TradingKey – While investors are focused on Micron Technology, which is leveraging machine learning to spark a historic memory boom, big changes are also occurring in other areas of the storage market. Once considered just another cyclical hard drive manufacturer, Western Digital is steadily establishing itself as a key source of infrastructure for the machine learning era.
This shift is occurring as hyperscale cloud service providers rush to increase machine learning capacity, significantly increasing the demand for high-bandwidth memory and NAND flash, as well as for long-term, cost-effective data storage. As such, the recent surge in demand has changed pricing trends across the storage industry, driving up the value of companies whose value is tied to the data infrastructure needed to support machine learning.
AI infrastructure spending growth outpaces memory chips
While the recent AI market has been primarily driven by the shortage of GPUs and the impressive growth in sales of Micron’s high-bandwidth memory, there is a secondary and less-discussed constraint that comes from the creation of persistent data storage brought about by building AI around training, inference, agent AI, and enterprise-scale applications that generate vast amounts of data that must be archived and retrieved at a reasonable cost. As a result, demand for NAND flash and high-capacity hard drives (HDDs) continues to grow.
After spinning off its flash business to SanDisk in 2025, Western Digital became an independent supplier solely focused on the HDD segment, with its product offerings primarily aimed at cloud and enterprise customers. Hyperscalers continue to strengthen their position as they make large-scale storage deployments related to AI infrastructure a key priority.
Almost all of Western Digital’s 2026 HDD capacity is already under contract, with long-term contracts with customers extending through 2028 and 2029, management indicated.
Western Digital’s profits are influencing investment decisions in new ways
Based on Western Digital’s latest report on quarterly earnings, it’s clear that the company is starting to experience significant positive changes in its revenue as the AI storage cycle continues to gain momentum.
In its third fiscal quarter ending March 2026 (Q3 FY26), the company reported revenue of $3.34 billion, up 45% year over year. Also noteworthy in the quarterly results was adjusted gross margin exceeding 50% (50.5%) for the first time in company history. Last year it was 40%. Adjusted EPS nearly doubled to $2.72.
Pricing played a big role in these improvements. Average selling price per exabyte (“ASP”) increased 9% year-over-year, and exabytes sold (exabytes shipped) increased 34%. This significant increase reflects strong enterprise demand for high-capacity storage systems.
The company has also significantly ramped up its introduction of UltraSMR technology, which can dramatically increase storage density. Management said two of its largest customers utilize UltraSMR technology to meet nearly all of their exabyte-based storage requirements.
Additionally, cash generation has also improved significantly. The company generated approximately $1 billion of free cash flow during the quarter, resulting in significant and continued positive shareholder returns and continued balance sheet improvement. The company increased its quarterly dividend by 20% after increasing the dividend by 25% over the past six months.
Storage solutions expand due to demand from AI growth
In addition to the imbalance in supply of DRAM and high-bandwidth memory that has driven up Micron’s price, eliminating opportunities for price increases, demand from AI is starting to trickle into other adjacent areas.
The latest evidence of this is Micron’s high volume sales of advanced DRAM and high-bandwidth memory. These two technologies are essential to operating AI acceleration and inference workloads. Therefore, with sales volumes and associated price increases during this period of supply constraints, it is not surprising that Micron has previously stated that its entire supply of advanced memory for 2026 is completely sold out.
Similarly, the same demand pressures that are driving DRAM sales are now impacting the NAND flash and HDD markets as well.
Western Digital’s NAND flash business was recently acquired by Sandisk (previously it went bankrupt in a company split). The company’s recent quarterly revenue growth has exceeded 200% and gross margins have exceeded 78%. These results reflect the rapid growth in ASP across the flash storage market.
Unlike previous memory cycles, most storage providers have long-term, multi-year supply contracts with hyperscale customers to reduce revenue volatility. This includes providing volume commitments and establishing pricing frameworks to reduce cash flow volatility.
The overall premise of this situation is: Existing and future AI systems will significantly require large amounts of computing power. At the same time, AI systems require large amounts of cost-effective and persistent storage infrastructure. As a result, storage infrastructure demand cycles are likely to be much longer than previous cloud infrastructure investment and development booms.
Despite strong fundamentals, risks remain high
Western Digital and other AI storage stocks remain highly risky despite improving fundamentals.
Western Digital’s growth relies on a small number of hyperscale cloud customers for much of its revenue. A slowdown in hyperscale cloud spending on AI infrastructure, changes in sourcing behavior, and reductions in enterprise capital spending will put significant downward pressure on product pricing and associated margins.
Execution risk continues to be another major issue facing Western Digital. The company continues to work on the commercialization phase of its next-generation HAMR drives, and management says it continues to optimize yields and improve reliability ahead of production increases that are not expected to occur until at least 2027.
Valuations also remain a major concern. Over the past 12 months, Western Digital’s stock price has risen more than 200% as investors enthusiastically embraced the AI storage story. There are growing concerns that the market is already pricing in demand for AI infrastructure to accelerate for years, as was the case with Micron and SanDisk.
There appears to be a growing debate among online investors, both optimistic and pessimistic, surrounding the longevity of the AI infrastructure cycle, which has a history of large boom-bust price swings.
What investors should pay attention to next
The emerging AI storage deal includes companies that only make memory semiconductors (such as Micron), as well as a variety of other storage infrastructure companies, such as Western Digital and Sandisk.
The key consideration investors are looking at is whether current market conditions represent only a temporary supply squeeze, or are we actually at the beginning of a much larger supply cycle for artificial intelligence that will require the use of memory-based storage for inference processing, long-term data retention, and the deployment of enterprise-grade AI systems within the enterprise?
If the number of AI workloads deployed (besides training) increases while buyer interest from hyperscalers remains strong, the outlook for how long demand for new memory-based storage will sustain could change. This will further drive Western Digital’s focus on high-margin enterprise HDDs, potentially making this transition longer than all previous cycles.
On the other hand, if spending on AI continues to slow or storage prices normalize much faster than expected, this could mean a return to normal cyclicality for this sector, and given how volatile this sector’s revenues have been historically, it could recover very quickly. Therefore, given the current environment, Western Digital is an interesting investment in the AI infrastructure space and could offer long-term growth opportunities. However, this also involves significant cyclical and execution risks.
Disclaimer: The content of this article represents only the personal opinions of the author and does not reflect the official position of Tradingkey. It should not be considered investment advice. This article is for reference purposes only and readers should not make investment decisions based solely on its contents. Tradingkey is not responsible for any trading results caused by reliance on this article. Additionally, Tradingkey cannot guarantee the accuracy of article content. Before making any investment decisions, we recommend that you consult an independent financial advisor to fully understand the risks involved.
