Unverified50% confidenceFactExact time
NVIDIA's B200 is based on the Blackwell architecture, manufactured using TSMC's 4NP process, consists of two dies coupled via a 10 TB/s chip-to-chip interconnect, and contains a total of 208 billion transistors.
1
Sources
50%
Confidence
Long-term
Relevance
8/8/2026
First Seen
Sources
软件优化让单张B200媲美专用推理芯片
Related Claims
UnverifiedBlackwell (B100/B200 series) was released in 2024, manufactured using TSMC's 4NP process, with a single GPU integrating 208 billion transistors73% similarUnverifiedBlackwell is NVIDIA's latest-generation data center architecture (B100/B200 series), featuring a dual-chip package design (MCM), and a single GPU can deliver up to 20 petaFLOPS of FP4 compute performance.65% similarUnverifiedThe NVIDIA GB300 GPU is the flagship model of the Blackwell architecture, integrating dual B200 GPU chips with a Grace CPU, delivering approximately 4x the FP8 training throughput compared to the H100.64% similarUnverifiedBlackwell单颗GPU晶体管数量是前代Hopper架构H100的约两倍63% similarUnverifiedBlackwell采用双芯片互联设计，两颗GPU裸片通过10TB/s的芯片间互联（NV-HBI）连接，在操作系统层面呈现为单一逻辑GPU63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/707601API
curl https://kongchang.com/api/v1/knowledge/claims/707601MCP
get_claim(id=707601)