In my years of watching markets and analyzing financial flows, I have rarely encountered situations where a fundamental revaluation of a business occurs with such aggressive speed. The shares of SanDisk (SNDK) accomplished an incredible leap: from depressed levels of $45 to levels above $900. The capitalization of the company surpassed the mark of $135 billion.
Presently, ahead of the Q1 2026 earnings report, the market is holding its breath.
The figures for the fourth quarter of 2025 have already outlined the scale of this paradigm shift: a jump in revenue to $3 billion and a takeoff of net profit from deep losses to $800 million. A natural question arises from the perspective of value investing: is this merely a speculative bubble on the hype of artificial intelligence or a fair valuation for a business that has transitioned into a qualitatively new state? I assume the latter.
The market did not simply pump the stock price. It rearranged the company onto a new fundamental step.
The Core Thesis: Beyond Computing Capacity
To understand the nature of the revaluation of SanDisk, it is necessary to look at how the physical architecture of modern data centers changed. The rules of the game in the AI hardware market fundamentally changed.
In 2023–2024, the industry was in the training phase. The absolute focus of attention was fixed on computing capacities—graphic processors. Data was simply absorbed by models during the training process. In this very period, the broad memory market was grappling with a deep surplus, suffocating from overproduction and low prices, due to which investors erroneously considered this sector dying and cyclical.
They were fundamentally wrong.
By 2026, we have firmly transitioned into the era of inference and mass exploitation of AI. Artificial intelligence, from a consumer of data, turned into the primary driver of data generation. Synthetic data for the additional training of other models, HD video, endless logs of multimodal dialogues—this volume of information is growing exponentially.
All this mass of data needs to be continuously recorded and stored.
The "Bottleneck" of Data Storage
The industry quickly realized that the traditional architecture of equipment is completely not ready for such a tsunami of data. Ordinary hard drives in server racks became—or rather, proved to be—a fatal bottleneck.
Modern AI models with trillions of parameters demand the regular saving of states, forcing the system every 30 minutes to dump arrays in terabytes, where reliance on slow HDDs literally paralyzed work, forcing ultra-expensive GPUs to stand idle waiting for recording.
Simultaneously with this, data centers hit the Memory Wall. The problem of the shortage of local video memory for supporting huge context windows was solved by offloading a part of the "consciousness" of AI onto ultra-fast Enterprise SSDs. As a result, memory ceased to be simply a slow, passive archive; it became an active continuation of the computing process.
More than that, due to harsh energy limits, data center architects today are massively abandoning energy-consuming HDDs in favor of SanDisk's solid-state drives, which provide the necessary density and speed.
Financial Retro-analysis: The Death of Historical Cyclicality
This technological shift coincided with the most important corporate transformation of the company itself, which created the effect of a perfect storm for the shares. For a long time, SanDisk remained a hostage of the parent company, Western Digital (WDC), where debts and the stagnating segment of HDDs diluted the overall margins.
The finalized spin-off at the beginning of 2025 allowed the market to begin valuing SanDisk as an independent pure-play player in the flash memory market at exactly the moment when this product became critically scarce.
The historical cyclicality died. Earlier, the memory market was subject to three- to five-year cycles of renewal of consumer electronics, depending on when people would buy new PCs or smartphones. Today, due to the development of AI, the generation of data became a continuous routine process of interaction of machines with machines and people. AI does not sleep, does not go on weekends, and does not wait for Christmas sales to generate video or synthetic datasets.
The array of information generated today constantly grows. Under these conditions, classic cyclicality dies out, giving way to a constant, structural base demand.
This shift gave SanDisk unprecedented pricing power. In 2026, the industry faced the heavy consequences of underinvestment in past years: warehouses are empty. The sharp deficit of ultra-fast corporate SSDs allowed SanDisk to dictate prices to hyperscalers.
In the semiconductor industry, the base production cost of a chip—silicon, lithography, amortization of equipment—is a relatively fixed cost. In the period of overproduction, selling prices fell almost to the level of production costs, leaving manufacturers a minimal margin. However, right now, when contract prices flew up against the backdrop of a harsh deficit, all this price markup directly converts into net profit.
Expenses on production remained the same, and revenue from every chip grew severalfold. This colossal effect of operating leverage instantly turned flash memory chips from a low-margin mass commodity into a super-profitable product, driving the explosion of net profit to $800 million in the fourth quarter of 2025. This fully justifies the current capitalization of $135 billion because the business generates real cash right now and does not live on promises.
Financial Modeling: Deconstruction of the Forward P/E
The main fear of the crowd is that a 20-fold stock growth automatically means an inflated bubble with sky-high multiples. To dispel this optical illusion, it is enough to look at Wall Street consensus valuations and extrapolate the data from the last financial quarter.
Using Seeking Alpha's data, we see a striking picture: despite a capitalization of almost $136 billion and a share price above $910, the forward P/E multiple is only 21x. For the AI-infrastructure sector, where companies often trade at 50x or 80x multiples, an indicator of 21x looks like classic value investing.
www.barchart.com
How did such math become possible? The answer hides in the radical change of the company's financial basis, visually confirmed by the reporting for the quarter ended in January 2026. The company generated $3.02 billion in total revenue. Much more important is that gross profit constituted $1.54 billion. This means the gross margin surpassed the incredible mark of 50%.
From this revenue, the company squeezed out $803 million in net profit for a single quarter. If we take this baseline of $800 million and simply multiply it by four quarters, we get an annualized profit of $3.2 billion. However, market consensus forecasts factor in further growth of contract prices and delivery volumes. The expected forward earnings per share is currently valued by consensus estimates at $43.64. Given the current structure of the company's share capital, achieving such an EPS means that SanDisk is projected to generate about $6.5 billion in net profit for the upcoming year.
The valuation is well supported by the numbers. If we divide the current capitalization of $136 billion by the forecast net profit of $6.5 billion, we get that very P/E multiple in the area of 21x. The valuation is fully adequate because it is based on the already occurred explosive growth in profitability. The business model rebuilt itself, memory chips became a high-margin product, and the market simply brought capitalization into accordance with the new, multiplied cash flows.
The Math of Valuation: Comparison with Micron
One question still remains: why does the main competitor, Micron (MU), cost $500 billion, while SanDisk costs $135 billion? The temptation arises for many to say that SanDisk must catch up to the leader, but such a direct equalization would be fundamentally erroneous.
The reason hides in the simple difference in revenue volumes and target market coverage. Micron plays in two fields simultaneously today: the company occupies a huge share in the RAM market and is also present in the storage market. SanDisk is a pure-play entity that specializes exclusively in flash storage.
Let's look at the hard numbers. For the last quarter, SanDisk generated $3 billion in revenue. If we extrapolate this pace taking growth into account, we get an annualized revenue in the area of $12-13 billion. If we glance at Micron, its revenue for the last 12 months already exceeded $37 billion, and according to consensus forecasts amid the AI boom, it is rushing toward the $50 billion mark.
The difference in revenue generation is almost fourfold. This is a natural ceiling. Being a pure-play company in one market, it physically cannot generate the same cash flows as a diversified giant that also sells super-margin HBM memory for graphic processors. This is precisely why the gap in capitalization is absolutely logical.
A valuation of $135 billion is not a sign of undervaluation in relation to Micron, but a highly precise reflection of SanDisk's own revenue. Within the frames of its target market, the company is valued fairly and strongly.
Strategy of "Smart Expansion"
Valuing the prospects of SanDisk, I look not only at the price chart but also at how much concrete is being poured into the foundations of new factories. The memory market is a capital-intensive oligopoly, and the strategy of capital expenses here decides everything.
Competitors right now are obsessed with massive scaling. Samsung (SSNLF) is aggressively expanding the mega-cluster in Pyeongtaek, attempting to crush the market by mass. SK Hynix is pouring $15 billion into a new cluster in Yongin, and Micron is plowing tens of billions into new factories in Idaho and Singapore, leaning on U.S. state subsidies for the former project.
Against this backdrop, the alliance of SanDisk and Japanese Kioxia (KXHCF) demonstrates superb capital efficiency. Instead of breaking ground on new facilities during a period of peak interest rates, they are launching mass production of the 10th generation of NAND at the already completed K2 shell in Kitakami. SanDisk simply installs new equipment into existing clean rooms, minimizing downtime and capital expenses. In 2026, this strategy gives them a colossal advantage: they capitalize on the deficit and high prices right now, while competitors burn cash on construction sites.
Fundamental Risks
Although the current situation is extremely favorable, investors must preserve objectivity. It is exactly from the aggressive race of competitors, mentioned above, that the sole fundamental risk for our investment thesis stems. It is directly tied to the 2027-2028 period.
When the massive new factories of Micron and SK Hynix are completely built and become operational, a temporary supply overhang can arise in the market. If new production capacities outpace structural demand from AI, pricing power could weaken in the short term, and sector margins could contract.
This risk is significantly mitigated by two factors. First, in the current year of 2026, pricing power is completely on SanDisk's side, and they pragmatically convert it into colossal profit. By the time competitors' new capacities come online, SanDisk will have built a gigantic financial cushion, which it can direct toward its own expansion.
Second, the global data generation market does not stand still. The array of information grows so fast that by 2028, the world economy will require incommensurably more memory. What today seems like future "overproduction" can tomorrow be fully absorbed by the new base demand from data centers, excluding a return to the deep cyclical crises of the past.
Outcome: Navigating the Volatility
The AI hardware market has passed the point of no return. The non-stop data generation of neural networks broke the old cyclicality of the memory market, turning Enterprise SSDs from an expensive option into a base infrastructure element. SanDisk's business has transitioned into a qualitatively new state, and a return to losses and quotes at $45 looks highly improbable from a fundamental standpoint.
However, a stock that has grown 20-fold possesses colossal historical amplitude. Euphoria and fear will regularly swing the quotes, creating high volatility. Given this extreme amplitude, rather than relying on strict directives, it might be more prudent to evaluate the stock within a hypothetical valuation corridor. Assuming the company is currently priced relatively fairly around its current levels, we can outline potential zones of interest.
If broader market corrections or localized profit-taking press the price closer to the $700 mark, the stock could be considered relatively cheap, potentially offering a more attractive margin of safety. Conversely, if a wave of market euphoria pushes the shares toward the $1200 level, SanDisk's capitalization will approach $180 billion, breaking its proportional parity with Micron. At those marks, the stock would likely be viewed as fundamentally expensive, presenting a scenario where one might consider it a less favorable entry point and a potential area to re-evaluate exposure.
Comments