2025-11-28_tpuv7-google-takes-a-swing-at-the::TPU7-0005
① SA Source
- Source: 開啟完整 SA 文章
- Section:
Google AI Infrastructure Supremacy: Systems Matter More Than Microarchitecture - Line hint:
93
Context Before
Source: SemiAnalysis TCO Model ↗, Anthropic and OpenAI
Evidence
Context After
Our Accelerator Industry Model ↗, Datacenter Industry Model ↗ and Core Research ↗ subscribers saw the industry implications well before this was announced and became market consensus. In early August, we shared with our Accelerator Model clients ↗ that we saw massive upward revisions for Broadcom / Google TPU orders in the supply chain for 2026. We also revealed that the reason for these order increases was the fact that Google would begin selling systems externally to multiple customers ↗. In early September, we revealed that one of the big external customers will be Anthropic, ↗ with demand of at least 1 million TPUs. This was officially confirmed by Anthropic and Google in October ↗. We also called out Meta as a big TPU customer ↗ on the 7th of November, weeks before others. In addition we have discussed other customers as well.
As a result, our institutional clients have had plenty of heads up on one of the largest performance dispersions in the AI Trade to date. SemiAnalysis was the first to break all these insights because no other research firm can connect the dots from the fabs to supply chain through the datacenters to the labs. To get access to these insights as and stay ahead of the curve: sales@semianalysis.com
② Atomic Claim
SemiAnalysis 估計 OpenAI 尚未部署 TPU 前，就因競爭壓力讓 Nvidia fleet 成本約下降 30%。
- Epistemic Mode:
ESTIMATED - Mapping Status:
COMPLETE
③ Semantic Frame
{
"attribute": "nvidia_fleet_cost_saving_from_tpu_competition",
"context_nodes": [
{
"id": "04_knowledge_base/TPU",
"label": "TPU"
},
{
"id": "02_companies/NVDA",
"label": "Nvidia"
}
],
"entity": {
"id": "02_companies/OpenAI",
"label": "OpenAI"
},
"frame_type": "ATTRIBUTE",
"qualifiers": {
"condition_text": null,
"numeric_mentions": [
"30%"
],
"temporal_mentions": []
},
"value": {
"numeric_mentions": [
"30%"
],
"value_text": "~30%"
}
}④ Canonical Entity Mapping
⑤ Human Review
請在 Properties 逐項確認：
- 原文 → Atomic Claim 是否忠實
- Atomic Claim → Semantic Frame 是否忠實
- Canonical Entity mapping 是否正確
- Epistemic mode 是否保留原文語氣
- 最後選擇
review_action
Review state
Markdown 內文不是正式 approval。只有 Apply bridge 寫入的 Decision Ledger event 才是正式決策。