explainx.ai0k
TrendingAI News TodayPathwaysSkills
Pricing
explainx.ai

Upskill in AI — 16 free pathways, live workshops & bootcamps, and 50+ courses from practitioners. Plus the skills, tools, and MCP servers to practice on.

follow us

follow on google

Add explainx.ai as a preferred source

corporate training

support@explainx.ai

get started

Find your pathTake Free Evaluation

community

Join the community

learn

mind: share how you thinkpathways — start freeworkshopsbootcampscoursescompare Explainxcertificationsmock testsexplainx universitycorporate traininglearn skills & mcp

discover

skillsmcp serversexplainx mcptoolsmdx readeragentsllmsdesignsdictionarypeopleagi trackerfelony benchranks

company

aboutvisionmissionteaminstructorsteach on explainxpartnershipscommunityhackathonscareers

content

daily AI newsstate of AI — live resultsblogreleasespromptsgeneratorsresource libraryfor LLMsexplainx.ai kids

solutions

all solutionsdeveloper upskillingmarketing upskillingproduct manager upskillingleadership upskilling

newsletter · weekly

Get AI news, tools, and insights in your inbox.

supportcontactprivacytermsdata rightshow we create contentsubmission guidelines

© 2026 AISOLO Technologies Pvt Ltd

explainx.ai

On this page

  • TL;DR: the confirmed numbers
  • What the press release confirms
  • Why an 85% gross margin matters
  • Reading the three segments
  • The balance sheet and cash flow clues
  • Why flash matters for AI work
  • The cycle risk
  • How it fits the memory squeeze
  • What this means for what you build or pay
  • What is not confirmed
  • How to read the numbers yourself
  • Related reading
← Back to blog

explainx / blog

Sandisk Q1 FY2027: Data Center Revenue Hits $4.8B as AI Storage Demand Reshapes the Flash Market

Sandisk, AI Hardware, Memory, Data Centers, Industry

Part of AI Chips and Infrastructure

Sandisk reported $11.0B revenue and $4.8B data center sales, up 61% in a quarter, at 85% gross margin. What is confirmed and what it means for AI storage costs.

Oct 30, 2026·8 min read·Yash Thakker
add explainx.ai
go deep
Sandisk Q1 FY2027: Data Center Revenue Hits $4.8B as AI Storage Demand Reshapes the Flash Market

Sandisk's fiscal first quarter of 2027 is a clean snapshot of what AI demand is doing to the storage market. The company reported $11.0 billion of revenue for the quarter ended October 2, 2026, with data center sales of $4.802 billion, up 61% from the prior quarter, according to its October 29 press release. Gross margin was 85.1% on a non-GAAP basis.

This is a finance story, so we keep it to what builders can use: what is confirmed in the filing, how it fits the memory and storage squeeze we have been tracking, and what it implies for the cost of running AI.

TL;DR: the confirmed numbers

table · 4 cols
ItemQ1 FY2027Prior quarterYear ago
Revenue$11,031M$8,965M (up 23%)$2,308M (up 378%)
Data center$4,802M$2,977M (up 61%)$269M
Edge$5,832M$5,432M (up 7%)$1,387M
Consumer$397M$556M (down 29%)$652M
Non-GAAP gross margin85.1%84.6%29.9%
GAAP net income$8,101M$6,903M$112M
Non-GAAP EPS$49.51$39.25$1.22

Source: Sandisk's press release. Q2 guidance: revenue $12.2B to $12.6B, non-GAAP EPS $55.00 to $57.00 pre-split.

What the press release confirms

  • Revenue of $11.0 billion, up 23% sequentially, with GAAP net income of $8.1 billion.
  • Data center at $4.8 billion, now the fastest-growing segment, though edge (devices and PCs) is still larger at $5.8 billion.
  • A $5.0 billion share buyback in the quarter, about 3.2 million shares, and an eight-for-one stock split effective November 19, 2026.
  • Adjusted free cash flow of $5.696 billion, a 51.6% margin.
  • CEO David Goeckeler said: "robust and growing demand for our products driving outpaced growth in our data center business."

Yahoo Finance's earnings reaction piece and earnings call highlights cover the same quarter. The release does not say how much of data center revenue is specifically AI. Sandisk's description of itself lists "enterprise SSDs powering AI infrastructure," and third-party summaries tie the growth to AI storage and hyperscaler customers, but the AI share is an inference, not a disclosed number.

Why an 85% gross margin matters

A year earlier the same company earned a 29.9% non-GAAP gross margin on $2.3 billion of revenue. Revenue is now nearly five times higher and gross margin is 55 points higher. Flash is a commodity business that normally swings between glut and shortage. Margins this high mean pricing power: buyers are paying a lot more per bit and competing for supply.

The release's risk language mentions "pricing trends and fluctuations in average selling prices" and "volatility in demand," and it notes reliance on long-term agreements and its partner Kioxia. It does not break out price versus volume. Both clearly rose, but we cannot say by how much.

Reading the three segments

Data center ($4.802B) is the story. It grew from $2.977B a quarter earlier and from $269M a year earlier. Within a year it went from a small slice of revenue to 44% of the total. Enterprise SSDs sit behind this line: the drives in servers that feed GPUs with training data, hold checkpoints, serve embeddings and retrieval indexes, and store the model artifacts that inference clusters load.

Edge ($5.832B) covers flash in devices such as phones and PCs. It still produces more revenue than data center, and it grew 7% sequentially and 320% year over year, so prices and volumes are higher here too. AI on devices is part of why: bigger local models and bigger caches need more storage per device, though the release does not attribute the growth to that.

Consumer ($397M) is the weak spot, down 29% sequentially and 39% year over year. Retail cards and drives are the lowest-priority customers when supply is tight, which fits the pattern of capacity being steered toward large contract buyers.

The balance sheet and cash flow clues

The cash flow statement has details that the headline does not.

  • Accounts receivable rose from $4.708B to $6.585B in a quarter, so about $1.9 billion of the revenue growth is not yet cash. That is normal when sales jump, but it is worth watching.
  • Inventories rose modestly, from $2.698B to $3.016B. Inventory is not piling up, which fits a supply-constrained market rather than an oversupplied one.
  • Prepayments and deposits from new business models reduced adjusted free cash flow by $465 million in the quarter, after $1.938 billion in the prior one. The company's risk factors refer to "long-term agreements, or New Business Models," which suggests customers are committing money ahead of delivery. The release does not detail the terms.
  • Net income included a $513 million gain on equity securities, which Sandisk excludes from non-GAAP results. That is why GAAP EPS ($52.60) is above non-GAAP ($49.51).

None of this is advice on the stock; it is a way to see that demand is backed by commitments, not only spot purchases, and that part of the profit is not operating.

Why flash matters for AI work

GPU and HBM shortages dominate the AI hardware conversation, but real systems also depend on storage in several places:

  • Training data pipelines stream large datasets from SSD to accelerators; slow storage leaves expensive GPUs idle.
  • Checkpointing writes large model states often, so write endurance and speed matter.
  • Retrieval and vector databases keep indexes on fast flash for low-latency search.
  • Inference at scale loads model weights quickly and increasingly offloads parts of the key-value cache to storage to save GPU memory.
  • Agent workloads generate logs, traces and artifacts that need somewhere to live.

As models and context windows grow, more of the stack spills from GPU memory to host memory to flash. That raises the amount of storage per accelerator, which is one reason demand rose faster than GPU counts alone would suggest. Again, that is our reading of the architecture trend, not a statement in Sandisk's release.

The cycle risk

Memory and flash have a long history of boom and bust. High prices bring new capacity, buyers stockpile and then pause, and prices fall sharply. Nothing in this release says the cycle has turned, and guidance for next quarter is higher. But an 85% gross margin on a commodity is the sort of number that invites supply responses, from competitors and from Chinese producers, as the Samsung and SK Hynix reporting on China NAND plants hints. Builders should plan for high prices now and a possible swing later, rather than assume either lasts.

How it fits the memory squeeze

Sandisk is one data point in a pattern we have covered across the supply chain this month:

  • Kioxia's quarter, where flash was described as sold out into 2027.
  • Samsung's record Q3 profit tied to HBM4 contracts.
  • SK Hynix's record quarter and its view of shortages stretching years.
  • CXMT's revenue jump as AI memory prices lifted China's DRAM maker.
  • Michael Dell on the memory shortage lasting years.

The common thread: HBM and DRAM got the headlines first, and flash storage for training data, checkpoints, vector databases and inference caches is following the same path.

Weekly digest3.6k readers

Catch up on AI

Curated AI updates on agents, skills, and MCP — delivered to your inbox. Unsubscribe anytime.

What this means for what you build or pay

Storage is a smaller line than GPUs in an AI budget, but it is no longer free.

  1. Cloud storage and SSD prices are unlikely to fall soon. A supplier at 85% gross margin has no reason to cut. That is our inference from the numbers, not a forecast from Sandisk.
  2. Local AI hardware costs more. Fast NVMe for model weights is part of any local rig. The same squeeze showed up in consumer devices, for example the DGX Spark memory cut.
  3. Design for less data movement. Cache wisely, compress checkpoints, delete stale artifacts, and tier cold data to cheaper storage.
  4. Budget with a buffer. If you are planning a data pipeline or a self-hosted stack for 2027, assume flash costs flat to higher, not lower.
  5. Watch consumer weakness. Consumer revenue fell 29% sequentially. Supply is being steered to data centers, which can leave phones and PCs with thinner or pricier storage.

What is not confirmed

  • Which customers are driving data center revenue. Third-party summaries mention U.S. hyperscalers, but the press release as we read it does not name them.
  • The AI share of data center sales.
  • How much of the growth is pricing versus volume.
  • How long it lasts. Guidance for the next quarter is higher, but it is guidance.
  • Any effect on the prices you pay. That depends on your cloud or retailer.

Figures such as year-over-year percentage growth in data center vary by outlet because the year-ago base was tiny ($269 million); Sandisk itself marks that comparison as not meaningful.

How to read the numbers yourself

Open the press release, find the end-market table, and compare data center to the total. Divide to see the share: $4,802M of $11,031M is about 44%, up from roughly 33% in the prior quarter ($2,977M of $8,965M). The earnings presentation on Sandisk's investor site has more detail, and the 10-Q will follow with final numbers that may differ slightly, as the release itself warns.

Related reading

  • Kioxia Q2: sold out into 2027
  • Samsung Q3 2026 record profit and HBM4
  • SK Hynix Q3 2026 and the memory shortage
  • CXMT Q3 2026 revenue
  • Michael Dell on the memory shortage
  • AI chip supply chain: TSMC, Samsung, GlobalFoundries

Figures are from Sandisk's October 29, 2026 press release and are subject to change when the 10-Q is filed.

Spotted something out of date? Let us know.
Yash Thakker

Written by

Yash Thakker

Yash is an AI expert with over 300K learners. Join his workshops →

View Yash Thakker in People in AI →

Related posts

Oct 30, 2026

CXMT Q3 2026: Revenue Up 613% to About $16.7B as AI Memory Prices Lift China's DRAM Maker

CXMT, China's largest DRAM maker, reported third-quarter revenue of about $16.7B and a net profit of about $8.17B on rising AI-driven memory prices. This post separates the reported figures from the company's earlier guidance and explains what a slowing growth rate means for AI builders.

Oct 30, 2026

Kioxia Q2: ¥1.995T Operating Profit on AI NAND Demand, but Q3 Guidance Misses; Is 2027 Really "Sold Out"?

Kioxia reported July to September operating profit of about 1.995 trillion yen, a first dividend, and an October to December forecast of about 2 trillion yen, roughly 15 percent below analyst estimates. A claim that its entire 2027 NAND output is sold out is circulating; here is what Kioxia has actually said.

Oct 30, 2026

Runware Wants to Build AI Data Centers Like Cars: 30 MW a Day by 2028, Online in 48 Hours

Runware told TechCrunch it will scale its Romanian factory to 30 megawatts of AI data center pods per day by 2028 and bring them online within 48 hours. We separate the company claims from what its own launch post documents and flag the risks.