Sandisk AI Strategy: A Lights-Out Factory Leads the Way
How Sandisk uses AI on its own work: a 95% lights-out factory in Penang, an internal AI model called PrimeGPT and 32% lower factory costs. Updated November 2026.
Flash memory and storage, including NAND chips, SSDs and memory cards
Sandisk has no AI strategy document for its own work. Its approach comes from a 2025 Dell Technologies case study that quotes Sandisk’s automation and analytics leader.
Read the company sourceWhere Sandisk Is Going with AI
Sandisk uses AI to run its Penang factory almost entirely lights-out. The plant reached 95% lights-out operations, according to a 2025 Dell Technologies case study. Sandisk has published no AI strategy for its own work, so this profile is inferred from that case study and the Sandisk remarks in it.
Generative AI reaches the factory team as well as the machines. Peng Koon Hew, Senior Director of Automation and Analytics, said in the case study:
“Ranging from operations to engineering, GenAI has proven to be advantageous for our factory team.”
The focus has four parts.
- Factory automation: robotics and AI that let 95% of operations at the Penang plant run lights-out
- Engineering and operations: PrimeGPT, an internal multimodal AI model that helps staff find information in standards documents, write code and work on product designs
- Energy: Vision AIoT, which uses factory sensor data to switch lights on only when people are present
- Quality and planning: AI and analytics that cut defects, lead times and inventory
What It Takes: Its Own AI Platform, Generative AI for Staff and Outside Recognition
Sandisk built its own AI platform for the factory. The Sandisk AI Platform runs on Dell servers with NVIDIA accelerated computing and NVIDIA AI Enterprise software. It combines on-premises and cloud systems to keep data secure.
Factory staff use generative AI in daily work. Hew said staff use PrimeGPT to pull insights from industrial standards documents, write code more efficiently and reach secure databases for product design work.
Outside recognition backs the factory claims. The World Economic Forum named the Penang factory to its Global Lighthouse Network, as Asia’s first Sustainability Lighthouse and Malaysia’s first Advanced Industry 4.0 Lighthouse.
The next step is agents. The case study calls PrimeGPT a big step toward agentic AI, which puts Sandisk at the gap between chat and agents.
Headcount has held steady while revenue tripled. Sandisk had about 11,100 employees at the end of fiscal 2026, against about 12,000 two years earlier. The sources reviewed do not link Sandisk’s headcount to AI.
AI fluency, coded against the Lead with AI Fluency Matrix (September 2026 edition):
- Fluency floor: Not published. Sandisk has set no AI training requirement or usage standard for all staff.
- Highest level evidenced: Level 3 (Integrating), company-claimed. AI is built into factory operations and engineering work with measured results. Lights-out operations rest on robotics and automation as well as AI, and Sandisk has not shown AI-run workflows outside the factory, so Level 4 is not evidenced.
- Portfolio design: Sandisk concentrates AI in manufacturing, where automation, quality and energy savings drive results.
How Sandisk Measures Results
Sandisk reports factory results and no company total. All figures below are company-claimed, from the 2025 case study:
- Automation: lights-out operations up from 80% to 95%
- Cost: factory costs down 32%
- Energy and emissions: energy use down 46% and CO2 emissions down 45%
- Quality: defective parts per million down from 800 to 100
- Delivery: 99% on-time delivery, with lead time down 54.5%
The results cover one factory. Sandisk has not published AI results for engineering, sales or corporate functions, or a company-wide figure. Extending the measures beyond Penang is the step from AI adoption to AI value.
Next markers: AI results outside Penang, agentic uses of PrimeGPT, an AI strategy for the whole company, and headcount for fiscal 2027.
Change log
- February 2025: Sandisk separates from Western Digital.
- August 2025: A Dell Technologies case study reports 95% lights-out operations and 32% lower costs at Penang.
- July 2026: Fiscal 2026 ends with revenue of about $20.2 billion and about 11,100 employees.
The Bottom Line
- Disclosure: 3 of 4. Sandisk answers where it will focus and how it measures results, and in part what it will take, through a vendor case study. It has published no AI strategy for its own work.
- The Penang factory runs 95% lights-out, with AI in quality, energy and planning.
- PrimeGPT gives factory staff a generative AI tool for documents, code and design.
- Factory costs fell 32%, by Sandisk’s count.
- The say-do gap is scope. Strong factory results stand alone, with no published AI plan for the rest of the company.
- Watch the rest of the company. The test is whether Sandisk takes its factory approach to engineering and corporate work.
Revenue, Headcount and Revenue per Employee
What the numbers show: Revenue tripled from fiscal 2024 to fiscal 2026 while headcount fell 7%, so revenue per employee rose more than threefold, to about $1.82M. The jump mostly reflects flash memory prices, which rose sharply in fiscal 2026. These figures do not show that AI caused any change.
Revenue, headcount and revenue per employee, FY2024 to FY2026
- FY2024: $6.66bn revenue, 12,000 employees, $555k per employee
- FY2025: $7.36bn revenue, 11,000 employees, $669k per employee
- FY2026: $20.25bn revenue, 11,100 employees, $1.82M per employee
Change, FY2024 to FY2026: revenue up 204% (3.0x), headcount down 7%, revenue per employee up 229% (3.3x).
Notes: Revenue is as reported. Headcount is approximate, at fiscal year-end (late June or early July). Sandisk separated from Western Digital in February 2025, so fiscal 2024 is a carve-out from Western Digital’s accounts. Only three years are shown because Sandisk reported no standalone headcount before fiscal 2024. Fiscal 2026 had 53 weeks. Flash revenue moves with memory prices, which rose sharply in fiscal 2026.
These figures give context on scale and productivity. They do not show that AI caused any change. Sources: Sandisk 10-K and Form 10 filings.
Frequently Asked Questions
What is Sandisk’s AI strategy for its own work?
Sandisk uses AI mainly to automate and improve its factories. Its Penang plant runs 95% lights-out, and staff use an internal generative AI model called PrimeGPT.
What is PrimeGPT?
PrimeGPT is Sandisk’s internal multimodal AI model. Factory staff use it to find information in standards documents, write code and work on product designs.
Who leads AI at Sandisk?
CEO David Goeckeler sets the company’s direction. Peng Koon Hew, Senior Director of Automation and Analytics, described the factory AI work in 2025, and the sources reviewed name no company-wide AI leader.
Is Sandisk cutting jobs because of AI?
The sources reviewed do not link Sandisk’s headcount to AI. Headcount was about 11,100 at the end of fiscal 2026, against about 12,000 two years earlier.
How is this profile put together?
Lead with AI profiles use public sources only, such as annual reports, earnings materials, filings and on-record interviews, dated at the time of writing. AI fluency is coded against the Lead with AI Fluency Matrix, and company figures are marked as company-claimed unless they have been independently verified.
About the AI Strategy Index
The Lead with AI AI Strategy Index profiles how the largest companies in the US and Europe approach AI, starting with the S&P 100 and the STOXX Europe 50. Each profile answers four questions: where the company is going with AI, where it will focus, what it will take and how it will know it is working. AI Pacesetters and notable companies outside the two indexes are profiled for contrast and are not counted. Profiles use public sources only, and AI fluency is coded against the Lead with AI Fluency Matrix.
See all company profiles