Nvidia's dominance in the processor market for artificial intelligence is beginning to waver. The growing pressure from competitors like AMD and Intel, coupled with the development of proprietary chips by giants such as Microsoft and Google, has generated a wave of uncertainty among investors. The company no longer has a clear path, and its once unquestionable leadership now faces an uncertain future where it must demonstrate its ability to adapt.
The Blackwell architecture and Nvidia's technological response 🚀
Faced with this scenario, Nvidia is preparing its next architecture, Blackwell, designed to compete in efficiency and performance. These new chips are expected to integrate next-generation HBM4 memory and a more advanced manufacturing process, seeking to maintain an advantage in model training. However, the real test will be whether they can offer a lower cost per inference than solutions from its rivals, such as AMD's APUs or Google's TPUs, which are already gaining traction in data centers.
When your customers become your worst enemies 😅
Nvidia must be feeling that classic uncomfortable heat you get when your best customers, those who buy by the pallet, decide that making their own chips is more fun than continuing to pay your premium prices. It's like your best friend deciding to open a pizzeria right next to yours, but with sourdough and lower prices. The difference is that here the pizza costs millions and the oven is called artificial intelligence.