📰 DigiTimes 实时DigiTimes RealtimeDoubao · 实盘不做把握 5/10不做原因:无增量信息
北京时间·🕒 发布 09-14(台北日)仅日期·时区推断·⟳ 抓取 12:23:06·⚡ 决策 09-15 12:27:22 +4m16s·42.4 天前
Sandisk在韩大举招聘NAND/HBM设计工程师，反映AI存储人才争夺升温
这是一篇产业趋势/人才争夺报道：核心事实（Sandisk在韩招工程师）本身无金额、无订单、无指引、无客户定点，对任一上市公司都不构成可量化的短线盈利增量；支撑性数据（FY2026营收、HBF标准化、闪存需求CAGR）均为7月或更早的公开旧信息，新闻已发布约一天（Newsis 9/13首发，DigiTimes 9/14），不具备低扩散预期差。
📄 原文DigiTimes Realtime↗ 原文
Sandisk recruits engineers in Korea as memory talent race heats up
中文網 繁體 简体 SUBSCRIBE My account Sign out SUBSCRIBE Trending Apple Huawei Nvidia AI Search Scaling AI Glasses Inference E…
不做 Sandisk（SNDK.US）
事件本质是工程招聘+行业景气综述，无金额/订单/指引等可量化增量；引用的FY2026营收与HBF进展均为旧信息，软催化对大市值标的缺乏短线edge，且美股当前处盘前不可即时成交，不追长期叙事
不做 SK Hynix（000660.KS）
仅作为HBF合作方与人才竞争背景出现，无公司层面新增事实，HBM/HBF逻辑已为市场充分共识
展开详情(成交明细 / 研究档案 / 原文全文 / 决策记录链)
研究档案
结论摘要
Sandisk在韩招聘NAND/HBM工程师印证AI存储（HBF）架构长期趋势，但属人才布局与行业景气叙事，无新增订单或业绩数字，引用的财务与需求数据均为旧信息，对Sandisk、SK Hynix等均缺乏可交易的短线edge，故回避。
🔎 新鲜度:最早出现 2026-09-13 · 扩散 medium · price-in 风险 medium ↗
关键理由
- HBF/AI推理存储是真实的结构性方向（闪存需求CAGR、数据中心46%、AI推理86%），但仅支持中长期产业判断
- 招聘属先行能力建设，不代表收入兑现，且无任一公司获得可量化新增事实
主要风险
- 若据此趋势追多Sandisk，可能买入已被FY2026业绩和HBF标准化充分定价的旧逻辑
- HBF替代部分HBM的节奏存在不确定性，2038年才可能超过HBM的第三方预测时间跨度极长
反向证据
- 做多前查反向 · 来源 web · 日期 2026-09-14 · 影响 weakens
本报道未给出任何新增订单、客户或上修指引；招聘不等于需求兑现，缺乏做多所需的超预期增量
检索:Sandisk HBF 订单 客户定点 指引 2026 原文 ↗
先例如何影响了这次决策
对本次决策的影响:先例缺位叠加软催化性质，支持回避而非据趋势追多
方向:无可对照的同机制（半导体公司海外工程招聘）价格先例；招聘/人才类事件本身不构成被市场定价的离散催化，历史上多被视为长期投入信号
涨幅量级:缺乏aligned先例，且无增量财务事实，无法校准短线幅度
形态:不适用
工具执行结果 · 成功 1 · 降级 0 · 失败 0 · 取消 0
- search_processed_events 成功 stock_tools HTTP · 14 ms
{"result": "search_processed_events:agent query=`sandisk hbf sk hynix 人才争夺 | sandisk korea recruitment nand design engineers newsis | sandisk 韩国 招聘 nand hbm 工程师`；结构化索引覆盖完整 72 小时；旧记录兼容扫描近 0.0/72 小时、0 条本臂决策(完整)，返回 0 条候选。未命中只表示这些 agent query 表达式没有召回候选，不能据此证明消息首发或不是重复；按当前事件继续正常研判。"}
原文全文
展开原文 · 7,302 字
中文網
繁體
简体
SUBSCRIBE
My account
Sign out
SUBSCRIBE
Trending Apple
Huawei
Nvidia
AI Search
Scaling AI Glasses
Inference Era AI
Advanced Packaging
Trending
SUBSCRIBE
CONNECT WITH US
Sign out
REALTIME NEWS
Corning sees CPO and glass core as new semiconductor battlegrounds
Tomorrow's Headlines
Sep 14, 18:47
China optical transceivers avoid US FCC curbs as NPO and CPO compete in high-speed links
Tomorrow's Headlines
Sep 14, 18:46
China's AIDC token costs hinge on access to advanced GPUs
Tomorrow's Headlines
Sep 14, 18:46
Home
Tech
Semiconductors
Sandisk recruits engineers in Korea as memory talent race heats up
Sherri Wang
, DIGITIMES, Taipei
Sep 14, 2026, 12:20
0
Sandisk Vice President Rajeev Nagabhirava discusses AI memory architecture at the SEMICON Taiwan 2026 Memory Summit. Credit: Sherri Wang
Sandisk is making an unusual push into engineering recruitment in South Korea, seeking talent across NAND system architecture, NAND core design, VLSI design, and mask design while listing experience in HBM, LPDDR, and GDDR among key qualifications.
Newsis reported September 13 that the openings range from internships to experienced positions, including roles requiring at least five years in the field. The move marks a departure for Sandisk's Korean operations, which have historically focused largely on sales and marketing rather than engineering.
The recruitment comes as Sandisk expands beyond conventional NAND into AI-oriented storage and system-level memory architectures. At its 2026 investor day, the company highlighted capabilities spanning NAND intellectual property, front-end manufacturing, controller and systems expertise, back-end manufacturing, and go-to-market operations.
Sandisk's financial performance has also strengthened sharply. Fiscal fourth-quarter revenue rose 50.7% sequentially to US$8.97 billion for the period ended July 3, taking fiscal 2026 revenue to US$20.25 billion, up 175% from a year earlier.
More significant for its engineering needs is the changing role of flash inside AI infrastructure. Sandisk's updated memory hierarchy places direct-attached SSDs, network-attached compute SSDs for inference context storage, and network-attached storage SSDs below system DRAM and GPU HBM, reflecting a growing role for flash in persistent key-value (KV) cache and other inference workloads.
High Bandwidth Flash (HBF) pushes Sandisk further into territory where flash and high-performance memory architectures increasingly intersect. Sandisk says HBF can deliver read bandwidth comparable to HBM with up to eight to 16 times the capacity and is aimed at inference workloads involving large mixture-of-experts models, long context windows, and large KV caches.
The company has outlined architectures in which HBF could augment or replace some HBM stacks, or store model weights and KV cache alongside HBM. That helps explain why experience in HBM and other high-speed memory technologies could be useful to a NAND supplier pursuing higher-bandwidth AI memory designs. The disclosed material does not identify any specific Korean company as a recruitment target.
Sandisk and SK Hynix are already cooperating separately on HBF standardization. Their joint agreement was unveiled in August 2025, followed by the formation of an HBF consortium and a dedicated Open Compute Project workstream in February 2026. The first HBF specification was released in August.
The market backdrop is strengthening demand for this kind of expertise. Kioxia's June investor presentation, citing TechInsights, projected global flash-memory demand to grow at a 22% compound annual rate from 2025 through 2028. Data center flash demand is expected to rise 46% annually, while AI inference-related flash demand grows 86%.
Micron sees a similar structural shift. AI system performance is becoming increasingly dependent on memory subsystem capacity and performance, creating a more complex memory hierarchy and elevating memory's strategic role. The company expects DRAM and NAND supply conditions to remain tight beyond calendar 2027, with NAND growth constrained partly by cleanroom space shifting toward DRAM and HBM consuming a growing share of manufacturing resources.
Competition for engineers is rising alongside that pressure on capacity. Micron has reportedly been approaching employees from Samsung Electronics, SK Hynix, and other Korean semiconductor companies through LinkedIn after recruiting at major Korean universities late last year across more than 10 roles spanning process, quality, and equipment.
Samsung and SK Hynix are stepping up recruitment of their own. Samsung's second-half hiring includes development work on HBM and other high-performance memory for AI, along with process and equipment automation. SK Hynix has been recruiting across design, devices, R&D process, and mass-production technology, while its monthly experienced-hire program covered 14 job categories, including 10 in technical R&D and seven related to HBM.
Fabs, equipment, and capital remain the most visible constraints on memory expansion. But the hiring push in Korea points to another increasingly important bottleneck: engineers able to work across more complex memory and storage architectures. As HBM, DRAM, and flash take on increasingly interconnected roles in data centers, competition for that expertise is likely to intensify.
Credit: Sherri Wang
Article
edited by Ysi Chen
AI Search
Related stories
Samsung pushes memory supply deals as demand spreads beyond HBM
SK Hynix and Sandisk move to standardize HBF, shaking up AI memory ecosystem
Citi flags post-HBM shift as edge memory, HBF advance
HBF likely overtaking HBM market by 2038 as commercialization speeds up
AI memory revolution: Sandisk and SK Hynix define new standard for HBF
Show more
Categories
Chips + components Memory chips
Tags
2026 demand design DRAM flash HBM manufacturing NAND performance SK Hynix South Korea talent
Share this article
Other links
Print
Related topics
Global supply chain: Advanced economies
Asia
Regions
BIZ FOCUS
Sep 14, 08:00
Hauman Unveils Micro LED Testing Breakthrough for AI Optical Interconnects
Monday 14 September 2026
Scaling Photonics: Assembly, Test and Fiber Automation at SEMICON Taiwan
Friday 11 September 2026
DAS Expands in Taiwan and Unveils World First at SEMICON
Friday 11 September 2026
Toward Technologies Redefines Relay Solutions at SEMICON Taiwan 2026
MOST-READ
7 DAYS NEWS
Google's v10 TPU opens door to AMD without sidelining Broadcom, MediaTek
Update: ABF substrate maker Unimicron searched in origin-labeling probe
Qualcomm's HBC play pulls Samsung and SK hynix into a new race beyond HBM
SK keyfoundry expands 8-inch capacity as China demand strains supply
Samsung, SK Hynix step up China NAND upgrades as US tool risks grow
The 1,000-watt problem: a TSMC veteran on the brute force forging Nvidia's AI chips
Wuliangye skips carmaking, turns liquor waste into silicon-carbon anodes
eCloudvalley expands enterprise AI Agent portfolio
CXMT rides AI boom to 80% operating margin
Nvidia, MediaTek deepen 20-year chip ties
Full list
Members only
×
Sorry, the page you are trying to open is available only for our paid subscribers.
Please login to read more
New users, please register first
Notification
×
繁體
简体
SUBSCRIBE
My account
Sign out
SUBSCRIBE
Trending Apple
Huawei
Nvidia
AI Search
Scaling AI Glasses
Inference Era AI
Advanced Packaging
Trending
SUBSCRIBE
CONNECT WITH US
Sign out
REALTIME NEWS
Corning sees CPO and glass core as new semiconductor battlegrounds
Tomorrow's Headlines
Sep 14, 18:47
China optical transceivers avoid US FCC curbs as NPO and CPO compete in high-speed links
Tomorrow's Headlines
Sep 14, 18:46
China's AIDC token costs hinge on access to advanced GPUs
Tomorrow's Headlines
Sep 14, 18:46
Home
Tech
Semiconductors
Sandisk recruits engineers in Korea as memory talent race heats up
Sherri Wang
, DIGITIMES, Taipei
Sep 14, 2026, 12:20
0
Sandisk Vice President Rajeev Nagabhirava discusses AI memory architecture at the SEMICON Taiwan 2026 Memory Summit. Credit: Sherri Wang
Sandisk is making an unusual push into engineering recruitment in South Korea, seeking talent across NAND system architecture, NAND core design, VLSI design, and mask design while listing experience in HBM, LPDDR, and GDDR among key qualifications.
Newsis reported September 13 that the openings range from internships to experienced positions, including roles requiring at least five years in the field. The move marks a departure for Sandisk's Korean operations, which have historically focused largely on sales and marketing rather than engineering.
The recruitment comes as Sandisk expands beyond conventional NAND into AI-oriented storage and system-level memory architectures. At its 2026 investor day, the company highlighted capabilities spanning NAND intellectual property, front-end manufacturing, controller and systems expertise, back-end manufacturing, and go-to-market operations.
Sandisk's financial performance has also strengthened sharply. Fiscal fourth-quarter revenue rose 50.7% sequentially to US$8.97 billion for the period ended July 3, taking fiscal 2026 revenue to US$20.25 billion, up 175% from a year earlier.
More significant for its engineering needs is the changing role of flash inside AI infrastructure. Sandisk's updated memory hierarchy places direct-attached SSDs, network-attached compute SSDs for inference context storage, and network-attached storage SSDs below system DRAM and GPU HBM, reflecting a growing role for flash in persistent key-value (KV) cache and other inference workloads.
High Bandwidth Flash (HBF) pushes Sandisk further into territory where flash and high-performance memory architectures increasingly intersect. Sandisk says HBF can deliver read bandwidth comparable to HBM with up to eight to 16 times the capacity and is aimed at inference workloads involving large mixture-of-experts models, long context windows, and large KV caches.
The company has outlined architectures in which HBF could augment or replace some HBM stacks, or store model weights and KV cache alongside HBM. That helps explain why experience in HBM and other high-speed memory technologies could be useful to a NAND supplier pursuing higher-bandwidth AI memory designs. The disclosed material does not identify any specific Korean company as a recruitment target.
Sandisk and SK Hynix are already cooperating separately on HBF standardization. Their joint agreement was unveiled in August 2025, followed by the formation of an HBF consortium and a dedicated Open Compute Project workstream in February 2026. The first HBF specification was released in August.
The market backdrop is strengthening demand for this kind of expertise. Kioxia's June investor presentation, citing TechInsights, projected global flash-memory demand to grow at a 22% compound annual rate from 2025 through 2028. Data center flash demand is expected to rise 46% annually, while AI inference-related flash demand grows 86%.
Micron sees a similar structural shift. AI system performance is becoming increasingly dependent on memory subsystem capacity and performance, creating a more complex memory hierarchy and elevating memory's strategic role. The company expects DRAM and NAND supply conditions to remain tight beyond calendar 2027, with NAND growth constrained partly by cleanroom space shifting toward DRAM and HBM consuming a growing share of manufacturing resources.
Competition for engineers is rising alongside that pressure on capacity. Micron has reportedly been approaching employees from Samsung Electronics, SK Hynix, and other Korean semiconductor companies through LinkedIn after recruiting at major Korean universities late last year across more than 10 roles spanning process, quality, and equipment.
Samsung and SK Hynix are stepping up recruitment of their own. Samsung's second-half hiring includes development work on HBM and other high-performance memory for AI, along with process and equipment automation. SK Hynix has been recruiting across design, devices, R&D process, and mass-production technology, while its monthly experienced-hire program covered 14 job categories, including 10 in technical R&D and seven related to HBM.
Fabs, equipment, and capital remain the most visible constraints on memory expansion. But the hiring push in Korea points to another increasingly important bottleneck: engineers able to work across more complex memory and storage architectures. As HBM, DRAM, and flash take on increasingly interconnected roles in data centers, competition for that expertise is likely to intensify.
Credit: Sherri Wang
Article
edited by Ysi Chen
AI Search
Related stories
Samsung pushes memory supply deals as demand spreads beyond HBM
SK Hynix and Sandisk move to standardize HBF, shaking up AI memory ecosystem
Citi flags post-HBM shift as edge memory, HBF advance
HBF likely overtaking HBM market by 2038 as commercialization speeds up
AI memory revolution: Sandisk and SK Hynix define new standard for HBF
Show more
Categories
Chips + components Memory chips
Tags
2026 demand design DRAM flash HBM manufacturing NAND performance SK Hynix South Korea talent
Share this article
Other links
Related topics
Global supply chain: Advanced economies
Asia
Regions
BIZ FOCUS
Sep 14, 08:00
Hauman Unveils Micro LED Testing Breakthrough for AI Optical Interconnects
Monday 14 September 2026
Scaling Photonics: Assembly, Test and Fiber Automation at SEMICON Taiwan
Friday 11 September 2026
DAS Expands in Taiwan and Unveils World First at SEMICON
Friday 11 September 2026
Toward Technologies Redefines Relay Solutions at SEMICON Taiwan 2026
MOST-READ
7 DAYS NEWS
Google's v10 TPU opens door to AMD without sidelining Broadcom, MediaTek
Update: ABF substrate maker Unimicron searched in origin-labeling probe
Qualcomm's HBC play pulls Samsung and SK hynix into a new race beyond HBM
SK keyfoundry expands 8-inch capacity as China demand strains supply
Samsung, SK Hynix step up China NAND upgrades as US tool risks grow
The 1,000-watt problem: a TSMC veteran on the brute force forging Nvidia's AI chips
Wuliangye skips carmaking, turns liquor waste into silicon-carbon anodes
eCloudvalley expands enterprise AI Agent portfolio
CXMT rides AI boom to 80% operating margin
Nvidia, MediaTek deepen 20-year chip ties
Full list
Members only
×
Sorry, the page you are trying to open is available only for our paid subscribers.
Please login to read more
New users, please register first
Notification
×
决策记录链 · 只追加
每次领取产生一条记录,重新决策追加一条并以 supersedes 指向被替代的记录(ADR 0014)。
这条决策早于决策账本(2026-10-08 起逐条追加),没有追加记录。