Nancy Pelosi's 3 Years of Stock Trades: Where Did the Money Actually Go?

A data-driven breakdown of every disclosed trade from May 2023 to January 2026 — mapped by fund size, sector, and timing.

Nancy Pelosi is the most-watched trader in Congress — not because she holds any committee power over the companies she buys, but because her portfolio has an uncanny habit of outperforming. Her husband Paul Pelosi executes most of the trades (disclosed as "Spouse"), and every move gets filed under the STOCK Act.

We pulled three full years of her disclosed trades — from the NVIDIA calls in late 2023 through the January 2026 batch — and ran the numbers. The picture that emerges is not random. It's a concentrated, AI-and-semiconductor-heavy portfolio with very deliberate position cycling.

Here's the full breakdown.

The Big Picture: Where Was Capital Deployed?

The STOCK Act requires disclosure in size ranges, not exact dollar amounts. We use the midpoint of each range to estimate allocation. Across all buy-side trades from May 2023 to January 2026, total estimated capital deployed was in the range of $15 million to $45+ million, with the following breakdown by stock:

Note: Percentages are estimates based on midpoints of disclosed size ranges. "Size" ranges used: <1K=0.5K, 15–50K=32.5K, 50–100K=75K, 100–250K=175K, 250–500K=375K, 500K–1M=750K, 1M–5M=3M, 5M–25M=15M.

The NVIDIA Obsession: A Play-by-Play

No stock tells Pelosi's story better than NVIDIA. She (through her husband's account) built, trimmed, and rebuilt the NVDA position seven times across the three-year period. This isn't passive investing — it's active cycling that generated substantial returns at nearly every turn.

Sector Allocation: It's an AI Portfolio

When you group by sector rather than individual stock, the thesis becomes crystal clear. This is a concentrated bet on the AI infrastructure supercycle.

NVIDIA and Broadcom together represent 55% of all buy-side capital — both pure-play AI chip companies. Add Palo Alto Networks (AI-enhanced cybersecurity) and you're at ~64% in AI infrastructure. The mega-cap tech holdings (GOOGL, AMZN, AAPL, MSFT) are all AI-adjacent. Even Vistra Corp, the power utility, fits the narrative — data centers need electricity.

The Sell Side: What She Got Out Of

The sell trades are equally revealing. Here's what was exited across the three years — and how well the timing held up:

Broadcom: The Sleeper Position

NVIDIA gets the headlines, but Broadcom (AVGO) may be the most interesting call in the entire three-year window. Pelosi bought AVGO twice — once in June 2024 at roughly $156.53 (adjusted entry) and again in June 2025 at an adjusted entry of $248.56. Current price: $425.28.

The June 2024 Broadcom buy came just as the market was starting to price in Broadcom's custom AI chip (XPU) business for hyperscalers like Google and Meta. The June 2025 add came as that thesis was being confirmed with massive earnings beats. Combined, the two Broadcom positions represent an estimated $6M–$10M in capital — easily a 70%+ return on the 2024 tranche.

The Apple Cycle: Buy Low, Sell High, Repeat

Apple is the most actively cycled position in the portfolio, with multiple buys and sells across the three years:

• Jun 2023: Buy $250K–$500K at $80 (adjusted) — huge entry, stock was around $185 split-adjusted
• Dec 2024: Sell $5M–$25M at $250.42 — locked in massive gains from the 2023 entry
• Dec 2024: Also buy $250K–$500K at $272.82 adjusted — rebuilding a smaller position
• Dec 2025: Sell $5M–$25M at $273.81 — near-flat from the Dec 2024 re-entry

Palo Alto Networks: Accumulated Over 14 Months

Palo Alto Networks was built patiently across three separate buys spanning more than a year — a sign of genuine conviction rather than a momentum chase:

• Feb 12, 2024: Buy $500K–$1M at adjusted $185.99
• Feb 21, 2024: Add $100K–$250K at adjusted $130.99
• Dec 20, 2024: Add $1M–$5M at $100.00/share (likely options, adjusted entry $186.78)

Current price: $183.67 — roughly flat on the position. PANW has been a volatile ride, but the Pelosis stayed in it. The cybersecurity thesis remains intact given the company's platform consolidation strategy.

The Options Question

Several trades show a disclosed price-per-share that is dramatically lower than the market price of the underlying stock at the time. This is a strong indicator of options contracts rather than direct stock purchases:

The Wildcard: Private & Illiquid Assets

Three trades in the dataset don't have tickers at all — meaning they're investments in private funds or LLCs:

• FORGE INVESTMENTS LLC — $1M–$5M buy in March 2024 (Spouse). No public market data.
• REOF XXVI LLC — $250K–$500K buy in August 2024 (Spouse). Likely a real estate fund.
• VERSANT MEDIA GROUP INC — <$1K exchange in Jan 2026 (Undisclosed). Tiny, likely a corporate restructuring.

Tempus AI: The Boldest Small-Cap Bet

Tucked into the portfolio is a $50K–$100K position in Tempus AI (TEM), a health-data AI company that IPO'd in mid-2024. The Pelosis bought it twice — once in January 2025 at an adjusted entry of $31.83 and again in January 2026 at $20/share (likely options, adjusted entry $70.33).

Current price: $53.49. The January 2025 tranche is up ~68% from entry. This is the portfolio's highest-risk, highest-upside conviction bet — a small AI healthcare company backed by Eric Lefkofsky (co-founder of Groupon). The position size is small relative to the rest, but the timing of the Jan 2025 entry — just as TEM was recovering from its post-IPO lull — was sharp.

Vistra Corp: The AI Power Play

Vistra Corp (VST) is the most unusual holding — it's an independent power producer, not a tech company. But the thesis is straightforward: AI data centers consume massive amounts of electricity, and Vistra's nuclear and natural gas fleet is positioned to supply it.

The Pelosis bought VST twice — in January 2025 (adjusted entry $169.35) and January 2026 (adjusted entry $166.37). Current price: $158.21 — currently slightly underwater on both entries. VST has been volatile, trading as high as $180+ before pulling back on power market uncertainty.

What 3 Years Tells Us

Three years of data reveals a coherent, concentrated strategy. A few key takeaways:

1. AI infrastructure is the core thesis. NVDA + AVGO = 55% of buy-side capital. This isn't diversification — it's conviction. Every major buy over three years has been tied, directly or indirectly, to the AI buildout.

2. Year-end is the active trading window. The vast majority of sells and new buys cluster in December–January, consistent with sophisticated tax-loss harvesting and portfolio rebalancing. This isn't reactive to market news — it's planned.

3. Options are used for leverage. Multiple trades at prices far below market value confirm options usage. This amplifies returns in bull markets but creates hidden leverage that the disclosed "size" ranges don't fully capture.

4. Position cycling is the strategy. Apple, NVIDIA, and GOOGL are all bought, trimmed, and re-entered. This isn't buy-and-hold — it's active position management around core convictions.

5. The sells have been more mixed than the buys. Early exits from Tesla, GOOGL, and Visa left significant returns on the table. The buy timing has been consistently excellent; the sell timing less so.

Whether you view this as exceptional investment acumen, pattern recognition built on decades inside the corridors of power, or simply a well-managed portfolio with a clear secular thesis — the data speaks for itself. Three years. One dominant theme. Semiconductors, AI, and infrastructure. And it's worked.

Want to track Nancy Pelosi's trades in real time? See all Pelosi trades on Polytick →