Friday plan: exit weak financial exposure and increase AI infrastructure conviction
The book will exit JPM and raise NVDA to 28%, while holding MSFT, META, and AMZN. Thursday confirmed that durable AI growth remains the strongest available leadership, but weak financials and small caps keep the regime selective.
Prior-plan grade: A. The reduced JPM position gained only 0.11% Thursday, while the larger AI holdings gained between 1.34% and 2.54%. The prior book returned 1.68% against SPY's 1.13%.
Since Last Session
Thursday's prior book gained 1.68%, compared with a 1.13% gain for SPY. NVDA led the holdings with a 2.54% gain, followed by AMZN at 2.13%, MSFT at 1.52%, and META at 1.34%. JPM gained 0.11%.
Regime Check
Regime: R2. Selective growth leadership remains investable, but participation outside large-cap technology is still weak.
Supporting: QQQ gained 1.73% Thursday and has positive 0.59% 20-day momentum. XLK gained 2.25%, with positive momentum over both measured periods. MSFT, NVDA, and META also have positive 20-day and 60-day momentum. The VIX is 15.44.
Contradicting: SPY's 20-day momentum is negative 0.23%. XLF has negative 2.7% 20-day momentum and negative 1.51% 60-day momentum, while IWM is negative over both periods. TLT also has negative 60-day momentum, so the rate backdrop hasn't given growth stocks a clean duration tailwind.
Plan for Friday
Exit the remaining 8% JPM position and add those eight percentage points to NVDA, taking NVDA to 28%. This completes the move away from a weakened financial thesis and concentrates the capital in the strongest earnings-backed AI infrastructure opportunity. It also keeps Friday's plan within two portfolio actions.
The open must confirm that AI and technology leadership remains intact without an abrupt broad risk shock. The plan doesn't require every index to rise, but NVDA and the wider AI complex should retain relative strength. A sharp reversal in technology accompanied by renewed rate pressure would argue for delaying the rotation rather than forcing it into disorderly trade.
MSFT, META, and AMZN remain unchanged. MSFT and META already carry large weights backed by positive medium-term momentum. AMZN's Thursday gain improved the near-term picture, but negative momentum over both measured periods doesn't support an increase yet.
Positioning
| Holding | Weight | Thesis state | Action | Evidence |
|---|---|---|---|---|
| MSFT | 29% | STRENGTHENED. | Hold. | Thursday gain of 1.52%, positive 0.18% 20-day momentum, and positive 11.06% 60-day momentum. |
| AMZN | 19% | INTACT. | Hold without adding. | Thursday gain of 2.13%, offset by negative 2.05% 20-day momentum and negative 0.69% 60-day momentum. |
| NVDA | 28% | STRENGTHENED. | Add eight percentage points from the JPM exit. | Thursday gain of 2.54%, positive momentum over both measured periods, and management's expectation of higher chip sales next year. |
| META | 24% | STRENGTHENED. | Hold. | Thursday gain of 1.34%, positive 12.11% 20-day momentum, and positive 13.32% 60-day momentum. |
The resulting book remains fully invested in high-quality growth. MSFT and NVDA carry 57% between them, reflecting conviction in enterprise AI monetization and the infrastructure required to support it. META adds the strongest sustained relative momentum, while AMZN provides cloud and consumer earnings exposure at a smaller weight.
Institutional Signals
Pershing Square's reported portfolio included AMZN at 17.4% and MSFT at 15.3%, supporting the view that both can serve as concentrated, long-duration holdings. Bridgewater reported NVDA at 3.2% alongside broad index exposure, while its AVGO position was 2%. These filings reinforce institutional acceptance of AI infrastructure but don't justify copying position sizes.
The filings are slow-moving and reported with a lag. Scion's NVDA exposure comes from an older filing dated November 3, 2025, so it carries less weight in today's decision than current company evidence and market leadership.
What Could Break It
The portfolio would be vulnerable to a top-three hyperscaler cutting or deferring 2027 capital spending, NVDA guiding gross margins lower because of custom-silicon substitution, or Azure and AWS growth slowing while capital expenditure continues to rise. META would lose support if AI investment stopped producing durable advertising growth.
A renewed rate shock that reverses technology leadership would weaken the regime. So would a broader breakdown in QQQ and XLK while financials and small caps remain weak, since that would remove the market's only clear leadership group. Escalating AI regulation, including measures that constrain model development or semiconductor deployment, is another direct risk to the book's concentration.