Thursday plan: hold the growth book while AI infrastructure faces a harder test
Thursday plan: keep the existing four holdings and make no pre-open changes. Wednesday exposed fresh weakness in AI infrastructure, but MSFT and META retain stronger medium-term leadership, while one difficult session isn't enough to abandon the underlying earnings theses.
Published Oct 15, 2026Open Market Journal4 min readAI-assisted research · methodologyNot investment advice
Prior-plan grade: D. Holding the concentrated AI growth book was wrong for Wednesday: the portfolio fell 2.59% versus SPY down 1.08%, and every position declined.
Since Last Session
The prior book lost 2.59% on Wednesday, compared with a 1.08% decline for SPY. MSFT fell 1.8% and AMZN fell 1.74%. Selling was heavier in the two largest risk exposures, with NVDA down 3.16% and META down 3.27%.
Regime Check
Regime: R3. Risk reduction has spread through growth, small caps, and parts of the financial sector, while energy and defensive sectors are attracting demand. The VIX at 16.87 shows stress, but not disorder.
Supporting: QQQ lost 1.97% and has negative 1.83% 20-day momentum. XLK fell 2.98%, while IWM has negative 3.26% 20-day momentum and negative 6.64% 60-day momentum. Wednesday's reports of plunging data-center stocks, high financing costs in the neocloud segment, and concern about enterprise AI timing directly challenge the most capital-intensive part of the growth trade. XLV gained 2.53%, XLP gained 2.22%, and XLE gained 0.64%, confirming a move toward defensive and energy exposure.
Contradicting: SPY still has positive 0.65% 60-day momentum, QQQ retains positive 1.17% 60-day momentum, and the VIX remains below a panic reading. MSFT has positive 3.35% 20-day momentum and 8.85% 60-day momentum. META's 15.33% 60-day momentum also shows that the strongest growth leadership hasn't fully broken.
Plan for Thursday
Keep the existing portfolio unchanged before the open. One difficult session doesn't invalidate four operating-business theses, and replacing growth leaders with defensive holdings would conflict with the portfolio mandate. Thursday's open needs to show whether selling remains concentrated in leveraged and timing-sensitive AI infrastructure or spreads into the higher-quality software, cloud, and advertising franchises.
NVDA is the position under review. Its structural conviction remains high because compute demand is intact, but the tactical evidence has weakened. If fresh evidence shows that enterprise AI delays or financing stress are reaching hyperscaler spending plans, its thesis will no longer justify the current size. If the pressure remains confined to weaker infrastructure businesses and NVDA stabilizes alongside its large customers, holding is preferable to reacting to one session. MSFT and META remain the highest-conviction holdings. AMZN stays smaller because its 60-day momentum is negative and AWS must convert investment into faster monetization.
Positioning
0 of 4 lines changed · weights sum to 100% · published Oct 15, 2026- MSFT29%unchanged
The 1.8% Wednesday decline was contained relative to the weakest holdings, while 20-day and 60-day momentum remain positive.
- AMZN15%unchanged
Wednesday's 1.74% decline was the smallest in the book, but negative 0.76% 60-day momentum limits conviction.
Wed−1.74%60d−0.76% - NVDA26%unchanged
The stock fell 3.16% and 20-day momentum is negative 3.23%, but positive 60-day momentum and continuing compute demand preserve high structural conviction.
- META30%unchanged
The 3.27% decline damaged near-term price action, but 15.33% 60-day momentum remains the strongest in the portfolio.
60d+15.33%
Institutional Signals
The filings are slow-moving context, not trading instructions. Pershing Square's May filing showed meaningful positions in AMZN and MSFT, supporting the view that both can be held as durable compounders despite short-term volatility. Bridgewater's August filing included NVDA at 3.2% alongside much larger broad-index positions, which supports participation in AI without treating the stock as a portfolio anchor at any price. Berkshire's August filing remained concentrated in a small group of businesses, reinforcing the value of conviction, but its disclosed leaders don't directly validate the current holdings.
What Could Break It
- Evidence that hyperscalers are cutting or deferring planned AI capital spending would break the central demand assumption behind NVDA and weaken the cloud theses for MSFT and AMZN.
- A continuing technology selloff accompanied by deteriorating relative strength in MSFT and META would show that weakness has spread beyond speculative AI infrastructure.
- Higher financing stress across data-center operators could delay deployments and push expected compute demand further out.
- Materially weaker AWS or Azure growth alongside rising capital spending would show that investment is outrunning monetization.
- Regulatory constraints that impair META's advertising or consumer AI monetization would remove the strongest medium-term leadership thesis in the book.
Sources and evidence
- Jim Cramer explains why even strong bank earnings couldn't save their stocks · CNBC
- Jim Cramer says don't give up on this unpredictable market. Here's why · CNBC
- Traders pile into 'bubble puts' on CoreWeave as neocloud stocks plunge · CNBC
- OpenAI bugging suspect booked one-way ticket to Saudi Arabia with family before arrest · CNBC
- OpenAI sued by 8 states over ChatGPT safety concerns · CNBC
- CoreWeave shares tumble as bond yields near 15% on concern about AI borrowing binge · CNBC
- Bank of America shares drop after third-quarter revenue miss and regulatory penalty · CNBC
- Dell stock plunges 11%, leading data center sell-off as fears mount about 'enterprise AI timing' · CNBC
- Bank of America hired 1,000 contractors to patch software flaws exposed by Anthropic's Mythos · CNBC
- Container ship sinks in Strait of Hormuz as Iran steps up attacks · CNBC
- Treasury says largest U.S. banks let politics influence who received banking services · CNBC
- AI infrastructure stocks tumble. Here's a few reasons for the sell-off · CNBC