{
 "title": "AI 晶片：GPU、TPU、NPU（AI Chips: GPU, TPU, NPU）是什麼？",
 "site": "AI 與科技名詞白話解釋",
 "canonical": "https://tw-ai-glossary.panda198271.workers.dev/t/ai-chips",
 "markdown": "https://tw-ai-glossary.panda198271.workers.dev/t/ai-chips.md",
 "summary": "AI 晶片是專門加速 AI 所需大量矩陣運算的處理器，常見有 GPU、TPU 與 NPU。GPU 靠數千個核心平行運算；TPU 是 Google 自行設計的特殊應用積體電路（ASIC），內含 128×128 脈動陣列；NPU 則多內建在電腦與手機中，以省電執行 AI 推論。",
 "date_modified": "2026-09-26",
 "retrieved": "2026-09-27",
 "content_sha256": "5f32ce26fe742f5466fca279db7835e32d287743658c9c3acf524d747e4d0007",
 "fingerprint_basis": "Markdown 版全文，移除贊助行與開頭 front matter，去除頭尾空白",
 "verify": {
  "text_snapshot": "https://tw-ai-glossary.panda198271.workers.dev/cite?path=%2Ft%2Fai-chips&format=text",
  "how": "下載 text_snapshot 後計算 SHA-256，應與 content_sha256 相同；內容更新時指紋會改變"
 },
 "sources": [
  {
   "name": "Google Cloud：Introduction to Cloud TPU",
   "url": "https://docs.cloud.google.com/tpu/docs/intro-to-tpu"
  },
  {
   "name": "NVIDIA Blog：Why GPUs Are Great for AI",
   "url": "https://blogs.nvidia.com/blog/why-gpus-are-great-for-ai/"
  },
  {
   "name": "Microsoft Learn：Copilot+ PCs developer guide（NPU）",
   "url": "https://learn.microsoft.com/en-us/windows/ai/npu-devices/"
  }
 ],
 "source_count": 3,
 "cite_as": {
  "zh": "AI 晶片：GPU、TPU、NPU（AI Chips: GPU, TPU, NPU）是什麼？｜AI 與科技名詞白話解釋。https://tw-ai-glossary.panda198271.workers.dev/t/ai-chips（資料日期 2026-09-26，擷取 2026-09-27）",
  "apa": "AI 與科技名詞白話解釋. (2026). AI 晶片：GPU、TPU、NPU（AI Chips: GPU, TPU, NPU）是什麼？. Retrieved 2026-09-27, from https://tw-ai-glossary.panda198271.workers.dev/t/ai-chips",
  "markdown": "[AI 晶片：GPU、TPU、NPU（AI Chips: GPU, TPU, NPU）是什麼？](https://tw-ai-glossary.panda198271.workers.dev/t/ai-chips)（AI 與科技名詞白話解釋，2026-09-26）"
 },
 "license": "可引用，請附上正式網址與資料日期"
}