Skip to main content

2 posts tagged with "MI450"

AMD Instinct MI450 加速器

View all tags

AMD Helios 正式出货:MI455X 机架 525 万美元一台,OpenAI / Meta / 甲骨文 / Anthropic 12GW 订单进入兑现期

· 6 min read
Industry Research Team

7 月 Advancing AI 2026 大会上的承诺正在兑现:AMD Helios 机架已进入全面量产,2026 年 Q3 末启动客户出货、Q4 放量。管理层透露客户拉货进度"领先于初始预测"——对英伟达而言,第一次有一个对手在"机架级系统"这个自己定义的战场上,拿到了真金白银的订单。


1. Helios:AMD 第一台真正意义上的"机架级计算机"

指标Helios(72 × MI455X)对比 Vera Rubin NVL72
显存31TB HBM4 池化(单卡 432GB)Rubin 单卡 288GB(+50%
显存带宽1.4PB/s(单卡 19.6TB/s)22TB/s(单卡)
FP4 推理2.9 ExaFLOPS3.6 EFLOPS
FP8 训练1.4 ExaFLOPS1.2 EFLOPS
Scale-up 互联260TB/s(UALink/UALoE)130TB/s(NVLink6,约 2 倍差距)
机架售价约 525 万美元
出货节奏2026 Q3 末起,Q4 放量2026 年 9 月批量出货

MI455X 采用 CDNA 5 架构:8 个 TSMC N2 制程加速 die + N3P 互联 die,256 个工作组处理器、192MB 全局 L2,12 层堆叠 HBM4 让它成为首款显存超过 400GB 的 AMD GPU。同系列还有面向主权 AI 与 HPC 的 MI430X(推理特化)、面向企业的 MI440X(8 GPU + 1 EPYC Venice 一体化服务器)——三条产品线覆盖从云到边缘的完整光谱,规格详见本站 MI455XMI440XMI430X 卡片。

配套的 EPYC Venice(Zen 6) 是业界首款 2nm x86 服务器 CPU,最高 256 核,性能较上代 Turin 提升约 1.7 倍,承担 Helios 机架内编排、数据搬运与 CPU 密集型负载。

2. 12GW 订单簿:四大锚定客户全部落地

AMD 在 AI 加速器市场最大的变化,是从"卖芯片"变成"签产能":

客户协议规模关键条款交付节奏
OpenAI6GW 跨代际首批 1GW MI450;附最高 1.6 亿股认股权证(与部署进度及股价挂钩)2026 H2 起部署,2027 加速
Meta最高 6GW / 五年最高 600 亿美元;Meta 获收购 AMD 最多 10% 股份的选择权2026 H2 部署 Helios
甲骨文5 万颗 MI450从 2026 Q3 起部署2026 Q3 起
Anthropic最高 2GWAMD 投资最高 50 亿美元;双方用 Claude 优化 ROCm首批 1GW 2027 H1 上线

分析师对 OpenAI 协议中 GPU 部分的收入估计约 800 亿美元;Piper Sandler 估计 Meta 协议五年内可为 AMD 带来约 1000 亿美元收入。合作模式也已升级——不再是简单的采购合同,而是"定制开发 + 资本投入 + 长期联合优化"的深度绑定:Anthropic 甚至披露 Claude 曾在一个周末内自主完成 AMD AI 服务器的配置

值得关注的是,Cerebras 也宣布与 AMD 达成合作,从 Cerebras 数据中心内部开始部署 AMD 服务器——推理芯片初创公司与 GPU 巨头的"竞合"正在成为新常态。

3. 最大的变量不是英伟达,是封装产能

AMD 2027 年数据中心营收翻倍的目标,瓶颈不在需求,在供给侧:

  • CoWoS-L 产能:汇丰分析指出,要达成目标,CoWoS-L 产能需求需比基础情景高出 12%–35%;AMD 选择将设备托管至 SPIL、PTI 等二线封装厂(一线供应商 2027 年前几乎没有产能富余),低良率风险不可忽视;
  • HBM 供应:MI455X 单卡 432GB HBM4 对存储供应提出极高要求,2027 年 HBM4 供给依然紧张(详见本站 HBM4 竞速分析);
  • 资本开支:AMD Q2 资本支出已跃升至 8.08 亿美元,专门用于向外包封测厂托管 CoWoS 设备以保障 Venice CPU 量产。

摩根大通提醒:这是 AMD 首次机架级产品量产,执行风险在 2027 财年内仍然显著,毛利率稀释效应将随出货量逐步显现。

4. ROCm 与 CUDA:真正的前线

AMD 高管提出"随着 PyTorch、vLLM 等高层抽象框架普及,底层 CUDA 或 ROCm 的差异对开发者日益模糊"。这个判断有部分道理——但 CUDA 的护城河不止是编程接口,还包括数学库、通信库、调试工具和二十年积累的硬件级优化能力。处理自定义算子、混合精度、稀疏计算和千卡集群故障排查时,底层软件栈依然决定成败。

AMD 的聪明之处在于借力打力:让 Anthropic 用 Claude 优化 ROCm——用 AI 优化 AI 芯片的软件栈,这可能是缩小生态差距最快的一条路。

5. 对采购方的启示

  • 第二供应源已是现实选项:2027 年规划推理集群时,MI455X/Helios 值得进入正式评估名单,其 432GB 显存对大参数模型推理的容量优势是实打实的;
  • 锁定交付窗口:Helios 为 OCP 开放机架参考设计,最终交付由 OEM/ODM 完成——下单时确认液冷、电源与系统集成责任方,避免"芯片到了机柜没到";
  • 关注 Q4 爬坡数据:年底前 Helios 是否真在 OpenAI 基础设施中大规模上线,是检验 AMD 机架级交付能力的第一个硬指标。

相关链接

参考资料


本文基于 AMD Advancing AI 2026 大会披露与后续供应链报道整理。订单规模、收入预测均为分析师口径,实际以 AMD 财报为准。

AMD Advancing AI 2026 Opens Tomorrow: Three CDNA5 MI400 Models, Helios Rack Hits 3 exaFLOPS, OpenAI + Meta Lock 12GW Deal

· 4 min read
AI Hardware Analyst

AMD has confirmed its flagship AI event Advancing AI 2026 will be held July 22-23, 2026 at the Moscone Center in San Francisco, with the keynote on July 23 hosted by Chair and CEO Lisa Su. The event will complete the Instinct MI400 series availability timeline, pricing, and independent benchmark data.

1. Instinct MI400 family: three CDNA 5 accelerators

AMD fully revealed the MI400 matrix at CES 2026; all three accelerators use CDNA 5 architecture, TSMC 2nm process, differentiated by precision and scenario:

ModelPositioningKey specs
MI455X (flagship)Large-scale train/inference (rack-scale)320B transistors, 12 chiplets, 432 GB HBM4 (12×36GB), 19.6 TB/s, FP4 40 PFLOPS / FP8 20 PFLOPS
MI440X (enterprise)Local enterprise AI (8-card node)Low-precision AI (FP4/FP8/BF16), direct MI300/MI350 replacement, compatible with existing power/cooling
MI430X (HPC/sovereign AI)High-precision scientific computing + AIFull FP32/FP64, already deployed at Oak Ridge Discovery and France's first exascale Alice Recoque

MI455X and MI440X target low-precision AI (FP4/FP8/BF16); MI430X fills traditional HPC high-precision needs — improving energy efficiency and cost-performance by "trimming execution units by precision." All three support UALink (among the first accelerators compatible with the standard) and Infinity Fabric die-to-die interconnect; rack scaling uses Ultra Ethernet.

Lisa Su confirmed on the Q1 2026 earnings call: MI455X samples have been sent to core customers, with demand "exceeding the company's internal expectations for 2027."

2. Helios rack: 3 exaFLOPS per cabinet

AMD enters the hyperscale market with the Helios rack-scale platform:

MetricHelios rack
Accelerators72 × MI455X
Aggregate HBM431 TB
Total memory bandwidth1.4 PB/s
Per-cabinet computeUp to 3 AI exaFLOPS (Q3 delivery target)
Target customersHyperscale train/inference clusters

Helios uses AMD's in-house Zen 6 EPYC Venice CPU (18 per rack) + Pensando Vulcano 800G NIC, integrated via the open ROCm software stack; AMD also plans a double-width 128-card Helios variant, pushing per-cabinet compute to the 3 AI exaFLOPS ceiling. Further out, the MI500 series (CDNA 6, 2nm, HBM4E) is planned for 2027, with official claims of up to 1000× AI performance vs MI300X.

3. 12GW deal: OpenAI + Meta dual endorsement

AMD holds two historic-scale compute agreements totaling about 12 GW, with lifetime potential revenue possibly reaching $100B:

CustomerScaleFirst deploymentStructure
OpenAI6 GW (multi-gen products)First 1 GW, H2 2026 on MI450"compute-for-upside": up to 160M warrants, vesting by milestone and stock-price targets
Meta6 GWCustom MI450 chips, from H2 2026Deployed in next-gen data centers

Financial expectations

Metric2026 forecast
MI400 series revenue~$7.2B (about 25% of data-center sales)
Data-center GPU revenue~$15B (up +114% YoY)
Total data-center revenuePossibly $28.7B (up +73% YoY)

⚠️ Execution risk: AMD has flagged that MI450's Q3 mass production will weigh on gross margin (new products below company average); advanced process and advanced packaging (TSMC CoWoS) capacity remain the main constraint.

Industry interpretation

  1. CUDA moat being pried open: when companies building the world's largest training clusters — Meta, OpenAI — bet on AMD silicon, AMD's long-standing 5-7% GPU share ceiling is being broken.
  2. Memory advantage as differentiation: 432 GB HBM4 / 19.6 TB/s vs NVIDIA Rubin's 288 GB offers capacity advantage, critical for large-model inference (KV Cache-constrained scenarios).
  3. Tight benchmarking pace: MI450 and NVIDIA Vera Rubin both ramp in H2 2026, with the two giants competing head-on over HBM4 supply and CoWoS capacity.

References


This article was written on the eve of Advancing AI 2026 (July 22-23, opening tomorrow); the keynote is July 23 hosted by Lisa Su, where MI400's final availability, pricing, and independent benchmarks will be revealed — we will update in sync.