7 posts tagged with "Broadcom"

2025 年 10 月 13 日，OpenAI 与 Broadcom 宣布共同开发并部署 10 吉瓦（GW）自研 AI 加速器，消息当日 Broadcom 股价涨超 10%。一年后的现在，这份协议进入真正的考验期：2026 下半年开始实际部署机架，2029 年底前完成全部 10GW。

一、协议要点回顾​

要素	内容
规模	10GW（约等于 800 万美国家庭用电，或 5 座胡佛大坝）
分工	OpenAI 主导芯片与系统设计，Broadcom 负责共同开发与部署
组网	Broadcom 以太网方案承担 Scale-up 与 Scale-out（非 InfiniBand）
时间线	2026 H2 开始部署 → 2029 年底完成
芯片定位	以推理为主的第一代专用加速器

二、推理专用：为什么第一颗芯片不做训练​

多家供应链报道指向同一个设计取向：OpenAI 第一代 ASIC 砍掉 FP64 流水线与通用图形逻辑，把晶圆面积全部留给推理负载——低秩注意力加速单元（面向 MLA 与低精度 KV Cache 解压）、放大片上 SRAM 以减少 HBM 往返、面向自回归解码的 INT4/FP8/FP4 量化矩阵单元。解码阶段的瓶颈在访存带宽而非算力密度，这一取舍与推理接棒训练的行业趋势互为印证。

工程团队由前谷歌 TPU 架构师 Richard Ho、Thomas Norrie 领衔，据报道已扩张至 40 余名 ASIC 设计人员；代工锁定台积电 N3P（后续 N2），并预留了 CoWoS-L 封装产能——先进封装而非前道制程，才是当前 AI 芯片产能的真瓶颈。

三、OpenAI 的三线算力版图​

自研芯片只是 OpenAI 算力组合的一根支柱：

• NVIDIA：计划投资至多 1000 亿美元，供应至少 10GW 数据中心系统
• AMD：6GW 芯片供应协议，含 OpenAI 认购 AMD 股权的期权
• Broadcom：10GW 自研加速器（本文主角）

三份多吉瓦级协议在数周内接连落地，OpenAI 同时成为英伟达最大客户、AMD 大客户和芯片设计方——这是对「单一供应商锁定」的对冲，也是对自身推理成本结构的长期下注。

四、对行业的三个信号​

1. 以太网 Scale-up 挑战专有互联：如果 OpenAI 机架全以太网组网并达到性能预期，将强化开放标准路线的议价能力
2. ASIC 从「降本工具」变「战略资产」：谷歌 TPU 已向 Anthropic 供应百万颗量级，OpenAI 再入场，自研芯片正式成为头部 AI 公司的标配
3. 资金问题仍未回答：OpenAI 未披露财务条款，多数分析师认为自研芯片短期内不改变英伟达的市场主导地位，但 2027-2028 年推理侧的份额争夺会真正开始

延伸阅读：谷歌 TPU 外部客户版图 · 微软 Maia 300 发布 · 自研芯片对比

截至 2026 年 10 月，OpenAI 的算力采购版图已经扩展到接近 26 吉瓦，其中自研芯片项目在 10 月迎来两个实质性节点：硬件平台选型曝光，融资安排浮出水面。这不再是简单的「买卡」故事，而是一家模型公司在重写自己的算力供应链。

一张吉瓦级采购清单​

OpenAI 在过去一年里先后与三家芯片供应商签下多吉瓦级承诺。2025 年 10 月宣布的博通合作是自研芯片的起点：双方共同开发 10 吉瓦定制 AI 加速器系统，博通负责硅实现与以太网组网，OpenAI 负责架构设计，部署窗口为 2026 年下半年启动、2029 年底完成。几乎同一时间，OpenAI 与 AMD 达成 6 吉瓦 Instinct GPU 协议，首批采用 MI450，并附带最高 1.6 亿股 AMD 股票的认股权证，按容量交付情况分批归属。英伟达方面承诺至少 10 吉瓦系统，并拟按每个吉瓦的实际部署进度逐步投资不超过 1000 亿美元，首个吉瓦计划于 2026 年下半年在 Vera Rubin 平台上线。

供应商	承诺规模	芯片/平台形态	关键条款
英伟达	至少 10 吉瓦	Vera Rubin 系统	拟投资不超过 1000 亿美元，按吉瓦进度释放
博通	10 吉瓦	OpenAI 自研定制加速器	2026 下半年部署，2029 年底完成
AMD	6 吉瓦	Instinct GPU，首批 MI450	含最高 1.6 亿股 AMD 股票认股权证

三份协议加起来接近 26 吉瓦。据公开报道，多家机构把这个体量换算成用电规模——单吉瓦 AI 数据中心的芯片采购成本就可能达到数百亿美元级别，26 吉瓦对应的硬件投入接近万亿美元量级。需要说明的是，其中相当一部分「已承诺容量」是分期交付的框架额度，而非一次性交付的现货。

Jalapeño 主机平台选型曝光​

10 月 4 日披露的一组细节，首次让人看清 OpenAI 自研推理芯片的机架级形态。据报道，这款代号 Jalapeño 的推理 ASIC 放弃了英伟达 Vera Rubin 加速器作为主机平台，改用 AMD EPYC Turin 处理器，每台主机搭载两颗 Turin 级 CPU 与 1.5TB 内存，经八根 PCIe 线缆连接到加速器托盘；相邻机架容纳 128 颗 Jalapeño 芯片，最多可通过 scale-up 网络互联 2048 颗加速器。功耗分配上，主机甲架约 31 千瓦，加速器机架约 130 千瓦，配对合计约 160 千瓦。

这套设计的逻辑是让 CPU 侧承担分词、请求处理、编排调度工作，避免昂贵加速器空转等待。Jalapeño 从设计到流片仅用九个月，博通负责硅实现与网络，Celestica 负责板卡与系统设计。OpenAI 硬件负责人将选择成熟 x86 平台称为务实之举——加速器本身没有 CUDA 那样的生态，软件栈验证容错空间极小，因此主机侧更倾向使用已有大量使用经验的成熟平台。值得注意的是，该项目内部代号为 Nexus，第一代与第二代芯片分别叫 Jalapeño 与 Serrano。

融资：从「买设备」到「买资产」​

10 月 8 日的消息把这条线索补完了。据《华尔街日报》与彭博社报道，博通近几周一直在为 OpenAI 的定制 AI 芯片安排超 500 亿美元融资，已与阿波罗全球管理、黑石集团等私募信贷机构展开早期洽谈，谈判尚处初期，规模可能变化；彭博社同期称早期磋商显示双方可能寻求筹集约 300 亿美元债务，尚未启动正式流程。

这并非孤立事件。博通此前已为 Anthropic 安排大规模债务融资，据报道正为 Anthropic 及其他公司筹集 600 亿美元新 AI 芯片融资，其中 A 类优先担保贷款银团分销函金额达 420 亿美元，黑石牵头 180 亿美元 B 类次级债务，并承诺从旗下基金出资 90 亿美元。甲骨文则与阿波罗、高盛洽谈专门融资，用于为一座 1 吉瓦数据中心采购芯片，结构可能是私人投资者设立表外实体购入芯片再租回给甲骨文，以填补硬件付款与云收入之间的时间缺口。

研判​

算力采购的竞争已经从芯片规格转向融资能力。当芯片厂商能把「交付硬件」和「安排资金」打包成一份方案，它锁定的就不只是订单，而是客户未来数年的路线图依赖。OpenAI 三线并行的策略本质是风险对冲，但代价是资本结构的复杂度同步上升——26 吉瓦的承诺对应的是尚未完全验证的收入能力。

真正的观察点在 2026 年下半年：Jalapeño 能否按期点亮送样、Vera Rubin 首个吉瓦能否如期上线、私募信贷对 AI 算力资产的风险偏好是否延续。这三件事的答案，会决定自研 ASIC 是真的降低成本，还是只是把成本挪到了资产负债表上。

英伟达的GPU是"通用答案"，而云巨头们心里都藏着一个念头：为自家工作负载量身定做一颗芯片，长期成本更低、能效更高。承接这个念头的，是两家设计服务巨头——博通与Marvell。它们不站聚光灯下，却决定着每一颗定制AI芯片的诞生。

双寡头格局：一份不对称的订单版图​

据多家行业机构测算，博通与Marvell合计占据定制AI ASIC设计服务市场六成以上份额，其中博通一家独占约55%至60%，Marvell约占15%至20%。第三方机构Counterpoint Research预计，到2027年博通仍将以约60%的份额领跑AI服务器计算ASIC设计合作伙伴市场。

两者的差距主要体现在收入体量上。据公开披露与行业估算，博通2025财年AI相关收入约200亿美元，同比增长65%，2026年预计迈向250亿美元；Marvell同期AI收入约39亿美元，目标超过50亿美元。博通的AI收入大约是Marvell的4到5倍。

维度	博通	Marvell
AI ASIC市场份额	约60%	约15%至20%
2025财年AI收入	约200亿美元	约39亿美元
核心客户	谷歌、Meta、OpenAI、字节	亚马逊、微软、谷歌
标志项目	谷歌TPU全系、OpenAI十吉瓦	Trainium、Maia 2026量产
收入目标	2026年约250亿美元	2028年70至80亿美元

博通：深度绑定式的碾压​

博通的霸权建立在与谷歌长达十年的深度绑定之上：从第一代TPU开始，博通全程参与设计与制造，双方合作已推进至3纳米工艺的第六代TPU。据公开报道，2023年谷歌在TPU相关合作上向博通支付约35亿美元，2024年翻倍至约70亿美元，第七代TPU的供应合同也已提前锁定。除谷歌外，Meta的MTIA、字节的自研芯片均出自博通之手；2025年博通更与OpenAI启动了10吉瓦级定制加速器项目，整套系统依托博通的以太网方案扩展集群。

博通的护城河不止于设计能力：业界最全的高速SerDes IP、HBM控制器IP与先进封装经验，构成定制芯片的"标准零件库"；"定制加速器加高速互连"的组合拳让客户难以拆分采购。这种模式不承担库存风险、不赌制程迭代，是典型的卖铲子生意。

Marvell：差异化补位的追赶者​

Marvell的策略是不与博通正面硬刚，走差异化路线。它给亚马逊量产Trainium训练芯片（据行业报道占AWS自研ASIC出货量的绝大部分），拿下微软Maia芯片2026年的量产订单，同时为谷歌代工Axion ARM CPU——与博通的TPU业务形成互补而非对撞。2026财年第一财季，Marvell数据中心收入14.41亿美元，同比增长76%，占总营收76%，几乎全部由AI定制芯片大规模出货驱动。

技术上，Marvell在数据中心互连与光电共封装领域积累深厚，其定制的112G SerDes与Die-to-Die互连技术在能效与互联带宽上形成差异化卖点，公司也据此定下2028年AI收入冲刺70至80亿美元的目标。

研判：双雄之外，第三势力已在敲门​

这场双雄之争并非终局。联发科已切入谷歌第八代TPU的I/O Die设计，预计2026年第四季度开始贡献营收，其管理层公开表示有信心在2026年实现超过10亿美元的数据中心ASIC营收——SerDes技术正是它敲开大门的钥匙。对博通而言，谷歌分散供应商的意愿是最大的长期变量；对Marvell而言，能否把亚马逊、微软之外的第四、第五个大客户装进管道，决定它能否真正缩小与博通的量级差距。ASIC战争的本质是信任战争：云厂商每多押一家设计伙伴，都是在为供应链的议价权与冗余度下注。

Hot Chips 2026 上，OpenAI 与博通合作的首颗物理芯片 Jalapeño 完整规格曝光：700W 纯推理 ASIC、216GB HBM4、13.4 PFLOPS MXFP4，从初始设计到流片只用了约 9 个月。

Jalapeño 是什么​

综合 Data Center Dynamics、Tom's Hardware、ServeTheHome 在 Hot Chips 2026 上的报道，Jalapeño 是 OpenAI 与博通合作的首颗物理芯片，定位非常纯粹：一颗纯推理 ASIC，功耗 700W，专门针对 OpenAI 大规模推理服务（如 ChatGPT 等）的实际瓶颈设计。

与通用 GPU 不同，Jalapeño 不追求覆盖训练、微调、推理的全场景，而是把全部资源押在一件事上：以最低成本、最高密度跑好 OpenAI 自己的推理负载。这是自研芯片路线中最激进的一种——不做通用产品，只为自家业务画像画芯片。

完整规格​

项目	规格
定位	纯推理 ASIC（OpenAI × 博通首颗物理芯片）
功耗	700W
内存	216GB HBM4
算力（MXFP4）	13.4 PFLOPS
算力（MXFP8）	3.4 PFLOPS
架构	NUMA 式架构，64 个内存与计算切片
流片速度	从初始设计到流片约 9 个月
工艺	博通 3nm/2nm 混合设计，台积电代工
首批部署	目标 2026 年底

架构上，Jalapeño 采用 NUMA 式设计，共 64 个内存与计算切片。这种把内存与计算绑定为切片、再横向扩展的思路，本质上是围绕 OpenAI 自己的模型结构和流量特征做定制——博通在定制 ASIC 上最擅长的，正是这种"按客户负载画芯片"的方法论。

集群规模：单 pod 最高 27 EFLOPS​

单颗芯片只是起点，Jalapeño 的部署规划更能说明 OpenAI 的推理盘子有多大：

• 128 颗芯片组成的部署规模约 1.7 EFLOPS FP4，合计 27.5TB HBM4；
• 单个 pod 最多容纳 2048 颗加速器，最高 27 EFLOPS MXFP4。

作为背景，OpenAI 与博通的定制算力协议总规模约为 10GW，Jalapeño 只是这盘棋的第一颗子。如果 2026 年底首批部署如期落地，OpenAI 的推理成本结构将出现一个全新的变量。

背景与资金面​

围绕这颗芯片的产业与资金背景同样值得梳理：

• OpenAI 与博通签有约 10GW 算力的定制协议，芯片采用博通 3nm/2nm 混合设计、台积电代工，目标 2026 年底首批部署；
• 博通正为 OpenAI 定制芯片筹资超 500 亿美元，其中早期磋商约 300 亿美元为债务融资，接续此前为 Anthropic 安排的债务融资；
• FT 报道 OpenAI 年化营收约 500 亿美元，低于此前约 700 亿美元的口径；
• 与此同时，Meta 也在与博通合作开发下一代 MTIA 芯片。

定制芯片是资金密集型游戏：芯片要钱，HBM 要钱，数据中心还要钱。博通为 OpenAI 安排的数百亿美元债务融资，说明这类项目的资金结构已经接近基础设施建设，而非传统的芯片采购。

分析：专用化的赌注​

**纯推理专用 vs GPU 通用性。**已有分析师对 Jalapeño 的路线提出质疑：纯推理 ASIC 缺乏灵活性，一旦模型结构或推理范式发生变化，芯片可能无法跟上。OpenAI 的赌注是：自家推理需求足够大、足够可预测，专用化带来的成本与密度收益足以覆盖灵活性损失。这是一个用业务确定性换取硬件效率的交易——赌对了，推理成本大幅下降；赌错了，芯片就是沉没成本。

**9 个月流片的工程方法论。**reticle 级定制 ASIC 通常需要数年才能流片，Jalapeño 从初始设计到流片约 9 个月，是异常速度。这背后是三重支撑：博通成熟的 IP 库与设计流程、台积电先进工艺的稳定保障，以及 OpenAI 作为单一客户时的高度确定性需求——需求越明确，设计迭代越少，周期越短。

**博通成为定制 ASIC 时代最大赢家。**OpenAI 的 Jalapeño、Meta 的下一代 MTIA、Google 的 TPU，再加上为 Anthropic 安排的债务融资——几乎所有超大规模玩家的定制芯片路线，最终都汇入博通的订单簿。GPU 时代的赢家是英伟达，定制 ASIC 时代的赢家，很可能是那个卖铲子的博通。

小结​

Jalapeño 的完整规格披露，让外界第一次看清 OpenAI 自研芯片的成色：700W、216GB HBM4、13.4 PFLOPS MXFP4，9 个月流片，单 pod 最高 27 EFLOPS MXFP4。它不追求通用，只追求把 OpenAI 自己的推理账单打下来。这颗芯片能否成功，取决于推理需求是否如 OpenAI 预期的那样持续且可预测；而无论成败，博通都已经稳稳站在了赢家的位置上。

据IEEE Spectrum 2026年9月14日报道，OpenAI的Jalapeño AI加速器正逐步揭开面纱。这不仅是一颗芯片的诞生，更是一场芯片设计方法论的革命——由LLM深度参与设计、不足百人团队、从架构概念到首次流片不足20个月。

Jalapeño的核心规格​

根据OpenAI官方口径，Jalapeño AI加速器的关键规格如下：

规格项	Jalapeño参数
峰值算力	最高13.4 petaflops（4-bit精度）
内存容量	232GB
内存带宽	15.4TB/s
设计定位	专为LLM推理设计
合作伙伴	博通（Broadcom）
研发周期	从首个架构概念到首次流片不足20个月
团队规模	平均不足100人

其中最引人注目的是4-bit精度下13.4 petaflops的峰值算力。这一数字定位清晰：Jalapeño并非追求通用计算的"全能选手"，而是聚焦于大语言模型推理这一特定场景，在低精度计算上做到极致。232GB内存与15.4TB/s带宽的组合，则为KV Cache密集型的长上下文推理提供了吞吐保障。

性能对标：较GB300最高降低3.6倍延迟​

OpenAI引用的基准测试显示，在端到端延迟方面，Jalapeño较NVIDIA GB300最高降低3.6倍。对于推理业务而言，延迟直接决定用户体验——聊天机器人的首字响应、Agent系统的多轮迭代速度，都取决于每一毫秒。

需要指出的是，这一数据来自OpenAI自己的引用口径，测试条件与模型负载尚未完全公开。但即便打上折扣，一颗自研ASIC在特定推理负载上跑赢NVIDIA旗舰级GB300平台，本身就说明定制芯片在"场景足够聚焦"时确实具备结构性优势。

与博通合作：多代规划与吉瓦级部署​

Jalapeño由OpenAI与博通合作设计，规划了多代产品路线，目标直指吉瓦（GW）级部署。据公开报道，该芯片据称搭配AMD EPYC Turin主机CPU，而非完全依赖NVIDIA系统——这意味着OpenAI正在构建一条绕开NVIDIA生态的完整推理基础设施。

博通作为全球定制ASIC（XPUs）设计服务的头部玩家，此前已深度参与多家云巨头的自研芯片项目。OpenAI选择博通，等于站在了成熟的多项目晶圆与SerDes/IP积累之上，这也是20个月流片速度的重要支撑。

自定义芯片的适用条件​

Jalapeño的故事之所以成立，背后有一组苛刻的前提条件，并非所有公司都能复制：

1. 海量且可预测的推理需求。ChatGPT的token消耗量巨大且增长曲线稳定，只有这种量级才能摊薄流片与迭代成本。
2. 稳定的模型架构。Transformer架构多年未发生根本变化，才让针对特定计算模式的ASIC有了"押注"的价值。
3. 全栈软件掌控。OpenAI同时拥有模型、推理框架和基础设施，软硬件协同优化可以贯穿到底。
4. 足够的部署量。吉瓦级部署规划意味着单位芯片成本可以压到极低。

四个条件缺一不可。缺了需求，流片成本摊不掉；缺了架构稳定，芯片投产即过时；缺了软件栈，性能优势发挥不出来。

定制ASIC不会取代GPU​

一个常见的误读是"自研芯片将终结GPU"。但行业现实是：定制ASIC与GPU长期共存、各司其职。

• 训练仍是GPU的主场。训练负载多变、需要频繁调试与实验，通用性至关重要。
• 多模态与快速演进的架构同样属于GPU。新算子、新注意力变体往往先在GPU上验证。
• 推理中负载高度稳定、模型固化的部分，才是ASIC的收割区。

换句话说，NVIDIA的GPU生态不会因为Jalapeño而瓦解，但推理市场的价值分配确实开始被重新切分。对NVIDIA而言，最大的挑战或许不是失去训练市场，而是越来越多头部客户把"最大的一块推理蛋糕"留给自己。

LLM设计芯片：方法论的革命​

比规格更值得关注的，是Jalapeño的设计方式。据IEEE Spectrum报道，OpenAI的LLM深度参与了芯片设计——从首个架构概念到首次流片不足20个月，团队平均不足100人。传统芯片设计动辄数百人团队、三到五年的开发周期，Jalapeño把这个数字压缩到了令人难以置信的程度。

LLM辅助设计带来的是研发效率的量级变化：架构探索空间的快速评估、RTL代码生成与验证、设计文档与工具链自动化。如果这套方法论被验证可复制，那么芯片设计的门槛将被实质性改写——未来不仅是"大厂造芯"，而是"AI造芯"成为常态。

小结​

Jalapeño是2026年定制ASIC浪潮中最具标志性的产品之一：13.4 PFLOPS 4-bit算力、232GB内存、15.4TB/s带宽的规格专为LLM推理而生，与博通的合作、多代规划与吉瓦级部署展现了OpenAI摆脱NVIDIA依赖的决心。而LLM参与设计、不足百人团队、20个月流片的研发模式，其影响或许远超这颗芯片本身。定制ASIC不会取代GPU，但它正在重新定义推理经济的成本结构——对整个AI算力产业而言，这是一场刚刚开始的游戏。

The Tuesday-afternoon AI session at Hot Chips 2026 this August was the most historically significant of the conference — not because any single chip was so powerful, but because almost everything on stage was a "hyperscaler de-NVIDIA-ization" custom ASIC: Google's 8th-gen TPU, OpenAI's first self-designed chip, Microsoft Maia, Meta MTIA, and Cerebras wafer-scale racks, all on one stage. When the world's largest AI compute buyers start treating GPUs as "one of the options," the power structure of AI hardware is loosening.

---

1. OpenAI Jalapeno: Building a Chip in 9 Months​

On June 24, 2026, OpenAI, together with Broadcom, unveiled its first self-designed inference ASIC, Jalapeno — the fifth member of the "custom inference chip club."

Dimension	Jalapeno
Partner	Broadcom + TSMC manufacturing
Positioning	Inference-specific ASIC
Design cycle	9 months end-to-end (Greg Brockman says aided by OpenAI's own models)
Cost target	~50% lower token cost vs general-purpose GPU stack
Commercial timing	First deployments by end-2026; long-term goal 10GW of self-designed chips
Deal scale	Up to $10B strategic partnership with Broadcom (accelerators + networking by 2029)

The talk title "You Can Just Build Things … Chips" is itself a signal: the largest AI compute buyer no longer defaults to GPU as the only path.

---

2. Google TPU v8: The Biggest Architectural Pivot in a Decade — Train/Infer Split​

Google has the longest custom-chip history (2016 to now), and its 8th-gen TPU for the first time splits the product line in two:

Model	Codename	Partner	Positioning	Key Specs
TPU 8t	Sunfish	Broadcom	Training	9,600 cards per pod, 121 FP4 ExaFLOPS, 2PB shared HBM, 2× ICI bandwidth
TPU 8i	Zebrafish	MediaTek	Inference	288GB HBM, 384MB on-chip SRAM (3× prior gen), 19.2 Tb/s ICI

On capacity, Morgan Stanley estimates based on supply-chain interviews that Google TPU production in 2026 may exceed 3 million units (a brokerage estimate, not an official target). Google is also the only vendor to achieve large-scale custom-chip deployment and sell compute externally (Gemini runs on TPUs).

---

3. Meta MTIA: From Recommendation Systems to a GenAI Dual Mission​

Meta's custom journey has the clearest starting point — MTIA was originally built for recommendation ranking hardware and is being pulled toward a dual mission by generative AI.

• MTIA 300 is deployed; 400 / 450 / 500 are planned at roughly one new model every 6 months through 2027;
• Based on RISC-V, Meta claims up to 25× compute gain;
• Node evolves with industry cadence: 100 (7nm) → 200 (5nm) → 300 series (3nm + CoWoS);
• In partnership with Broadcom; another chip codenamed Iris reportedly passed testing in July 2026;
• Meta plans to start volume production of one of them in September 2026, doubling its overall compute.

---

4. Microsoft Maia 200/300: Most Advanced Deployment​

Microsoft's Maia 200, released January 26, 2026, is the most advanced in deployment among the four:

Dimension	Maia 200
Process	TSMC 3nm, 140B+ transistors
Compute	10+ PFLOPS FP4 / 5 PFLOPS FP8
Memory	216GB HBM3E, 7 TB/s
Power	750W
Deployment	Already running in Des Moines data center, serving OpenAI GPT-5.2 and Microsoft 365 Copilot

Microsoft claims roughly 3× the performance of Amazon's Trainium on specific benchmarks. The short-term strategy is a dual track of "self-designed Maia + purchased NVIDIA" in parallel — self-designed chips need time from design to mass production, and NVIDIA's mature ecosystem cannot be replaced in the short term.

---

5. Amazon Trainium 3 and Anthropic's In-House Team​

• Amazon: The Trainium series is already commercial, with 1.4 million units cumulatively deployed (officially disclosed) — a multi-billion-dollar business; its strength is the AWS customer base, letting enterprises choose between NVIDIA GPUs and self-designed chips. Trainium 3 continues this path.
• Anthropic: In August 2026 announced the formation of an in-house chip team, with no tape-out or mass-production timeline yet; initially positioned as a complement (not a replacement) to existing partnerships with NVIDIA/AMD/AWS/Google Cloud, aiming to tailor-build for the Claude architecture and shed reliance on a single GPU.

---

6. NVIDIA's Answer: Not a Faster GPU, But Full-Stack​

It's easy to simplify the narrative to "four companies build chips, NVIDIA defends GPU." But NVIDIA took 6 slots at Hot Chips: a RISC-V tutorial, the Vera CPU, the Rubin GPU, the BlueField-4 DPU, the Spectrum-X multi-plane network, and an LPU accelerator.

A hyperscaler ASIC replaces only one of those five pillars. If the CPU, NIC, switching fabric, and software all come from the same vendor, what you save by swapping out the accelerator is far less than the accelerator line item on the bill suggests. Rubin's play is a full-stack AI factory platform spanning seven chips and five racks — the competitive answer is "full-stack positioning," not "a faster single chip."

---

7. Trend Judgment: Inference De-GPU-izes, Training Still GPU-Led​

• Inference side: The CUDA moat visibly shallows. Inference is parallelizable and replaceable at the endpoint; custom ASICs trade away the generality tax (implementing only the operations LLMs actually execute) for lower cost/token. Groq LPU, Cerebras, and various TPU/ASIC players all compete on the same metric.
• Training side: Foundation models are still trained on GPUs, with no serious challenger in the short term. NVIDIA's three training moats (fastest silicon + NVLink + CUDA) remain firm.
• Conclusion: Custom chips are not "replacing NVIDIA," but giving buyers a credible external negotiation option in the largest and fastest-growing battlefield — inference. That alone is enough to reshape the economics of AI infrastructure.

Related Links​

• Inference Accelerator Market 2026
• Hot Chips 2026 Full Recap
• Google TPU 8t Spec Page
• Google TPU 8i Spec Page

References​

• Hot Chips 2026: 4 Hyperscalers Put Custom AI Silicon on One Afternoon - Sean Kim
• Custom AI inference silicon: why OpenAI built Jalapeno - nexi.fund
• Anthropic Builds Custom AI Chips for Claude - Studio Global AI
• 所有大厂都在做同一件事：自己造芯片，少买英伟达的 - Toutiao

---

This article is compiled from August 2026 Hot Chips on-site reports, corporate announcements, and industry analysis. Some capacity and performance figures are brokerage estimates or vendor-disclosed figures; actual results are subject to mass-produced products.

On June 24, 2026, OpenAI and Broadcom jointly announced their first in-house AI inference chip, Jalapeño. This ASIC designed specifically for large language model inference went from design to tape-out in just 9 months and cuts inference cost by roughly 50%, marking OpenAI's transformation from a pure model company into a full-stack AI infrastructure provider.

---

1. Core conclusions (read this first)​

Dimension	Jalapeño	Current GPU solution	Advantage
Inference cost	-50%	Baseline	✅ Half the cost
Performance per watt	Clearly superior	Most advanced accelerator	✅ Energy-efficiency lead
Design cycle	9 months	~18 months	✅ 2× faster
Positioning	Inference ASIC	Train+inference GPU	Dedicated optimization
Supply	Internal only	Market purchase	⚠️ Not for sale

One-line summary: Jalapeño is a key step in OpenAI's full-stack AI strategy, using in-house silicon to cut inference cost 50% while opening a new paradigm of "AI-assisted design of AI chips."

---

2. What is Jalapeño?​

2.1 Basic information​

Item	Detail
Name	Jalapeño (a chili pepper)
Type	Application-specific integrated circuit (ASIC)
Positioning	Large language model inference
Announced	2026-06-24
Taped out	Sep 2025 (est., 9-month rapid tape-out)
Deployment	End of 2026 (gigawatt-scale data centers)
Partners	Broadcom, TSMC, Celestica
Process	TSMC 3nm
Architecture	Systolic Array
HBM	8 stacks (est. HBM3E or HBM4)

2.2 Why "Jalapeño"?​

Jalapeño is a Mexican chili known for "medium heat, strong flavor." OpenAI's naming hints that the chip:

• ✅ Medium heat: not the most aggressive architecture (vs Cerebras WSE), but effective enough
• ✅ Strong flavor: strong presence in inference scenarios (50% cost reduction)
• ✅ Appetizer: just "the first step of a multi-generation roadmap" (Broadcom CEO Hock Tan)

---

3. Deep technical analysis​

3.1 9-month rapid tape-out: the new paradigm of AI-assisted chip design​

Normally, designing an ASIC from scratch takes 1.5 to 2 years. Jalapeño went from initial design to manufacturing tape-out in just 9 months.

Key reason: deep software-hardware co-development

Technique	Description
AI-assisted architecture exploration	OpenAI used its own frontier models (GPT-5.3-Codex-Spark) to explore chip architecture design space
AI power simulation	AI models for power simulation and optimization
RL optimization	RL to optimize chip placement and routing
Broadcom silicon implementation	Broadcom provides top-tier ASIC implementation (network, switch chip experience)

OpenAI President Greg Brockman said:

"We use the frontier models that serve our users to optimize the infrastructure that runs the models of the future."

3.2 Architecture optimized for inference​

Unlike general-purpose GPUs, Jalapeño is an ASIC built from scratch around OpenAI's deep understanding of LLM inference workloads:

Architecture feature	Description
Reduce data movement	Core principle is minimizing data movement (the main bottleneck in inference)
Balanced compute-memory-network	Resource allocation optimized for inference, bringing real utilization closer to theoretical peak
High throughput + low latency	Aims to combine the throughput of leading accelerators with the low latency of the fastest dedicated inference systems
Future model support	Supports not only current models (GPT-5, GPT-5.3) but adapts to next-gen inference needs

3.3 Full-stack platform: more than a chip​

Jalapeño is a multi-generation compute platform, not just a chip:

Component	Supplier	Description
Accelerator chip	OpenAI design, TSMC fab	TSMC 3nm, 8-stack HBM
Network switch chip	Broadcom Tomahawk	High-speed interconnect (competes with NVIDIA NVLink)
Board, rack, system	Celestica	Full-rack solution
Software stack	OpenAI	Deep adaptation for GPT, Codex, Agent products

Deployment target: gigawatt-scale data centers

Broadcom CEO Hock Tan said:

"Jalapeño will begin deployment this year in gigawatt-scale data centers with Microsoft and other partners."

---

4. Performance and cost analysis​

4.1 Inference cost cut 50%​

Although OpenAI's official release was conservative on Jalapeño's cost savings — only stating its "performance per watt is substantially better than today's state of the art" without a specific percentage — per Bloomberg, Broadcom CEO Hock Tan revealed:

Early internal tests show Jalapeño achieves roughly 50% inference cost savings versus today's mainstream AI GPUs.

Significance for OpenAI:

Item	Current (GPU)	Jalapeño	Savings
Daily API calls	Hundreds of millions	Hundreds of millions	—
Inference cost share	~60-70% of operating cost	~30-35%	-50%
Annual compute spend	Billions of dollars	Hundreds of millions	Saves billions

4.2 Performance per watt clearly better than state of the art​

OpenAI's announcement states:

"Jalapeño engineering samples have successfully run complex reinforcement-learning tasks such as GPT-5.3-Codex-Spark at target frequency and power; early tests show performance per watt substantially better than today's most advanced AI accelerators."

Comparison target: NVIDIA Blackwell (today's most advanced AI accelerator)

Metric	Jalapeño	NVIDIA Blackwell	Note
Performance per watt	Clearly superior	Baseline	OpenAI official statement
Inference latency	On par with fastest dedicated inference systems	Baseline	Target
Throughput	On par with leading accelerators	Baseline	Target
TDP	Not disclosed (est. 400-700W)	700-1000W	Jalapeño possibly lower

---

5. Impact on the AI chip market​

5.1 "De-NVIDIA-ification" accelerates​

Jalapeño's launch is another footnote in big-tech's collective challenge to NVIDIA's market dominance:

Vendor	In-house chip	Type	Status	Relation to OpenAI
Google	TPU v6e / Ironwood	Train+inference	✅ Commercial	Google Cloud supplies OpenAI
Amazon	Trainium 3	Training	✅ Launched	AWS supplies OpenAI
Microsoft	Maia 100	Train+inference	✅ Launched	OpenAI exclusive partner
Meta	MTIA	Train+inference	✅ Launched	—
Apple	Neural Engine	On-device inference	✅ Commercial	—
OpenAI	Jalapeño	Inference	🚧 Deploy end of 2026	Internal + possibly sold to third parties

5.2 OpenAI is not about to fully "abandon" NVIDIA​

Brockman admitted:

"We simply cannot get compute fast enough."

Currently OpenAI is simultaneously procuring chips from NVIDIA, AWS, AMD, and Cerebras; Jalapeño is a structural supplement to its explosive compute demand, not a replacement.

5.3 Possibly sold to third parties​

Broadcom CEO Hock Tan specifically emphasized:

"This is just 'the start of a multi-generation roadmap'; OpenAI and Broadcom aim to jointly build gigawatt-scale compute clusters."

This means OpenAI may sell its hardware to third parties, provided it can secure enough supply from Broadcom and TSMC.

---

6. Jalapeño vs other in-house chips​

Metric	Jalapeño (OpenAI)	TPU v6e (Google)	Trainium 3 (Amazon)	Maia 100 (Microsoft)
Announced	2026-06-24	2024	Q4 2025	2023
Type	Inference ASIC	Train+inference TPU	Training ASIC	Train+inference
Process	TSMC 3nm	TSMC 4nm	TSMC 5nm (est.)	TSMC 5nm (est.)
For sale	❌ Internal only (maybe later)	✅ GCP	✅ AWS	❌ Internal only
Design cycle	9 months	~18 months	~18 months	~18 months
AI-assisted design	✅ First	❌ No	❌ No	❌ No
Cost advantage	-50% inference cost	Optimized	Optimized	Optimized

Key differences:

• ✅ Jalapeño is the first AI chip designed with AI assistance
• ✅ Jalapeño design cycle only 9 months (industry average 18 months)
• ⚠️ Jalapeño not for sale (at least for now)

---

7. Future roadmap​

7.1 Multi-generation chip platform​

Jalapeño is just "the start of a multi-generation roadmap":

Time	Event
End of 2026	Jalapeño initial deployment (gigawatt-scale data centers)
2027	Jalapeño v2 (est., architecture optimization)
2027-2028	Jalapeño training version (est., challenging TPU/Trainium)
2028 and beyond	Gigawatt-scale compute cluster fully built

7.2 OpenAI full-stack AI infrastructure strategy​

Layer	OpenAI in-house	Outsourced/procured
Models	✅ GPT-5, GPT-5.3, Codex	—
Chips	✅ Jalapeño (inference)	NVIDIA GPU, AWS Trainium, AMD GPU
Systems	✅ With Celestica	Microsoft Azure data centers
Network	✅ Broadcom Tomahawk	Microsoft Azure network
Cloud platform	❌ None	Microsoft Azure (exclusive partner)

---

8. Industry reaction and expert views​

8.1 Supportive views​

Expert	View
Broadcom CEO Hock Tan	"Jalapeño is just the start of a multi-generation roadmap; the goal is to jointly build gigawatt-scale compute clusters."
OpenAI President Greg Brockman	"We use the frontier models that serve our users to optimize the infrastructure that runs the models of the future."
Industry insiders	"Jalapeño's launch is another footnote in big-tech's collective challenge to NVIDIA's market dominance."

8.2 Skeptical views​

Concern	Description
Not for sale	Currently internal only; third parties cannot purchase, limited impact on NVIDIA's market share
Software ecosystem	OpenAI must build its own software stack; competing with CUDA is hard
Supply capacity	TSMC capacity is limited; can it meet OpenAI + Broadcom + other customers' demand?
Opaque performance data	OpenAI has not released specs (compute, memory, bandwidth, TDP), hard to assess objectively

---

9. Significance for developers​

9.1 If OpenAI sells Jalapeño to developers in the future...​

Scenario	Current (NVIDIA GPU)	Future (Jalapeño)
Inference cost	Baseline	-50%
Inference latency	Baseline	Possibly lower
Software stack	CUDA + TensorRT	OpenAI API (possibly open-source stack)
Procurement difficulty	High (export controls, supply shortage)	Low (OpenAI direct supply)

9.2 Worth watching even if not sold​

• ✅ 50% inference cost cut forces NVIDIA, AMD, Intel to lower GPU prices
• ✅ The AI-assisted chip design paradigm will be rapidly copied by the industry
• ✅ The 9-month tape-out cycle becomes a new industry benchmark

---

10. Summary​

Dimension	Assessment
Technology innovation	⭐⭐⭐⭐⭐ First AI chip designed with AI assistance, 9-month tape-out
Cost advantage	⭐⭐⭐⭐⭐ 50% inference cost cut, billions saved annually
Strategic significance	⭐⭐⭐⭐⭐ OpenAI transforms from pure model company to full-stack AI infrastructure provider
Market impact	⭐⭐⭐⭐ "De-NVIDIA-ification" accelerates, big-tech in-house chip camp grows
Openness	⭐⭐ Internal only for now, possibly sold to third parties later

Final recommendations:

• 🇨🇳 China market: Keep watching Huawei Ascend, Cambricon MLU, Moore Threads MTT (Jalapeño not sold to China)
• 🌍 International market: Watch whether Jalapeño is eventually sold externally and its impact on NVIDIA's market share
• 💡 Developers: Watch for possible OpenAI API price cuts (50% inference cost cut may partially pass through)

---

References​

• OpenAI official announcement (technical white paper pending)
• Broadcom CEO Hock Tan interview
• Bloomberg: Jalapeño inference cost cut 50%
• SemiWiki: Jalapeño technical details
• Tom's Hardware: Jalapeño architecture analysis
• ESMChina: Jalapeño 9-month tape-out

---

Disclaimer: Some specs in this article are estimates, subject to OpenAI's official technical white paper. OpenAI will release a detailed performance white paper in the coming months.

Last updated: June 26, 2026