wealth effects also turning, pushing the US toward an unmistakable recession.
AI Competition Upgraded: From "Single Benchmarks" to "Full-Stack Joint Optimization"
- Model firms designing chips: OpenAI aims to launch its "Jalapeño" inference chip in data centers by late 2026; Anthropic hired Amir Salek, who led Google’s TPU program up to Gen-7, to build a custom chip team.
- Chip firms building models: NVIDIA Corporation(NVDA.US) launched Nemotron 4 and licensed Poolside model tech for $600 million; AMD announced Instella running on its own GPUs.
- The focus is shifting from single-layer benchmark performance to integrated optimization across model, chip, software, and data center—pushing for vertical integration.
Hedge Fund Positioning: Embracing AI, but Selectively Concentrated
Q2 Rebalancing: Mega Tech Generically Trimmed, AMZN/MSFT Exception
| Stock | Net HF Position Change (#Funds) | Signal |
|---|---|---|
| Amazon.com, Inc.(AMZN.US) | +20 | Only clear add; 11th straight quarter atop VIP list |
| Microsoft Corporation(MSFT.US) | +10 | Slight increase |
| NVIDIA Corporation(NVDA.US) | 0 | Flat |
| Alphabet Inc. Class A(GOOGL.US) | -15 | Reduced |
| Meta Platforms(META.US) | -20 | Heavy reduction |
| Broadcom Limited(AVGO.US) | -25 | Heavy reduction |
| Apple Inc.(AAPL.US) | -25 | Heavy reduction |
| Micron Technology, Inc.(MU.US) | -30 | Largest trim |
Conclusion: Hedge funds broadly diversified within and outside the AI theme during the Q2 rally. Portfolio turnover hit the highest since 2021 (Infotech turnover the highest since 2011). Funds cut most mega cap tech, solely adding AMZN and MSFT.
Sector Level: Hedge funds raised Financials net allocation to the highest since pre-GFC, and Energy net allocation to the highest since 2015, also building Healthcare.
As of mid-August: Hedge fund AI net exposure has backed off the Q2 peak but remains above any pre-2026 period. Funds hold near-record net shorts in Nasdaq-100 futures. This week’s NVDA earnings and Jackson Hole are key directional catalysts for semis and market sentiment.
Seasonal Analysis: The Sep–Oct Effect in US Midterm Years
September is historically the weakest month in US midterm years, with market declines typically lasting into early October before a seasonal low forms. October has the highest volatility (“crash month”) but tends to have positive average returns and often marks lows (“bear market killer”).
Structural Causes: Most mutual funds’ tax year ends Oct 31 → large tax-loss selling; window dressing at Q3 end; in midterm years (e.g., 2026), these effects are stronger—weakness clusters in Jun–Oct, Sep especially poor, Oct often a turning point. Since the early 1960s, the 12 months after midterms have always had positive returns, averaging double-digits.
Institutional S&P 500 index(SPX.US) Target Price Summary
| Institution | 2026 YE Target | 2027 Target | 2026E EPS | 2027E EPS |
|---|---|---|---|---|
| Goldman | 8,000 | 8,385 (12M: 8,300) | $340 (+24% YoY) | $385 (+13%) |
| UBS | 8,100 | 8,400 (Jun 2027) | $350 | $400 |
Disclaimer: The content is provided as general information only and should not be taken as investment advice. All the contents shall not be taken as a recommendation to buy or sell any security or financial instruments. Any action you take resulting from information, analysis, or commentary on this article is your responsibility. Please consult your investment advisor before making any investments.