The Elec EnglishCodex · 实盘不做把握 4/10不做原因:证据不足
北京时间·🕒 发布 17:29:54(首尔 18:29)时区推断·⟳ 抓取 17:31:31·⚡ 决策 08-25 17:38:53 +7m22s·63.1 天前
英伟达Groq 3 LPX量产并获Nebius采用，三星4纳米代工读穿不足以下单
事件方向偏正面，但英伟达主标的相对市值缺少可量化增量，三星供应链读穿缺少金额和利润弹性，Nebius客户侧收益链不直接；同时无真实行情和高相似先例支撑可执行入场。
📄 原文The Elec English↗ 原文
Nvidia Begins Mass Production of AI Inference Rack 'Groq 3 LPX'
Nvidia said on Aug. 24 that it has begun mass production of its rack-scale artificial intelligence (AI) inference accel…
不做 英伟达（NVDA.US）
Groq 3 LPX量产和Nebius采用方向正面，但对超大市值英伟达缺少订单金额、出货量和指引增量，且无法用真实行情设置可执行入场。
不做 三星电子（005930.KS）
三星4纳米代工读穿成立但金额、晶圆量、毛利和占分部收入比例未知，不能证明短线正期望。
不做 Nebius Group（NBIS.US）
Nebius是客户采用方，平台性能提升尚未转化为新增收入或利润承诺，且可能伴随资本开支。
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
Groq 3 LPX量产和Nebius采用把英伟达AI推理硬件从路线图推进到商业部署阶段，The Elec补充三星4纳米代工提供供应链正向读穿；但官方披露没有订单金额、采购承诺、出货量或收入指引，英伟达和三星体量都很大，Nebius也只是客户采用方且收益未量化。直接载体虽覆盖正股与单股杠杆/反向产品，但底层edge不足，最终不做。
消息传播链
- 英伟达官方新闻稿首次确认Groq 3 LPX全面量产，并披露Nebius为首个采用AI云。
- 英伟达博客同步扩散Vera Rubin LPX、Spectrum-X和Nebius等平台采用信息。
- 英伟达技术博客披露3431 tokens/s等benchmark和架构细节，强化产品性能证据但仍未披露商业金额。
- The Elec转述英伟达量产消息并补充Groq 3由三星电子4纳米代工，形成三星供应链读穿。
关键理由
- Groq 3 LPX已量产且Nebius首个采用，证明英伟达Vera Rubin推理平台商业化进度推进。
- The Elec称Groq 3由三星4纳米代工，若放量可改善三星先进制程利用率和AI客户叙事。
- 现有产品销售/公司收入基数巨大，而本事件新增可触达量、采购金额和利润率均未披露，无法估算对英伟达、三星或Nebius收入利润及市值的占比。
主要风险
- 英伟达8月24日官方新闻稿已先于The Elec披露量产和Nebius采用，The Elec的主要新增是三星代工读穿，短线新鲜度有限。
- 若后续披露采购规模显著或三星获得更多AI ASIC客户，需重新评估；当前缺少金额和行情验证，不适合先行下单。
反向证据
- 做多前查反向 · 来源 web · 日期 2026-08-24 · 影响 weakens
英伟达官方新闻稿只有量产、性能和Nebius采用表述，没有最低采购承诺、订单金额或收入指引。
检索:NVIDIA Groq 3 LPX official press release order amount Nebius full production 原文 ↗ - 做多前查反向 · 来源 web · 日期 2026-08-22 · 影响 weakens
英伟达否认中国特供LPU路线，削弱把Groq 3量产解读为出口限制下中国增量需求的逻辑。
检索:NVIDIA Groq 3 LPU China denial Vera Rubin not approved sale China 原文 ↗ - 做多前查反向 · 来源 web · 日期 2026-08-20 · 影响 weakens
公开报道只支持三星4纳米AI客户和产能紧张背景，未给出Groq 3对应晶圆量或三星收入弹性。
检索:Samsung 4nm Nvidia Groq 3 foundry revenue wafer volume 原文 ↗
先例如何影响了这次决策
对本次决策的影响:先例不足叠加金额缺失，使正向事件停留在基本面叙事层，未升级为ENTER。
方向:同类AI芯片/平台量产与客户采用通常方向偏正面，但缺少同市场同相位且首发可核验的高相似样本支持短线交易。
涨幅量级:没有足够高相似先例可用于设定tp/sl；本次又未披露金额，因此不能从历史涨幅外推幅度。
形态:产品性能和量产类新闻在超大市值标的上容易被视为路线图兑现，存在见光后反应有限的风险，但本轮未取得可量化样本。
原文全文
展开原文 · 1,777 字
Nvidia said on Aug. 24 that it has begun mass production of its rack-scale artificial intelligence (AI) inference accelerator, the NVIDIA Groq 3 LPX.
Among the customers disclosed so far is AI cloud company Nebius. Nebius will deploy the Groq 3 LPX in its production inference platform, Nebius Token Factory.
According to Nvidia's internal testing, the system delivered 35 times higher throughput when running AI models with trillions of parameters. Nvidia benchmarked the system using the open-source agentic AI model Gemma 4 31B, achieving an output speed of 3,400 tokens per second. It recorded the fastest performance for the model.
The Groq 3 LPX also offers up to 10 times greater efficiency than existing products. Nvidia said operating Vera Rubin and Groq 3 together could generate an estimated $300 billion in revenue, based on throughput per gigawatt (GW) and estimated cost per million tokens. The comparable figure for Nvidia Blackwell is about $30 billion.
The Groq 3 LPX is designed for AI agent systems. It expands the inference capabilities of the Vera Rubin NVL72 and handles large contexts. Nvidia said it delivers up to four times faster response speeds than alternative platforms for agentic workloads such as coding.
"Groq 3 LPX can push the performance limits of Vera Rubin NVL72 even further," Nvidia CEO Jensen Huang said.
The Groq 3 chip used in the Groq 3 LPX is a language processing unit (LPU) specialized for AI inference. It was co-designed with Nvidia's recently launched Vera Rubin NVL72 rack and optimized to work with graphics processing units (GPUs).
Groq 3 is manufactured by Samsung Electronics' foundry business using a 4-nanometer process. The rack provides a total of 128GB of SRAM and memory bandwidth of 40 petabytes per second (PB/s).
Among the customers disclosed so far is AI cloud company Nebius. Nebius will deploy the Groq 3 LPX in its production inference platform, Nebius Token Factory.
According to Nvidia's internal testing, the system delivered 35 times higher throughput when running AI models with trillions of parameters. Nvidia benchmarked the system using the open-source agentic AI model Gemma 4 31B, achieving an output speed of 3,400 tokens per second. It recorded the fastest performance for the model.
The Groq 3 LPX also offers up to 10 times greater efficiency than existing products. Nvidia said operating Vera Rubin and Groq 3 together could generate an estimated $300 billion in revenue, based on throughput per gigawatt (GW) and estimated cost per million tokens. The comparable figure for Nvidia Blackwell is about $30 billion.
The Groq 3 LPX is designed for AI agent systems. It expands the inference capabilities of the Vera Rubin NVL72 and handles large contexts. Nvidia said it delivers up to four times faster response speeds than alternative platforms for agentic workloads such as coding.
"Groq 3 LPX can push the performance limits of Vera Rubin NVL72 even further," Nvidia CEO Jensen Huang said.
The Groq 3 chip used in the Groq 3 LPX is a language processing unit (LPU) specialized for AI inference. It was co-designed with Nvidia's recently launched Vera Rubin NVL72 rack and optimized to work with graphics processing units (GPUs).
Groq 3 is manufactured by Samsung Electronics' foundry business using a 4-nanometer process. The rack provides a total of 128GB of SRAM and memory bandwidth of 40 petabytes per second (PB/s).
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。