Sandisk Posts Record $11 Billion SSD Quarter, Forecasts $12.4 Billion Next

Sandisk booked $11 billion in SSD revenue in its first fiscal 2027 quarter, up 378 percent year over year and a third consecutive record after $9 billion and $6 billion in the prior two quarters. The company guided the next quarter to $12.4 billion plus or minus $200 million, a 310 percent annual rise at the midpoint.

GAAP net income reached $8.1 billion, up 7,133 percent. Operating cash flow was $6.4 billion against $488 million a year ago, and free cash flow hit $5.7 billion against $448 million. Sandisk attributes the durability of this demand to long-term customer supply and purchasing agreements running to 2028 and sometimes beyond.

CEO and Chairman David Goeckeler said: "We have started fiscal year 2027 with another quarter of strong performance, with robust and growing demand for our products driving outpaced growth in our data center business. We are executing on the strategy that we outlined at Investor Day, delivering growing and durable free cash flow through our technology leadership, deep customer partnerships, and disciplined financial management."

Which segments drove the numbers?

Sandisk reports three market sectors, and they diverged sharply:

• Datacenter: $4.8 billion, up 1,682 percent year over year
• Edge: $5.8 billion, up 320 percent
• Consumer: $397 million, down 39 percent, which the company attributed to a market adjusting to higher pricing

Datacenter is now the fastest-growing sector and could overtake edge within a quarter or two, particularly with PC sales held back by high memory prices. The company said its expectations are "reinforced by a structural, AI-driven increase in NAND demand and NBM customer signals extending beyond fiscal 2030." NBMs are long-term supply and purchasing agreements with large customers.

Sandisk stated: "Customer signals extend beyond fiscal year 2030 through our NBM partnerships; recent engagements with leading AI companies indicate further demand upside." It added that the "NBM pipeline exceeds our supply through the end of the decade."

That is a strong claim. It is also a claim about agreements and pipeline, not booked revenue, and the company did not disclose how much of the pipeline converts to binding volume.

What is the AI inferencing opportunity?

Sandisk expects AI inferencing to sustain demand beyond model training. The company said: "We expect KV cache to be ~one-third of the AI datacenter flash market by decade-end, with the AI datacenter opportunity approaching the size of today's NAND industry."

It assigned roles to its two flash types: "TLC serves latency-sensitive KV cache and active datasets; QLC adds density for data lakes, vector databases that power AI search, and retrieval repositories." That forecast sits at the far end of the decade. It is a roadmap, not a result, and the market size comparison rests on Sandisk's own estimate.

On edge, Sandisk described a market spanning smartphones, PCs, automotive, tablets and physical AI applications. It said deepening customer engagements "reinforce our confidence that Automotive can become one of our largest markets within Edge over time."

The consumer segment is now 3.7 percent of total revenue. Sandisk calls it "a differentiated strategic asset" that "provides a durable foundation for our business" and "early insight into shifts in end-user demand." The reader may reasonably ask whether that is strategy or rationalization of a shrinking business, since those NAND bits could presumably earn more sold to datacenter and edge customers. The confetti era of USB sticks at supplier events appears over, and prices are high enough that it may not return.

How does the Elliott split look now?

The numbers vindicate the corporate activists, though by accident. Elliott Management pushed Western Digital to split into WD for disk and Sandisk for flash, arguing the separation would enhance shareholder value. The then-unexpected AI boom delivered the rest. WD recently reported $4.4 billion in quarterly revenue; Sandisk is growing much faster. AI workloads need SSDs more than HDDs.

Against fab joint-venture partner Kioxia, which holds a strong datacenter presence, Sandisk's revenue is falling behind, weighed down by its historically dominant but slower-growing edge business and the declining consumer segment. Goeckeler is rebalancing toward datacenter to close that gap.

The share price has risen enough that Sandisk announced an eight-for-one forward stock split, effective November 19, 2026, to improve accessibility including through its employee stock purchase plan. Kioxia is splitting its stock for the same reasons.

Will high-bandwidth flash find a customer?

Sandisk is betting with SK Hynix that stacked flash can follow the HBM playbook into AI accelerators. High-bandwidth flash stacks layers of 3D-NAND dies with through-silicon vias onto a base logic die, then connects at high bandwidth to a GPU, aiming to beat NVMe SSD limits on bandwidth and latency. So far, no GPU supplier has committed to the technology.

The company reported a concrete step: "HBF reached a milestone in FQ1 with tape-out of our first HBF memory die; samples of our first HBF-based products remain on track for next calendar year." Tape-out is progress. A committed GPU partner would be proof.

For calendar 2027, Sandisk expects datacenter to represent over 55 percent of its market, with datacenter bit demand growing about 40 percent year over year on AI-driven capital spending, after more than doubling in 2026. Those figures are company forecasts. The NBM pipeline stretching past 2030 suggests the pricing power lasts; the absence of an HBF customer suggests the next technology bet is still unproven.