Seoul Economic Daily EnglishCodex · 实盘不做把握 5/10不做原因:风险收益比不佳
北京时间·🕒 发布 08:30:18(首尔 09:30)·⟳ 抓取 08:54:32 +24m14s·⚡ 决策 09-03 08:59:45 +5m13s·54.2 天前
NC AI与LG电子入选韩国K-On-Device AI休머노이드国策项目
事件方向正面,但它是政府研发项目入选和技术分工确认,非商业订单、非指引上修,金额与业绩弹性缺失;对NC和LG电子的短线交易edge不足。
📄 原文Seoul Economic Daily English↗ 原文
NC AI, LG Electronics Win State Project for Autonomous Humanoid Robots
NC AI, a physical artificial intelligence company, will develop an autonomous humanoid robot that transfers human movem…
不做 NC(036570.KS)
NC AI入选国策研发项目是正面验证,但未披露金额和商业订单,且子公司体量相对NC游戏主业较小,短线edge不足。
不做 LG电子(066570.KS)
LG电子Physical AI叙事已高传播,本次只是研发项目细分确认,缺少相对公司体量可量化的新增业绩弹性。
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
本次新闻确认NC AI与LG电子在韩国K-On-Device AI项目中承担休머노이드研发任务,能强化NC的Physical AI和LG电子机器人叙事;但韩国政府项目和LG机器人布局在2025-2026年已多次公开,8月31日已有媒体报道最终财团匹配,且本条未披露NC AI或LG电子的收入、利润或订单金额。对NC而言,NC AI账面规模相对游戏主业较小;对LG电子而言,机器人AI已是高传播主题且公司体量大,当前新增事实不足以支持追多。
消息传播链
- Edaily报道NC AI称其与LG电子财团入选K-On-Device AI休머노이드课题,并说明NC AI负责动作重定向。
- 英文版完整传播同一事件,距detect约24分钟被抓取。
- ZDNet此前已报道K-On-Device最终财团匹配,其中LG电子休머노이드课题、机器人公司与Mobilint等参与者已被市场知悉。
- 韩国工业部官方披露K-On-Device AI半导体项目启动框架,面向汽车、家电、机械机器人和国防等终端领域。
关键理由
- 正面因素是NC AI获得国家项目任务,有助于证明其动作数据、world model与机器人重定向能力进入实机研发阶段。
- LG电子在项目中整合机器人硬件、数据、AI模型和国产on-device AI芯片,延续其Physical AI战略。
- 但现有披露只有研发入选和技术分工,未能量化新增收入;相对NC近期季度收入7705亿韩元和LG电子体量,短期业绩增量无法坐实。
主要风险
- 8月31日韩媒已报道K-On-Device最终财团和LG休머노이드课题,9月3日新闻的新鲜度和预期差有限。
- 韩国AI机器人主题在6-8月已有多轮涨跌和高波动,软催化容易冲高回落。
反向证据
- 做多前查反向 · 来源 web · 日期 2026-07-06 · 影响 weakens
韩国AI机器人受益股在Jensen Huang相关主题高峰后大幅回撤,LG电子和斗山机器人从高点显著下跌,提示主题拥挤和软催化兑现风险。
检索:LG Electronics Physical AI robot rally correction risk 2026 原文 ↗ - 做多前查反向 · 来源 web · 日期 2026-08-31 · 影响 weakens
8月31日行业媒体已披露最终财团与LG休머노이드课题,9月3日NC AI新闻存在部分已传播风险。
检索:K-On-Device AI final consortium LG humanoid Mobilint August 31 2026 原文 ↗ - 做多前查反向 · 来源 web · 日期 2026-09-03 · 影响 weakens
可见报道没有披露NC AI或LG电子本次项目的具体金额、收入确认或利润贡献。
检索:NC AI LG Electronics selected project amount revenue contract 原文 ↗
先例如何影响了这次决策
对本次决策的影响:先例使本次不追NC和LG电子;若有直接上市的NC AI或明确订单金额才可能重新评估。
方向:低到中相似样本显示Physical AI合作会带来主题性正反馈,但直接机器人小票弹性更强,大型公司和间接母公司反应不稳定。
涨幅量级:没有8分以上同市场、同机制、同标的阶段的高相似先例,不能据此设定可交易tp/sl幅度。
形态:韩国AI机器人主题此前出现冲高后大幅回撤,说明软催化见光死和拥挤风险较高。
历史先例(按相似度)
- 相似度 5/10 066570.KS 2026-08-14 LG-Nvidia机器人联盟深化后LG电子盘中一度涨近9%,随后回落至约3%涨幅。 原文 ↗
事件后 +1日 +4.1% · +2日 +0.7% · +3日 -2.4% · +4日 -1.9% · +5日 -5.8%
⚠锚前1-2日已异动14.0%>锚定日4.1%·盘中首发但只有日线,混入首发前走势 — 该行涨幅别用来定止盈止损
相似度依据:同为LG电子Physical AI/机器人叙事,但8月事件是Nvidia合作深化与券商上调目标价,商业关注度强于当前研发项目入选。
经验:机器人AI软催化可推动盘中脉冲,但需要更强交易对手或订单化信息才能延续。 - 相似度 5/10 454910.KS 2026-07-30 CMES Robotics宣布与NC AI签署Physical AI合作MOU后盘前触及涨停。 原文 ↗
事件后 +1日 -3.6% · +2日 +7.1% · +3日 +12.0% · +4日 +19.1% · +5日 +27.0%
⚠锚前1-2日已异动13.1%>锚定日3.6% — 该行涨幅别用来定止盈止损
相似度依据:同为NC AI Physical AI合作,但交易标的是机器人公司CMES,直接敞口强于NC或LG电子。
经验:小型机器人直接标的对NC AI合作更敏感,大型或间接标的不能照搬涨停反应。 - 相似度 4/10 066570.KS 2026-07-06 Seoul Economic Daily称LG电子和斗山机器人在Jensen Huang访韩后一个月自高点接近腰斩。 原文 ↗
事件后 +1日 +1.8% · +2日 +5.4% · +3日 -4.2% · +4日 -1.8% · +5日 -0.1%
相似度依据:不是新利好事件,而是同主题交易拥挤后的回撤复盘,可作为风险反例。
经验:Physical AI主题在韩国市场存在预期快速透支和大幅回撤风险。
工具执行结果 · 成功 4 · 降级 0 · 失败 0 · 取消 0
- web_search 成功 ? · 0 ms
- web_search 成功 ? · 0 ms
- web_search 成功 ? · 0 ms
- web_search 成功 ? · 0 ms
原文全文
展开原文 · 3,113 字
NC AI, a physical artificial intelligence company, will develop an autonomous humanoid robot that transfers human movements to machines in partnership with LG Electronics.
NC AI said on the 3rd that it has been selected, as part of a consortium with LG Electronics, to carry out the humanoid segment of the K-On-Device AI Semiconductor Technology Development Project, run by the Ministry of Trade, Industry and Energy and the Korea Evaluation Institute of Industrial Technology (KEIT). The project centers on developing autonomous humanoid robots equipped with domestically made on-device AI chips that can perform human-level tasks in complex everyday spaces.
NC AI will take sole charge of building initial learning-based General Motion Tracking (GMT) data and a human-to-robot retargeting pipeline that transfers human movements to robots. The technology precisely accounts for complex physical variables unique to humanoids — joint limits, body proportions, foot support and balance conditions — so that robots can maintain balance and move reliably in real environments.
Converting human work movements into training data for robots is considered a key element that widens a humanoid's range of motion and underpins precise control. Drawing on motion data assets built up over years and its advanced animation production pipeline, NC AI plans to convert human work movements into data optimized for humanoid training, raising the maturity of full-body control technology.
To that end, the company will build a digital twin environment to secure large volumes of training motion datasets covering a range of scenarios without physical constraints. It also plans to combine its motion control capabilities with its own world model technology, advancing the technology so robots can predict changes in their surroundings on their own and autonomously carry out complex, multi-stage long-horizon tasks in sequence.
The project is seen as significant because LG Electronics and other consortium members will integrate robot hardware, data and AI models, and domestic on-device AI chips into a single development framework. The ultimate goal is to secure, ahead of rivals, the high-performance, low-power on-device AI chips and core robotics technologies essential for real-time motion control in unstructured everyday spaces.
The technology will be applied broadly to general-purpose humanoids that help with organizing, carrying, serving and handling tools in the home, as well as to industrial and public sectors requiring advanced autonomous control, including manufacturing and logistics support, disaster response, facility inspections and work in hazardous areas.
"This project carries great significance in that the motion control technology and data capabilities we have accumulated over a long period will be realized as the autonomous movement of an actual humanoid," said NC AI CEO Lee Yeon-soo. "We will work closely with partners including LG Electronics to maximize the competitiveness of humanoids based on domestic on-device AI chips and take the lead in securing national strategic technologies."
NC AI said on the 3rd that it has been selected, as part of a consortium with LG Electronics, to carry out the humanoid segment of the K-On-Device AI Semiconductor Technology Development Project, run by the Ministry of Trade, Industry and Energy and the Korea Evaluation Institute of Industrial Technology (KEIT). The project centers on developing autonomous humanoid robots equipped with domestically made on-device AI chips that can perform human-level tasks in complex everyday spaces.
NC AI will take sole charge of building initial learning-based General Motion Tracking (GMT) data and a human-to-robot retargeting pipeline that transfers human movements to robots. The technology precisely accounts for complex physical variables unique to humanoids — joint limits, body proportions, foot support and balance conditions — so that robots can maintain balance and move reliably in real environments.
Converting human work movements into training data for robots is considered a key element that widens a humanoid's range of motion and underpins precise control. Drawing on motion data assets built up over years and its advanced animation production pipeline, NC AI plans to convert human work movements into data optimized for humanoid training, raising the maturity of full-body control technology.
To that end, the company will build a digital twin environment to secure large volumes of training motion datasets covering a range of scenarios without physical constraints. It also plans to combine its motion control capabilities with its own world model technology, advancing the technology so robots can predict changes in their surroundings on their own and autonomously carry out complex, multi-stage long-horizon tasks in sequence.
The project is seen as significant because LG Electronics and other consortium members will integrate robot hardware, data and AI models, and domestic on-device AI chips into a single development framework. The ultimate goal is to secure, ahead of rivals, the high-performance, low-power on-device AI chips and core robotics technologies essential for real-time motion control in unstructured everyday spaces.
The technology will be applied broadly to general-purpose humanoids that help with organizing, carrying, serving and handling tools in the home, as well as to industrial and public sectors requiring advanced autonomous control, including manufacturing and logistics support, disaster response, facility inspections and work in hazardous areas.
"This project carries great significance in that the motion control technology and data capabilities we have accumulated over a long period will be realized as the autonomous movement of an actual humanoid," said NC AI CEO Lee Yeon-soo. "We will work closely with partners including LG Electronics to maximize the competitiveness of humanoids based on domestic on-device AI chips and take the lead in securing national strategic technologies."
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。