Sandisk forecasts profit surge as AI boom drives flash storage demand

Summary

Sandisk projected a sharp jump in revenue and profit on Thursday, signaling that the artificial intelligence boom is now spilling decisively into the data storage market. The company also extended a long-term supply partnership, positioning itself to meet rising demand from AI-focused data centers.

Sandisk forecast fiscal third-quarter revenue of about $4.6 billion at the midpoint and adjusted earnings of roughly $14 per share — far above Wall Street expectations of $2.77 billion in revenue and $4.37 per share in profit, according to LSEG data.

The upbeat outlook underscores a key shift in the AI infrastructure race. While investor focus has largely centered on advanced memory chips close to processors, AI systems also rely heavily on flash storage to hold massive datasets used in training and inference.

AI inference drives storage spending

Sandisk CEO David Goeckeler said demand is being fueled by the rapid build-out of data centers designed for AI inference — the stage where trained models respond to user queries in real time.

That process requires constant access to enormous volumes of stored data, boosting demand for high-capacity solid-state drives built on NAND flash memory.

As hyperscalers race to deploy AI infrastructure, storage supply has tightened.

“Customers prefer supply over price,” Goeckeler said, highlighting the urgency among AI operators to avoid deployment delays.

Strong quarterly results reinforce momentum

For its fiscal second quarter, Sandisk reported $3.3 billion in revenue and adjusted earnings of $6.20 per share, comfortably beating analyst forecasts of $2.64 billion and $3.33 per share.

The results add to growing evidence that AI capital spending is broadening beyond processors into storage, power, cooling, and data center infrastructure — increasingly the real bottlenecks of AI growth.

Long-term supply secured with Kioxia

To support future demand, Sandisk extended its NAND flash supply agreement with Japanese partner Kioxia Corp through 2034, pushing out the previous 2029 expiration.

The two companies operate a long-running joint venture manufacturing flash chips in Japan and continue investing heavily in capacity expansion.

The deal gives Sandisk long-term visibility into supply as AI-driven demand tightens global memory markets.

Why This Matters

• AI is lifting the entire hardware stack, not just GPUs — storage is becoming a critical growth engine
• Flash memory demand is exploding as inference workloads scale globally
• Supply security is now strategic, not just cost-driven
• AI infrastructure bottlenecks are shifting toward storage, power, and cooling

This signals a new phase of the AI boom — where every layer of data center hardware becomes a profit driver.

FAQs

Q1. Why is Sandisk benefiting from the AI boom?

AI systems require massive fast storage to hold training data and support real-time inference workloads.

Q2. What is AI inference and why does it drive storage demand?

Inference is when trained models respond to users. It constantly pulls huge datasets from storage, boosting flash usage.

Q3. How is flash memory different from DRAM?

DRAM handles active processing near chips, while flash stores large volumes of data permanently. AI needs both.

Q4. What is Sandisk’s deal with Kioxia?

A joint manufacturing partnership producing NAND flash chips in Japan, now extended through 2034.

Q5. Why are customers prioritizing supply over price?

AI deployment speed is critical — delays cost far more than higher component prices.

Q6. What does this signal for tech markets?

AI investment is expanding across the full infrastructure chain, not just semiconductors.