BREAKING
OpenAI Unveils Jalapeno AI Chip
Who Built Jalapeno
Design & Manufacturing
●
Design led by OpenAI
●
Manufacturing by Broadcom
●
Inference-focused ASIC
Integration
●
Celestica handles rack integration
●
Networking and assembly
●
Samples already in lab testing
0
mo
design to tape-out
0
%
cheaper than typical GPU
0
GW
accelerators targeted
Rollout Through Early 2028
1
First deployment late 2026
↓
2
Ramp-up in 2027
↓
3
Full operation by early 2028
Custom Silicon vs NVIDIA
The Case For
●
Cuts GPU waste via inference focus
●
Rivals Blackwell and TPU
●
Nine-month build pace
The Caution
●
No independent benchmarks yet
●
Specs like core counts undisclosed
●
NVIDIA's CUDA ecosystem
Broadcom Shares Rose 3%
AI NEWS BLITZ
OpenAI has revealed Jalapeno, its first in-house AI chip built with Broadcom.