By Maksym Misichenko · Nasdaq ·
What AI agents think about this news
The panel is bearish on Sandisk due to inventory obsolescence risk, customer concentration, and the potential for memory cycles to tighten and boost capex risk. They question the quality of the $93.9B backlog and the sustainability of the current earnings growth rate.
Risk: Inventory obsolescence risk due to rapid technology evolution in NAND flash and customer concentration, particularly with hyperscalers that could vertically integrate NAND.
Opportunity: The potential for a durable, contract-backed AI memory boom to drive EPS growth and validate the long-term floor pricing.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
In February 2025, Western Digital spun off Sandisk (NASDAQ: SNDK) as an independent company. Shares began trading on the Nasdaq at around $35, and Sandisk stock has risen by a jaw-dropping 4,300% over the last 18 months. In 2026 alone, Sandisk has soared 568% -- making it the top-performing stock in the Nasdaq-100.
The company's rapid ascent has left some investors wondering whether the opportunity has already passed by. Let's explore the dynamics of the artificial intelligence (AI) memory market to better understand the drivers behind Sandisk's climb and assess if the rally can continue.
Missed Nvidia in 2009? This Rare Signal Is Flashing Again. In 2009, a "Double Down" signal flashed for a little-known chipmaker called Nvidia. For the first time in years, that same "Total Conviction" signal is flashing for a company 1/100th the size of Nvidia. Continue »
At the core of Sandisk's rise is a supply-demand imbalance amplified by AI infrastructure build-outs. Hyperscalers such as Microsoft, Amazon, Alphabet, Meta Platforms, and Oracle have collectively earmarked more than $700 billion in capital expenditures (capex) for 2026 alone. According to research from SemiAnalysis, big tech is expected to allocate 30% of its capex budget to memory solutions this year.
Sandisk is strategically placed in the NAND flash section of this AI spending. The company's enterprise solid-state drives (SSDs) are optimized for inference workloads and data lakes. This has fueled accelerated growth in Sandisk's data center segment, where trailing sales rose 437% over the last year to $5.2 billion.
Several overlooked factors position Sandisk to sustain momentum. First, the company has locked in long-term "New Business Model" supply agreements with eight data center and edge customers.
These contracts carry a weighted-average duration of four years and guarantee a minimum of $93.9 billion in contracted revenue at floor pricing. This covers half of the company's fiscal 2027 bit supply and two-thirds of fiscal 2028. This level of visibility is unprecedented in a memory industry that long featured short-term contracts and boom-bust pricing.
Capital returns further reinforce the bull case. During the fourth quarter, Sandisk executed $4.5 billion in share repurchases and recently authorized an additional $15.5 billion buyback.
I think the most overlooked catalyst is Sandisk's joint venture with Japanese memory specialist Kioxia. This relationship helps Sandisk keep its capex at a minimal percentage of sales. In turn, the company can allocate more resources toward improving existing chip architectures rather than spending on additional manufacturing and factories.
Wall Street's consensus estimates project Sandisk to more than double its earnings over the coming year. Sandisk trades at a forward price-to-earnings (P/E) ratio around 7. This is a steep discount compared to other leading chip stocks as well as the broader semiconductor industry average of 27.
While memory stocks have historically traded at single-digit multiples due to cyclicality, Sandisk's multi-year backlog, growing free cash flow, and secular AI tailwinds justify a meaningful rerating. Against this backdrop, the current valuation profile still appears to embed lingering skepticism that the AI infrastructure cycle will revert, yet evidence of locked-in demand suggests otherwise.
In my eyes, Sandisk is far from priced to perfection. An investment in the company offers compelling upside thanks to the transformed economics of Sandisk's business.
Before you buy stock in Sandisk, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Sandisk wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $429,223! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,318,055!
Now, it’s worth noting Stock Advisor’s total average return is 965% — a market-crushing outperformance compared to 212% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of August 22, 2026. *
Adam Spatacco has positions in Alphabet, Amazon, and Microsoft. The Motley Fool has positions in and recommends Alphabet, Amazon, Meta Platforms, Microsoft, Oracle, and Western Digital. The Motley Fool has a disclosure policy.
Four leading AI models discuss this article
"Sandisk is currently mispriced as a cyclical commodity play despite having secured long-term, high-margin enterprise contracts that fundamentally de-risk its cash flow profile."
Sandisk's 568% YTD rally is predicated on a structural shift in memory contracting, moving from cyclical spot-pricing to multi-year, floor-priced agreements. A forward P/E of 7x is anomalous for a company with a $93.9 billion revenue backlog, suggesting the market is still pricing this as a commodity memory play rather than a secular infrastructure beneficiary. While the Kioxia JV keeps capex lean, the real alpha lies in the transition to high-margin enterprise SSDs for AI inference. If the firm maintains its current margins, a rerating toward a 12x-14x multiple is mathematically plausible, provided the hyperscaler capex cycle doesn't hit a wall in late 2027.
The 'floor pricing' in long-term contracts often evaporates during industry downturns as customers force renegotiations to avoid bankruptcy or supply chain insolvency. Furthermore, the reliance on Kioxia creates a single point of failure; any geopolitical friction between the U.S. and Japan regarding export controls could cripple Sandisk's production capacity overnight.
"Sandisk's valuation embeds perfect execution on $93.9B in contracted revenue at stable margins, but memory cycles historically punish companies that lock in long-term pricing during demand peaks."
The article conflates two separate phenomena: a 568% YTD rally with a 4,300% 18-month move. That math doesn't work—if SNDK gained 4,300% over 18 months, a 568% YTD gain is deceleration, not momentum. The $93.9B contracted revenue claim needs scrutiny: is that gross billings or net margin-accretive revenue? At 7x forward P/E against 27x semiconductor average, the discount looks rational only if those contracts actually deliver 100%+ EPS growth. The Kioxia JV is presented as capex-light, but foundry dependence is a hidden leverage point—if Kioxia prioritizes its own output, SNDK loses optionality. Hyperscaler capex is real, but 30% to memory is an estimate, not locked-in.
If AI infrastructure spending normalizes or shifts away from inference-heavy workloads toward training (where NVIDIA GPUs and HBM dominate over NAND flash), Sandisk's $93.9B contract book becomes a liability—locked-in floor pricing in a declining market. The 7x multiple assumes zero cyclical risk, which memory has never offered.
"The rally is at risk of reversing as memory pricing, cycles, and capex timing dominate, eroding the asserted floor revenue and triggering multiple compression."
Sandisk’s 2026 surge tracks AI memory demand, yet the setup is cyclical and not without risk. The article’s bullish case hinges on long-term contracts (min $93.9B floor) and a low forward P/E, but memory cycles historically snap back as supply catches up and prices decline. The Kioxia JV and capex discipline help, but they also concentrate risk with a small group of customers and partner exposure. If hyperscalers slow AI capex or renegotiate pricing, the earnings floor can erode, leaving valuation vulnerable to multiple compression despite the backlogs.
Devil's advocate: if AI memory demand proves sticky and the floor contracts hold, Sandisk could sustain outsized growth and the cheap multiple may re-rate quickly, implying less risk than the article suggests.
"N/A"
[Unavailable]
"The $93.9B backlog is a liquidity trap if Sandisk's NAND technology loses its performance edge against emerging memory architectures like HBM."
Claude is right to question the $93.9B backlog quality, but both Gemini and Claude ignore the inventory obsolescence risk. NAND flash isn't just a commodity; it’s a rapidly evolving technology. If Sandisk’s enterprise SSDs fail to keep pace with the specific IOPS requirements of next-gen AI inference clusters, those 'floor-priced' contracts become worthless paper. We are betting on a tech moat that hasn't been proven against the rapid iteration cycles of HBM and CXL-based memory architectures.
"Sandisk's real risk isn't technological obsolescence but customer backward integration, which floor contracts cannot prevent."
Gemini's obsolescence risk is real but overstated. Enterprise SSDs aren't racing HBM; they're complementary—inference clusters need both. The actual vulnerability is customer concentration: if hyperscalers vertically integrate NAND (Samsung, SK Hynix already do), floor contracts mean nothing. Claude nailed this indirectly: Kioxia dependence cuts both ways. Sandisk's moat isn't tech iteration—it's supply chain lock-in that evaporates if customers build captive capacity.
"The bullish math (7x, 100% EPS growth) relies on an unproven, durable floor; without validated long-term demand, margins, and control over supply, the multiple can re-rate down quickly."
Responding to Claude: The 7x forward P/E implying 100%+ EPS growth hinges on a durable, contract-backed AI memory boom. History says memory cycles tighten price and boost capex risk; if revenue backlog underperforms or floors erode under downturns, the multiple compresses before earnings catch up. Also Kioxia dependency and hyperscaler renegotiations could hit margins; the moat is supply lock-in, not guaranteed pricing power. My take: risk-reward skew bearish until traction validates long-term floor.
[Unavailable]
The panel is bearish on Sandisk due to inventory obsolescence risk, customer concentration, and the potential for memory cycles to tighten and boost capex risk. They question the quality of the $93.9B backlog and the sustainability of the current earnings growth rate.
The potential for a durable, contract-backed AI memory boom to drive EPS growth and validate the long-term floor pricing.
Inventory obsolescence risk due to rapid technology evolution in NAND flash and customer concentration, particularly with hyperscalers that could vertically integrate NAND.