Retail fund momentum is cooling, but AI and hyperscaler earnings remain the core trading themes

Overview

This issue of Retail Radar tracks U.S. retail trading activity from July 16 to July 22. The report’s core conclusion is that retail participation has not collapsed, but momentum is cooling; ETF buying has weakened significantly, especially as tech ETF flows turned negative, while single-stock buying remains concentrated in communications, technology, AI data centers, electrification, and Mag 7-related themes. At the same time, the rapid retracement in the momentum factor shows that crowded trades are releasing risk, while hyperscalers and the AI hardware chain remain key market focal points during earnings season.

Core views

First, retail fund flows are in a moderately weak state, with weekly inflows of $5.7B below the 12-month average of $6.8B; ETF buying has fallen to a very low percentile, but single-stock trading remains highly active. Second, the momentum factor retracement has characteristics of a crowded-trade unwind: the long-short portfolio has fallen 14% over the past month, accompanied by multiple high-sigma negative tail events, signaling continued short-term risk of factor air pockets and crowded reversals. Third, the AI capex supercycle still dominates earnings-season expectations, with the market closely watching Google Cloud growth, cloud and advertising monetization, enterprise demand, token pricing, frontier model competition, and returns on capex. Fourth, retail flows continue to favor large-cap tech and AI-related stocks such as NVDA, TSLA, MSFT, AMZN, META, and GOOGL/GOOG, while selling names such as AAPL, AAL, SNDK, PYPL, and CBRS.

Analysis framework

The report uses JPMorgan’s retail flow monitoring framework to break down retail trading behavior across ETFs, single stocks, thematic baskets, sectors, options, social media popularity, high short-interest stocks, and non-retail futures flows, and combines earnings-season events, factor crowding, momentum retracement, and AI capex themes to assess market implications.

Methodology notes

• alternative_dataRetail Trading Activity Monitor

  Retail fund flow and trading activity percentiles

  The report uses weekly retail fund flows, ETF and single-stock net buying, z-scores, and historical percentiles to measure changes in retail participation.

• factor_analysisMomentum Crowding and Unwind

  Momentum crowding and high-sigma negative tail events

  The report evaluates the severity of the momentum factor retracement through an S&P 500 market-cap-weighted quintile long-short portfolio and the frequency of negative sigma events.

• earnings_analysisHyperscaler Earnings Scorecard

  AI capex, cloud growth, and AI monetization

  The report summarizes the key variables in hyperscaler earnings season as AI capex, cloud business growth, AI monetization, enterprise demand, and returns on capex.

• positioning_riskMeme Battlegrounds

  Retail buying and squeeze risk against hedge fund shorts

  The report screens for stocks with high social media popularity, strong retail buying, and high hedge fund short interest to identify potential short squeezes or retail loss risks.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• AI Datacenter / Electrification basket

  Core retail buying theme

  Strengths

  Weekly retail turnover increased by $1.32B, monthly by $7.14B, and YTD by $44.72B, showing continued concentration of capital.

  Weaknesses

  Theme crowding is relatively high; if returns on AI capex or financing capacity are questioned, volatility may increase.

  Comparison

  Stronger than low-end consumer-sensitive, inflation-related, and wage-sensitive themes.

  Risks

  Slowing AI demand, an overly steep capex curve, insufficient debt-market absorption capacity, and increased supply plus efficiency breakthroughs compressing end-market profit margins.

• Mag 7

  Key retail positioning target

  Strengths

  NVDA, TSLA, MSFT, AMZN, and META received retail net buying, with earnings season and AI themes providing trading catalysts.

  Weaknesses

  AAPL saw -$114M in net selling, while GOOGL and TSLA showed weak post-earnings stock-price reactions.

  Comparison

  Retail activity at the single-stock level is stronger than in ordinary sector ETFs.

  Risks

  High valuations, disappointing earnings guidance, insufficient AI monetization, macro volatility, and weakening consumer demand.

• GOOGL/GOOG

  Retail pre-earnings buying and cloud-growth validation target

  Strengths

  Google Cloud grew 82% year over year, clearly above buy-side and Street expectations, and capex guidance was raised.

  Weaknesses

  The stock fell about 4% after hours following earnings, indicating continued market disagreement over capex and future returns.

  Comparison

  Alongside TSLA, it was one of the high-attention names that retail investors bought ahead of earnings that week.

  Risks

  2027 capex expectations, the pace of AI monetization, and volatility in advertising and cloud demand.

• TSLA

  Earnings-event and active retail trading target

  Strengths

  EV sales drove strong revenue growth, and retail investors bought $101M on 7/22.

  Weaknesses

  Earnings missed expectations due to lower prices and regulatory credit impacts, and the stock fell slightly after hours.

  Comparison

  Options trading activity remains among the highest, placing it with NVDA, AMZN, META, and AAPL on the core retail options list.

  Risks

  Pricing pressure, regulatory credit volatility, declining profitability, and weak consumer demand.

• Semiconductors / NVDA / TSMC / SMCI

  AI hardware and semiconductor earnings chain

  Strengths

  Semiconductor stocks rose about 7% in one week; TSMC earnings and capex guidance were strong, and SMCI rose about 23% on preliminary results plus order/gross-margin upside surprises.

  Weaknesses

  SNDK saw $113M of retail selling, and NVDA’s weekly retail activity was relatively mild.

  Comparison

  Semiconductors outperformed the S&P 500’s roughly 1% weekly gain.

  Risks

  Increased AI equity supply, capex slope, debt-financing absorption capacity, supply-demand balance, and debate over long-term profit margins.

• Retail ETFs

  Retail allocation channel but weakening momentum

  Strengths

  Broad large-cap equity ETFs still saw +$1.4B in inflows, with overall ETF net inflows of +$3.9B.

  Weaknesses

  ETF inflows fell to the 4th percentile, SOXL and SMH were among the more heavily sold ETFs of the week, and tech ETF inflows turned negative.

  Comparison

  Compared with warming single-stock activity, capital momentum at the ETF level has cooled significantly.

  Risks

  If ETF buying continues to weaken, broad market support may decline and style and sector volatility may rise.

Key data

• Weekly retail fund flows$5.7BBelow the 12-month average of $6.8B/week.
• Retail fund flow percentile35.3%ileBasically flat versus the previous week, indicating moderate participation.
• ETF inflow percentile4%ileContinued to decline from 18%ile, with tech ETF flows turning negative.
• Single-stock activity percentile71.8%ileRose from 51%ile, offsetting part of the ETF weakness.
• Retail ETF net inflows+$3.9BStill above single-stock net inflows, but momentum has weakened.
• Retail single-stock net inflows+$1.8BMainly driven by the communications and technology sectors.
• Momentum factor retracement-14%Calculated over the past month using a market-cap-weighted S&P 500 quintile long-short portfolio.
• Extreme negative momentum tail events1 -6 sigma day, 7 observations <= -3 sigma within 18 daysShowing that the crowded-trade unwind process is intense.
• Mag 7 retail buyingNVDA +$412M, TSLA +$319M, MSFT +$202M, AMZN +$177M, META +$21M, AAPL -$114MAAPL was the main net-sold name within the Mag 7.
• Top five retail-bought single stocksSPCX +$514M, NVDA +$412M, GOOGL/GOOG +$328M, TSLA +$319M, NU +$308MConcentrated in AI, technology, communications, and high-attention growth stocks.
• GOOGL earnings event7/22 buy $74.7M, 1.9z; after-hours down about 4%Google Cloud grew 82% year over year, above buy-side expectations of 70-75% and Street at 64%; 2026 capex guidance was raised to $195-205B.
• TSLA earnings event7/22 buy $101M, 0.4zRevenue growth was strong, but earnings missed expectations due to pricing and regulatory credit impacts, and the stock fell slightly after hours.
• TSMC earnings event2026 capex $60-64B, revenue guidance raised to 40%+ YoYAI demand commentary was strong, and gross margin reached 60%+ for the first time.
• Non-retail futures flowsNet selling of about $4.1BMainly driven by roughly $8.3B net selling in NQ, partially offset by net buying in ES and RTY.

Impact & implications

The report’s investment implication is that short-term market risk comes from the rapid unwind of crowded momentum trades and weaker ETF buying, especially as negative tech ETF flows may amplify style volatility; however, AI data centers, semiconductors, hyperscaler capex, and Mag 7 earnings can still attract retail capital and provide localized support. If earnings season continues to confirm cloud growth, AI monetization, and returns on capex, the AI chain may maintain relative strength; otherwise, if consumer sentiment, macro volatility, military activity, or concerns about financing AI supply weigh on guidance, downside sensitivity in crowded themes may increase.

Risks

• The unwind of momentum-factor crowding may continue to trigger high-sigma negative tail events and sharp style selloffs.
• Tech ETF flows have turned negative, potentially weakening the broad buying support previously seen for AI and tech themes.
• An overly steep AI capex curve, financing pressure, and increasing supply may compress valuations across the AI infrastructure chain.
• If hyperscalers cannot prove AI monetization and returns on capex, earnings-season guidance may trigger pullbacks.
• Weakening consumer sentiment, macro volatility, and military activity may weigh on the 3Q outlook.
• Stocks with high social media popularity and high short interest may experience short squeezes or retail chase-related losses.

What to watch

• The latest guidance from hyperscalers such as GOOGL, MSFT, AMZN, and META on AI capex, cloud growth, and AI monetization.
• Whether tech ETF flows remain negative, and whether selling pressure continues in semiconductor ETFs such as SOXL and SMH.
• Whether the momentum factor continues to see -3 sigma or even more extreme negative tail events.
• Retail fund flows and options activity in AI-chain stocks such as NVDA, TSLA, GOOGL/GOOG, SMCI, and TSMC.
• Capital dispersion within the Mag 7, especially whether selling in AAPL expands.
• The slope of retail buying and short-squeeze risk in stocks with high social-media popularity and high short interest.