⚡ Why OpenAI bet $20B+ on Nvidia's weakness.

Cerebras chips are wafer-scale — one processor the size of a dinner plate. The trade-off is speed for memory: blistering throughput, but the architecture is memory-bound on long context. For inference it runs roughly 15–75× faster than Nvidia GPUs depending on the model — not training, which is still Nvidia's game. That's why OpenAI committed $20B+. The IPO popped 68% to ~$95B, then cooled. A specialist today, priced as the future of inference.

Full thesis → convictioninvest.com

Sources: Cerebras · Artificial Analysis · CNBC
#Cerebras #AI #Nvidia #Inference #OpenAI #CBRS #Semiconductors #Shorts

Видео ⚡ Why OpenAI bet $20B+ on Nvidia's weakness. канала Alfred | Conviction Invest