After the stock price of Knight Hero surged by 1200%, the veteran in charge of storage was ordered to drive the super cycle! Striving to elevate eSSD to the position of AI computational power era.

The global leader in NAND flash memory, Kioxia Holdings Corp., based in Japan, has promoted veteran in the storage field, Hiroo Ota, to the position of CEO and President. This move aims to accelerate the expansion of its market share in the thriving storage chip market, as NAND flash memory transitions from a cyclical product to a core component in the AI computing infrastructure sector. Amidst the immense growth in storage demand brought about by the current AI computing trend, Kioxia, led by Ota, has seen its enterprise data center SSD (eSSD) become one of the hottest investment themes in the global AI computing market. It is reported that 63-year-old Ota will succeed 70-year-old Nobuo Hayasaka, who will assume the role of Senior Executive Advisor at Kioxia. The core supplier of storage chips for Apple Inc.'s iPhone stated in a statement on Thursday that the leadership change will officially take effect after receiving shareholder approval at the annual general meeting in June. In 2025, storage chip stocks and high-end product stocks were undoubtedly one of the hottest investment themes in the global stock market, and this trend has continued into 2026. For example, the cumulative increase in shares of leading data center enterprise SSD component supplier, SanDisk (SNDK.US), since 2026 has exceeded 122%, with a 580% increase in 2025. Despite the soaring trends of the super bull market in 2025 and the continuing bull market in 2026, global investors are not overly concerned about the sudden rise in valuations of these storage technology companies because they believe the unprecedented data center construction surge for artificial intelligence is changing the "strong cyclical nature" of the storage chip industry. Whether it's Alphabet Inc.'s massive TPU AI computing cluster, or the vast NVIDIA Corporation AI GPU computing cluster, the need for integrating HBM memory systems with AI chips, as well as the large-scale purchase of server-level DDR5 memory and high-performance enterprise SSD/HDD for new or expanded AI data centers by tech giants, has positioned companies like Samsung Electronics, SK Hynix, Micron Technology, Inc., and Kioxia, as leaders in various core storage categories: HBM, server DRAM (including DDR5/LPDDR5X), and high-end data center level SSD/HDD. These storage giants are the most direct beneficiaries of the "AI memory + storage stack" and are reaping the benefits of the AI infrastructure boom. According to Wall Street financial giant J.P. Morgan and other institutions, the biggest winners of this unprecedented "storage supercycle" may be Samsung, SK Hynix, Kioxia, and SanDisk in the eSSD sector. J.P. Morgan stated that the AI computing trend is transforming NAND flash memory from a "cyclical commodity" into a high-growth AI infrastructure asset. As AI workloads shift from training to inference and HDD supply bottlenecks in nearline storage, the focus on enterprise storage hot tiers like NVMe eSSD is experiencing unprecedented structural growth. Since its IPO, Kioxia's stock price has surged by 1200% Since its initial public offering (IPO) on the Japanese stock market at the end of 2024, Kioxia's stock price has violently risen by over 1200%. Following the announcement of this appointment, Kioxia's stock price in the Japanese stock market has risen by about 2%, highlighting the market's extreme trust in this veteran of the storage industry. Spun off from Toshiba Corp., Kioxia, as the inventor of NAND flash memory technology, has become one of the core beneficiaries of the AI infrastructure construction boom. This boom is driving the demand and prices of almost all storage chips in the DRAM and NAND storage sectors. Its high-performance storage solutions, including enterprise-level high-performance HDDs and eSSD, are becoming increasingly important for the accelerated construction of AI data centers globally, elevating its status from a commodity to a must-have for any tech company looking to succeed in the AI race. The incoming CEO, Hiroo Ota, has spent most of his career in the storage semiconductor industry; he joined Toshiba in 1985. He previously held core engineering positions at Toshiba, the former storage superpower, before the company was completely rebranded as Kioxia. "Being a CEO of a NAND flash company is an extremely difficult job," said Omdia analyst Akira Nanao. Given that the market is known for its unpredictable cycles between prosperity and recession, he noted that investment timing must be carefully managed over the long term. However, in the current supercycle, this "cyclical" label seems to no longer apply. "Leaders need to be at the forefront of the industry, frequently travel and deeply engage in sales, and be sensitive to subtle changes in customer perceptions." The company stated in a release that the board decided that now is an appropriate time for the leadership transition, as Kioxia celebrates its one-year anniversary since its IPO and aims to accelerate growth amid the surge in demand for eSSD storage chips driven by the AI computing trend. eSSD, likely to be the biggest winner of this super storage cycle The reason why Kioxia has become one of the core beneficiaries of this "storage supercycle" is fundamentally due to the AI data centers generating exabyte-scale AI training/inference workloads, pushing the "generation/retention/searchability of data" to high levels. Training and inference are continually raising the demand for high-capacity, low TCO (per TB cost + power consumption), and higher throughput/lower latency storage, all accurately controlled by Kioxia enterprise NVMe SSDs. Kioxia's QLC ultra-large capacity enterprise drives (LC9) + PCIe 5.0 enterprise/data center NVMe (CM/CD series) + low-latency SCM (XL-FLASH) form these three major enterprise storage product lines, specifically designed for AI inference. J.P. Morgan expects a significant 40% year-on-year increase in the mixed average selling prices of NAND industry in 2026, with this rise expected to be sustained, hovering near historical highs by 2027. There is a common misconception in the market regarding storage theme investments, where AI computing trends are seen as favorable to DRAM (especially HBM) in the long term, with NAND playing a supporting role. However, J.P. Morgan has outlined in a research report that in the era of AI inference, eSSD is just as crucial as HBM. Looking ahead, J.P. Morgan predicts that by 2026, the demand for storage capacity per AI server will exceed 70TB, at least three times that of general servers (around 20TB); by 2027, eSSD is expected to account for 48% of global NAND bit demand, surpassing the two long-standing pillars of NAND support - smartphones (30%) and PCs (22%). The recent major catalyst for eSSD undoubtedly lies in NVIDIA Corporation's new "context inference/long context inference" BlueField platform (ICMS: Inference Context Memory Storage), which can be seen as a significant long-term growth driver for "hot/warm-tired enterprise SSD/NVMe" and the cold/warm data ocean (Nearline HDD/object storage). NVIDIA Corporation specifically positions ICMS in a technical blog as a new G3.5 layer: a pod-level "Ethernet-attached flash tier" designed to store latency-sensitive, reusable inference contexts (KV cache) and emphasize its provision of "PB-level shared capacity" per GPU pod, supporting KV cache reuse and prefetch with higher bandwidth and efficiency to reduce GPU stalls and increase tokens/s. In terms of storage "benefit logic," this platform adds a high-bandwidth flash layer between local SSDs and shared storage for "context (KV cache)," leading to a greater need for enterprise SSD/NVMe (including SSD pools in NVMe-oF form).