The July Flush Was a Fund, Not a Fact

The most dangerous misread in the memory chip market right now is treating July as a verdict.

It was not. SanDisk closed July 30 at $1,279.96, down from a June 25 record high near $2,335, after closing at $1,015.89 on July 29, when Leopold Aschenbrenner’s Situational Awareness fund was forced to unwind its public-equity book after margin calls from its prime brokers. The stock did not collapse because NAND demand softened or because a single contract unraveled. It collapsed because a single leveraged fund ran out of margin.

Situational Awareness was holding SanDisk as a top position at roughly 4x gross leverage, absorbing what SpotGamma calculated as a roughly 120% hit to equity from a roughly 30% long-book decline. Margin calls from Goldman Sachs, JPMorgan, and Bank of America followed. When the position flushed to $1,015.89, dealers who had sold the fund call options unwound their long-stock hedges alongside the equity, a mechanical accelerant that turned selling into cascading selling. The 26% single-session recovery off that low, beginning the moment the block cleared, is the most precise measurement available of how much the final leg was flow rather than fundamentals.

Why Wall Street Cares

The distinction matters because institutional investors evaluating SanDisk today are not pricing a deteriorating business. They are pricing a business with one of the strongest quarterly performances in semiconductor history, sitting well below its high because a fund blew up.

SanDisk reported fiscal Q4 2026 revenue surging 372% year-on-year to $8.965 billion, with growth driven by the Datacenter and Edge segments while gross margins expanded to 84.6%. Adjusted EPS reached $39.25. Full-year fiscal 2026 revenue was $20.25 billion, up 175% year-on-year. These are not numbers a company posts during a cycle that is quietly rolling over.

The Q1 2027 guide is the number that should be anchoring this conversation. Management expects first-quarter 2027 revenue in the range of $10.30 billion to $10.80 billion, with non-GAAP diluted EPS in the range of $44.00 to $46.00. That is sequential revenue growth on top of an already-record quarter, guided by a management team that has beaten its own guidance four consecutive times.

The Bull Case

The structural argument begins with supply and ends with contracts. Deutsche Bank projects 2026 DRAM demand of roughly 2,261k WSPM against capacity of 2,051k WSPM, a 10% shortfall. By 2028, demand climbs to 3,563k WSPM while capacity reaches only 2,769k WSPM, a 29% gap. Deutsche sees the imbalance narrowing to 18% in 2029 and 11% in 2030, meaning shortage conditions persist for another four years.

SanDisk’s response to that dynamic is not to simply ride the cycle. It is to contractually lock it in. The company says it has signed 10 New Business Model agreements with eight customers representing $93.9 billion in minimum expected revenue assuming floor pricing, and those agreements are expected to represent more than 50% of bits in fiscal 2027 and approximately two-thirds in fiscal 2028. The financial guarantees backing these agreements total $16.5 billion.

That $93.9 billion is not a projection. It is a floor, the minimum customers are contractually obligated to pay under the agreements assuming the floor pricing terms apply. A commodity company that has negotiated downside protection for the majority of its production over a multi-year window is no longer being valued correctly when the market assigns it a forward multiple implying it will revert to commodity margins.

The geopolitical picture added a new dimension this month. Reports in late July said Apple had been testing memory from Chinese suppliers including CXMT and YMTC to ease shortages. That backdrop has kept Washington’s scrutiny on China-linked memory supply in focus, with non-Chinese suppliers positioned to benefit if U.S. pressure further constrains Chinese participation in global memory supply chains.

The Bear Case

The bear case is also coherent, which is why the analyst spread is unusually wide. Wall Street’s outlook on SanDisk remains far from unified. JPMorgan analyst Harlan Sur reinstated coverage with an Overweight rating and a $2,250 price target. Wells Fargo lowered its target to $1,400 from $1,620, signaling concerns about valuation and whether current market conditions can remain favorable.

The core skeptical argument is cycle recognition, not business quality. The memory chip market is cyclical, and once the supply crisis is resolved for memory chip makers, prices should come down. Nobody knows to what level they will fall. The market is skeptical about SanDisk’s unknown future, which is why the stock is priced so low relative to current earnings. Bears argue that 84.6% gross margins are a cycle peak by definition, and that the NBM agreements, while protecting against the worst-case scenario, do not eliminate the eventual normalization of NAND pricing.

Despite weaker momentum in smartphones and retail spot markets, major manufacturers are maintaining strict capital discipline while accelerating transitions to high-layer advanced nodes. The consumer segment tells a different story than the datacenter. Consumer revenue declined 32% sequentially in Q4, reflecting shifts in market equilibrium and pricing dynamics. If spot markets continue softening, the argument that enterprise pricing holds indefinitely becomes harder to make.

The Evidence

The most revealing data point is not in the income statement. It is in the technology stack.

On August 3, SanDisk and SK hynix released the High Bandwidth Flash technical specification through the Open Compute Project, advancing the standardization effort just six months after the consortium began work in February. Notably, Google and Tenstorrent joined as consortium members during the standardization process, contributing to technology validation. HBF is not a product launch. It is an attempt to define a new memory tier between HBM and traditional SSDs, and getting Google to validate the specification is the closest thing to a hyperscaler pre-commitment available before mass production begins.

HBF is designed to bridge the performance gap between HBM and traditional SSDs, with the companies describing bandwidth ranging from 0.4 TB/s to 3.0 TB/s and capacities of up to 512 GB. The AI industry is shifting from training to inference, which runs continuously across millions of devices, making these workloads memory-bandwidth and capacity hungry and increasingly power constrained. SanDisk is positioning its technology at the exact inflection point that shift creates. Mass production is not here yet, but SK hynix has said it plans to begin mass production of enterprise SSDs based on the technology early next year.

On X, Peter Diamandis wrote that “Memory, not compute, is the rate limiter of the Agentic Era.” Elon Musk replied: “Few realize this.” The observation is correct, and it is the structural thesis in two sentences.

The Mavens’ View

Professional investors are split on time horizon, not direction. The funds that added to positions through July’s chaos understand that the forced-selling event created a discount to intrinsic value. The funds that trimmed understand that even a discounted intrinsic value depends on assumptions about cycle duration that nobody can prove.

David Tepper trimmed a closely watched AI-infrastructure basket tied to the same theme, yet the value of Appaloosa’s remaining stake still rose sharply between March 31 and June 30. That is not a bearish data point. It is a data point about position sizing at cycle peaks. Tepper trimmed and still benefited from the run.

Several major tracking services show SanDisk still carries a Strong Buy consensus, with an average 12-month price target around the low-$2,000s. The range of individual targets, often spanning roughly $1,000 to $3,000, reflects genuine intellectual disagreement about the durability of the current cycle, not a broken investment case.

SanDisk’s New Business Models provide over four years of visibility, with more than 50% of FY2027 bits and approximately two-thirds of FY2028 bits already committed, backed by $16.5 billion in financial guarantees. The argument that this is still a pure commodity cycle becomes harder to sustain when the majority of near-term production is locked into floor-priced, multi-year agreements with third-party financial guarantees. Commodity companies do not negotiate those terms. Companies with pricing power do.

What Investors Are Missing

The consensus debate is framed around whether current NAND prices are sustainable. That is the right question for spot-market watchers. It is the wrong question for investors evaluating SanDisk’s actual business.

The question that deserves more attention is what happens to SanDisk’s competitive position when HBF enters mass production. HBF is a new storage tier designed to sit between High Bandwidth Memory and traditional SSD storage. It is not intended to replace either technology but to act as a high-performance bridge that eliminates bottlenecks in AI systems. SanDisk is a primary co-author of that standard. If HBF adoption follows the trajectory that Google and Tenstorrent’s participation suggests, SanDisk does not just benefit from higher NAND prices. It becomes the infrastructure layer for a new category of AI inference hardware.

The core SanDisk thesis is that AI-driven data center demand lets the company lean on higher-value enterprise SSDs and long-term hyperscaler deals instead of pure commodity NAND. HBF is where that thesis extends beyond the current cycle. Nobody is pricing that yet because HBF has no revenue. But Google does not join open-source standards consortia for products it does not intend to deploy.

The Apple-China testing reports add a second overlooked dimension. If Washington keeps tightening the aperture on Chinese memory supply, the demand those suppliers might have served has to go somewhere. SanDisk is positioned to be one of the beneficiaries.

Stocks to Watch

SanDisk (SNDK). The stock is still up several hundred percent year-to-date, yet down roughly 45% from its late-June peak. At 7.6 times forward earnings, this is not a stock the market is pricing for durability. The NBM agreements, the HBF standard, and any further policy pressure on China-linked memory supply are three separate catalysts that are not yet reflected in that multiple.

Micron Technology (MU). Micron closed at $940.76 on August 18. Micron is the DRAM complement to SanDisk’s NAND story. The same structural shortage applies. The same Apple demand redirection logic applies if policy pressure rises on China-linked supply.

SK hynix (SKHY). SK hynix’s ADRs began trading on Nasdaq on July 10. As SanDisk’s co-author on the HBF standard, hynix is the most direct second derivative of any HBF adoption wave. Its Nasdaq listing also means institutional access that did not exist in the same form earlier this year.

Western Digital (WDC). Western Digital shares rose about 4% on August 17, part of the same memory-sector move that lifted several storage and memory-linked names. WDC is the one legacy relationship that SanDisk’s spin-off left intact, and its HDD business provides a floor the pure-play NAND names lack. For investors unwilling to take full commodity-cycle risk, WDC is the more conservative expression of the same demand story.

Nebius Group (NBIS). Nebius reported Q2 2026 revenue of $582.3 million in mid-August. Nebius is not a memory company. It is the demand signal. When its backlog grows, SanDisk’s order book grows with it.