The Memory Stock Congress Is Buying and the One It Is Ignoring: A Tale of Two Markets

The Consensus Story: Politions Are Bidding Up Memory Stocks

Recent headlines focus on which memory stocks members of Congress are buying. The narrative is simple: lawmakers are piling into memory companies, confirming that AI-driven demand is powering the semiconductor recovery. Four representatives and one senator disclosed purchases of Micron Technology in 2026, with the largest position — Representative Tony Wied of Wisconsin — in the $250,000 to $500,000 range as of February 19. Representative Cleo Fields of Louisiana committed $100,000 to $250,000 on February 3. Smaller positions appeared from Representative Gilbert Cisneros of California and Senator John Fetterman of Pennsylvania, each in the $1,000 to $15,000 bracket.

SanDisk, which began trading independently on February 24, 2025, after its spin-off from Western Digital, has attracted far less political capital. Representative Ro Khanna bought shares below $50. Representative Lisa McClain executed a series of $1,000-to-$15,000 trades in October 2025. Representative Josh Gottheimer purchased a position in May 2026. The amounts are token. The disparity in political conviction between the two names is the first structural signal.

This Is Not a Single Memory Market. It Never Was.

The market treats Micron and SanDisk as co-equal beneficiaries of the AI memory boom. They are structurally different businesses competing in two distinct sub-markets, with different capex profiles, different constraint dynamics, and different paths to sustainability. The political money flow — concentrated in Micron, peripheral to SanDisk — tracks the AI-narrative thesis. It does not track the supply-side reality.

Micron is the last pure-play DRAM manufacturer in the United States. Its products — DRAM and high-bandwidth memory (HBM, the stacked memory architecture that feeds GPU compute) — sit at the point of maximum constraint in the AI data center. SanDisk is a NAND flash storage business that emerged from Western Digital's $20 billion acquisition in 2016, reconstituted as an independent company after a February 2025 spin-off. Its products serve consumer storage, enterprise SSDs, and embedded memory — important markets, but not the bottleneck.

Where the constraint sits determines who has pricing power. Right now, it sits in DRAM and HBM supply.

Table 1: Micron vs. SanDisk — The Structural Divergence

Metric	Micron (MU)	SanDisk (SNDK)
Current Price	$872	$1,231
YTD Return	+205%	+418%
Rolling 1-Year Return	+597%	+2,592%
52-Week Range	$113 – $1,255	$43 – $2,354
Revenue Growth YoY	+167%	+175%
Gross Margin	72.6%	71.5%
Operating Margin	65.6%	61.2%
Free Cash Flow Margin	29.0%	56.8%
Capital Expenditure (TTM)	$25.3B	$177M
Net Debt	-$20.3B (net cash)	$0
ROIC	58.6%	80.6%
Q2 Revenue vs. Consensus	$23.86B vs. $19.97B (+19.2%)	$3.03B vs. $2.69B (+12.7%)
Q2 EPS vs. Consensus	$12.20 vs. $9.19 (+32.8%)	$6.20 vs. $3.62 (+71.3%)

The table reveals the bifurcation. SanDisk's free cash flow margin — 56.8% — is extraordinary. Its ROIC of 80.6% dwarfs Micron's 58.6%. On a per-dollar-of-revenue basis, SanDisk is the more efficient business today. But its capital expenditure of $177 million over the trailing twelve months is not a business strategy; it is the residual of a spin-off that left the heavy manufacturing infrastructure with Western Digital. SanDisk is a fabless NAND design and branding operation sitting on a parent's capacity. That margin structure is the definition of unsustainable.

Micron's $25.3 billion in annual capex is not an expense — it is the supply discipline mechanism. That investment, announced as part of a $200 billion commitment to U.S. manufacturing and R&D in June 2025, covers fabs in Idaho, New York, and Virginia. It is what keeps DRAM and HBM supply constrained. It is what sustains the 72.6% gross margin.

The Recovery Is Not Driven by Demand. It Is Driven by Supply Discipline.

Consensus framing attributes the memory recovery to AI data center demand. The demand is real — AI infrastructure is consuming a disproportionate share of global memory output. But the supply side is the actual driver of pricing power, and it operates through two separate channels in these two companies.

In DRAM and HBM, the market has only three meaningful suppliers: Samsung, SK Hynix, and Micron. Post-2022, all three learned to constrain supply rather than chase volume. Capex was disciplined. Capacity additions were delayed. Technology migration to high-margin HBM products absorbed incremental investment without expanding the addressable market for standard DRAM. The result is a structural shortage. Prices rose because capacity did not, even as demand improved.

In NAND flash, the supplier base is broader. Samsung, SK Hynix, Micron, Kioxia, and Western Digital all produce NAND. SanDisk's NAND products feed a market where capacity has not been constrained to the same degree. SanDisk's current profitability reflects a temporary window — the post-spin-off period where it is harvesting margin from a supply chain it does not directly control. Once NAND manufacturers expand capacity to chase the same AI-adjacent demand, that 56.8% FCF margin compresses.

The Political Money Tells You Which Narrative Congress Believes

The congressional trading data is not a signal to follow. It is evidence of which narrative has institutional traction. The lawmakers who bought Micron for hundreds of thousands of dollars are buying the AI-HBM story. They are correct that Micron is positioned at the constraint. But they may be incorrect about sustainability.

The lawmakers who bought SanDisk for a few thousand dollars are participating in the tailwind without conviction. The spin-off created a stock that went from $35 in early 2025 to $2,354 at its 52-week high — a 66-fold return. No congressional position captured more than a trivial fraction of that move. The token investments reflect a market that recognizes the return but does not understand the mechanism.

Nancy Pelosi's history with Micron offers a cautionary frame. She purchased call options in December 2021 — investing between $250,000 and $500,000 — and sold them for between $15,000 and $50,000 in September 2022, losing at least $200,000. She missed the subsequent rally to current levels. The timing was wrong. The structural thesis about DRAM supply constraint was correct but executed too early, during the cycle trough when the market had not yet turned.

The Forward Question Is Not About Demand

The memory recovery has been priced into Micron's $1 trillion market capitalization and SanDisk's 2,592% rolling return. The question for investors is not whether demand will persist. It is whether supply discipline holds.

For Micron, the sustainability test is whether $25 billion-plus in annual capex translates into capacity that actually reaches market or remains absorbed by yield ramp and technology migration. The company has raised its U.S. manufacturing commitment to more than $250 billion through 2035. If that capital deploys efficiently, the DRAM constraint extends. If execution drags — if fabs come online faster than planned — the pricing power erodes.

For SanDisk, the sustainability test is structural. A fabless NAND business with $177 million in capex cannot maintain 56.8% free cash flow margins once the NAND cycle turns toward capacity expansion. The parent's spin-off was a financial event, not a structural one. SanDisk's profitability reflects a temporary arbitrage between its lightweight cost structure and the NAND supply it inherits. That arbitrage closes when NAND manufacturers compete on volume.

Investor Takeaway

The congressional money flow into Micron confirms the institutional belief that DRAM and HBM supply constraint will persist. That belief has structural merit — there are only three players, and none has an incentive to flood the market. But the $25 billion capex commitment is a double-edged sword. It sustains pricing discipline today but represents a massive overhang if the cycle turns and capacity comes online faster than demand absorbs it.

SanDisk's performance is the more interesting case. It has delivered the better return — 418% year-to-date versus Micron's 205% — but the mechanism is temporary. The fabless model that produces 56.8% FCF margins today produces normalized margins in the 15% to 20% range once the NAND capacity cycle completes. The stock's 2,592% rolling return reflects a post-spin-off repricing, not a durable operating advantage.

The key issue is not whether AI demand for memory remains healthy. The more important question is whether DRAM suppliers maintain the capex discipline that currently supports pricing, and whether NAND capacity expands fast enough to compress SanDisk's temporary margin structure. Micron's thesis depends on supply restraint. SanDisk's thesis depends on supply not catching up. Both are supply stories. Neither is a demand story.

Watching Micron's capex deployment rate and DRAM ASP trajectory through the end of 2026 is the more consequential check. SanDisk's margin compression risk is a second-derivative signal — but it will arrive faster than most investors expect.