📰 DigiTimes 实时DigiTimes RealtimeDoubao · 实盘不做把握 4/10不做原因:消息已过时
北京时间·🕒 发布 09-07 13:53:00时区推断·⟳ 抓取 09-07 14:05:05·⚡ 决策 09-15 13:18:14 +7d23h·42.1 天前
群联CEO潘健成称NAND供给2027年仍紧缺、公司逢低吸纳中国模组厂抛货，AI营收占比年底或超40%
报道首发于2026-09-07，距当前(09-15)已过6个交易日；内容是CEO在行业论坛的供需观点与定性展望，没有新增订单金额、客户、财报或正式指引等硬催化，属低增量软信息，新闻驱动的短线edge窗口已关闭。
📄 原文DigiTimes Realtime↗ 原文
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 群联电子（8299.TW）
陈旧软催化：9月7日论坛CEO观点，距今6个交易日，无订单/财报/指引硬增量，NAND紧缺叙事已被多篇报道连续消化；台股实时报价本轮超时无法确认剩余空间，风险收益不支持此时追入。
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
群联CEO论坛唱多NAND紧缺至2027、逢低买中国抛货库存、AI营收占比将超40%，方向正面但均为8天前的定性观点，无新增可交易硬信息，短线已过反应窗口，回避。
🔎 新鲜度:最早出现 2026-09-07 · 扩散 medium · price-in 风险 high ↗
关键理由
- NAND紧缺与AI存储需求叙事方向上利好群联等控制芯片/模组厂的定价与毛利
- 群联趁低价吸纳库存有望在价格平台期获得成本优势
主要风险
- 该等观点系论坛演讲而非新公告，38%→40% AI占比与2027紧缺均为既有叙事的延续，无超预期增量
- NAND价格自低点涨约十倍后涨幅放缓，潘健成本人也警示继续大涨会损害终端需求，消费电子需求受高成本压制
反向证据
- 做多前查反向 · 来源 web · 影响 weakens
正文自身承认月度营收出现环比下滑、价格涨幅放缓、高NAND价格压制消费电子需求，且供应商毛利率高达80%正引发大规模扩产，2027供需存反转风险，削弱单边做多
检索:群联 NAND 需求放缓 砍单 库存 月度营收 下滑 2026年9月 原文 ↗
先例如何影响了这次决策
对本次决策的影响:不据此设定任何止盈止损或入场，直接以陈旧+低增量回避
方向:无可用对齐先例；且本事件为过时软观点，缺乏再做事件研究的边际价值
涨幅量级:未校准幅度
形态:未检索到同相位首发先例
工具执行结果 · 成功 3 · 降级 2 · 失败 1 · 取消 0
- search_processed_events 成功 stock_tools HTTP · 14 ms
{"result": "search_processed_events:agent query=`phison k.s. pua nand supply tight 2027 semi forum | phison 群联 买入 低价 nand 库存 ai 营收占比 40% | 群联电子 nand 缺口 2027 采购 中国模组厂 抛货`；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。"} - stock_quote 降级 stock_tools HTTP · 10,099 ms
取价失败: TimeoutError: - stock_quote 成功 stock_tools HTTP · 753 ms
{ - normalize_ticker 成功 stock_tools HTTP · 16 ms
未归一到唯一代码;请用 stock_quote 的 candidates 消歧,或直接给代码。 - stock_quote 降级 stock_tools HTTP · 10,123 ms
取价失败: TimeoutError: - price_snapshot 失败 stock_tools HTTP · 10,135 ms
{
原文全文
展开原文 · 7,367 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
Phison snaps up low-cost China NAND inventory, AI revenue share set to top 40%
Siu Han
, Taipei
Sep 7, 2026, 13:53
0
Phison CEO K.S. Pua expects NAND Flash supply to remain tight into 2027 despite capacity expansion across major memory makers, arguing that forecasts of a rapid return to oversupply are "too naive."
Pua said two NAND suppliers recently told Phison that the size of the 2027 shortfall remains difficult to gauge. At the same time, soaring NAND prices have forced some Chinese consumer module makers to liquidate inventory, giving Phison opportunities to buy lower-priced stock. If supply remains secure, AI-related revenue could rise from 38% currently to more than 40% by the end of 2026.
Pua said major chipmakers, including TSMC, continue to add fabs as the AI ecosystem absorbs large amounts of semiconductor capacity. AI systems also require DRAM and Flash, keeping pressure on memory supply.
Cloud AI is already expanding rapidly, while on-premises AI has yet to reach full scale. Pua noted that PCs and smartphones developed over roughly 20-year cycles, while generative AI is only about four years old. He believes AI could consume semiconductor capacity for another 20-30 years, making a return to memory oversupply within one or two years unlikely.
Memory makers in China, South Korea and Japan, along with Micron, are expanding capacity. With supplier gross margins reaching as high as 80%, Pua said boards naturally have reason to approve more investment.
NAND prices have nevertheless risen about tenfold from their lows, lifting system costs and weighing on consumer electronics demand. Pua views the recent moderation in price increases as healthy, warning that continued sharp gains would damage the broader market.
With supply still constrained, competition is shifting toward access to upstream resources and allocation efficiency. Pua expects operating performance across the memory supply chain to diverge over the next six months. Prices have entered a plateau and should continue edging higher in 2027, reducing the advantage of previously accumulated low-cost inventory and placing greater weight on R&D.
Enterprise and AI demand remains strong, with visibility extending into 2027 and 2028.
High NAND prices are also reshaping China's module market. Pua said some suppliers remain heavily exposed to consumer applications and have yet to build sufficient cloud capabilities, prompting inventory sales to raise cash. Phison has been buying some of those lower-priced NAND stocks.
Phison aims to keep its inventory capacity broadly unchanged at each month-end while securing stable supply. Pua expects 2027 supply to remain extremely tight. If procurement stays on track, AI-related products could exceed 40% of revenue by the end of 2026. He added that Phison's existing enterprise and AI portfolio could support growth for the next three years without adding new product lines.
Phison reads memory slowdown through supply and AI
Pua said the recent pattern of monthly revenue declines alongside year-over-year growth reflects two forces.
First, limited NAND allocations force module makers to choose between maximising near-term sales and preserving inventory for future demand. Second, prices are no longer rising at the previous pace, a development he considers healthier for the industry.
Speaking at the SEMI Semiconductor Masters Forum, Pua joked that Phison has 100% market share "on the Moon and Mars," but said technological showcases do not necessarily make money. Innovation, he argued, must solve real market needs, with the ultimate goal to "survive and make money."
When the current AI cycle began in 2023, Pua initially saw little direct connection between NAND and AI. Phison later concluded that NAND could help offset DRAM limitations in computing workloads.
He cautioned that innovators can become overly convinced of their own technology and misjudge commercial demand. Phison spent more than three years and substantial resources validating its approach internally while facing market scepticism before turning the concept into commercial products.
Pua said execution remains the difference between an idea and a viable product. Phison's solutions now allow small and medium-sized enterprises to deploy on-premises AI with an initial investment of about NT$3-7 million (approx. US$95,000-221,000).
There is also "no permanent blue ocean" in technology, he said. Once a new application proves technically viable and profitable, capital and competitors flood in, followed by price and cost competition.
Phison began with USB products, later captured SSD growth and moved early into on-premises AI. Pua expects that market could become crowded within three to five years, making early positioning critical to securing market share.
Article
translated by Levi Li and edited by Jack Wu
AI Search
Related stories
Phison shifts to system AI solutions with 2027 memory crunch looming
Memory supply crunch pushes Phison to historic earnings
Phison targets AI storage and edge computing growth through Phison 3.0 strategy
Phison says NAND prices jumped 50% overnight as supply crunch continues
NAND flash prices to stay high in 2H25 with substrate shortage to hit soon, says Phison CEO
Show more
Categories
Chips + components Memory chips
Tags
2027 capacity CEO demand inventory low-cost market NAND NAND flash oversupply Phison revenue
Companies
Phison Electronics
Share this article
Other links
Print
Related topic
Global supply chain: China market
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
Column: When digital trust breaks — quantum risk rewrites supply chain resilience
Wuliangye skips carmaking, turns liquor waste into silicon-carbon anodes
The 1,000-watt problem: a TSMC veteran on the brute force forging Nvidia's AI chips
eCloudvalley expands enterprise AI Agent portfolio
CXMT rides AI boom to 80% operating margin
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
Phison snaps up low-cost China NAND inventory, AI revenue share set to top 40%
Siu Han
, Taipei
Sep 7, 2026, 13:53
0
Phison CEO K.S. Pua expects NAND Flash supply to remain tight into 2027 despite capacity expansion across major memory makers, arguing that forecasts of a rapid return to oversupply are "too naive."
Pua said two NAND suppliers recently told Phison that the size of the 2027 shortfall remains difficult to gauge. At the same time, soaring NAND prices have forced some Chinese consumer module makers to liquidate inventory, giving Phison opportunities to buy lower-priced stock. If supply remains secure, AI-related revenue could rise from 38% currently to more than 40% by the end of 2026.
Pua said major chipmakers, including TSMC, continue to add fabs as the AI ecosystem absorbs large amounts of semiconductor capacity. AI systems also require DRAM and Flash, keeping pressure on memory supply.
Cloud AI is already expanding rapidly, while on-premises AI has yet to reach full scale. Pua noted that PCs and smartphones developed over roughly 20-year cycles, while generative AI is only about four years old. He believes AI could consume semiconductor capacity for another 20-30 years, making a return to memory oversupply within one or two years unlikely.
Memory makers in China, South Korea and Japan, along with Micron, are expanding capacity. With supplier gross margins reaching as high as 80%, Pua said boards naturally have reason to approve more investment.
NAND prices have nevertheless risen about tenfold from their lows, lifting system costs and weighing on consumer electronics demand. Pua views the recent moderation in price increases as healthy, warning that continued sharp gains would damage the broader market.
With supply still constrained, competition is shifting toward access to upstream resources and allocation efficiency. Pua expects operating performance across the memory supply chain to diverge over the next six months. Prices have entered a plateau and should continue edging higher in 2027, reducing the advantage of previously accumulated low-cost inventory and placing greater weight on R&D.
Enterprise and AI demand remains strong, with visibility extending into 2027 and 2028.
High NAND prices are also reshaping China's module market. Pua said some suppliers remain heavily exposed to consumer applications and have yet to build sufficient cloud capabilities, prompting inventory sales to raise cash. Phison has been buying some of those lower-priced NAND stocks.
Phison aims to keep its inventory capacity broadly unchanged at each month-end while securing stable supply. Pua expects 2027 supply to remain extremely tight. If procurement stays on track, AI-related products could exceed 40% of revenue by the end of 2026. He added that Phison's existing enterprise and AI portfolio could support growth for the next three years without adding new product lines.
Phison reads memory slowdown through supply and AI
Pua said the recent pattern of monthly revenue declines alongside year-over-year growth reflects two forces.
First, limited NAND allocations force module makers to choose between maximising near-term sales and preserving inventory for future demand. Second, prices are no longer rising at the previous pace, a development he considers healthier for the industry.
Speaking at the SEMI Semiconductor Masters Forum, Pua joked that Phison has 100% market share "on the Moon and Mars," but said technological showcases do not necessarily make money. Innovation, he argued, must solve real market needs, with the ultimate goal to "survive and make money."
When the current AI cycle began in 2023, Pua initially saw little direct connection between NAND and AI. Phison later concluded that NAND could help offset DRAM limitations in computing workloads.
He cautioned that innovators can become overly convinced of their own technology and misjudge commercial demand. Phison spent more than three years and substantial resources validating its approach internally while facing market scepticism before turning the concept into commercial products.
Pua said execution remains the difference between an idea and a viable product. Phison's solutions now allow small and medium-sized enterprises to deploy on-premises AI with an initial investment of about NT$3-7 million (approx. US$95,000-221,000).
There is also "no permanent blue ocean" in technology, he said. Once a new application proves technically viable and profitable, capital and competitors flood in, followed by price and cost competition.
Phison began with USB products, later captured SSD growth and moved early into on-premises AI. Pua expects that market could become crowded within three to five years, making early positioning critical to securing market share.
Article
translated by Levi Li and edited by Jack Wu
AI Search
Related stories
Phison shifts to system AI solutions with 2027 memory crunch looming
Memory supply crunch pushes Phison to historic earnings
Phison targets AI storage and edge computing growth through Phison 3.0 strategy
Phison says NAND prices jumped 50% overnight as supply crunch continues
NAND flash prices to stay high in 2H25 with substrate shortage to hit soon, says Phison CEO
Show more
Categories
Chips + components Memory chips
Tags
2027 capacity CEO demand inventory low-cost market NAND NAND flash oversupply Phison revenue
Companies
Phison Electronics
Share this article
Other links
Related topic
Global supply chain: China market
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
Column: When digital trust breaks — quantum risk rewrites supply chain resilience
Wuliangye skips carmaking, turns liquor waste into silicon-carbon anodes
The 1,000-watt problem: a TSMC veteran on the brute force forging Nvidia's AI chips
eCloudvalley expands enterprise AI Agent portfolio
CXMT rides AI boom to 80% operating margin
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。