NVIDIA Corporation NVDA
Regular-session close
After hours · Nov 16, 6:10 PM ET
- YTD
- -16.28%
- 1 year
- -17.89%
Discussion (22926)
$NVDA If you have this stock last summer until now, then you nearly earn nothing about this stock without any action.
Trump taking the country backwards, with his B.S tariffs we got the lowest GDP 1.4% and India get 7.5 %. All lies he spews
$NVDA hard to keep $180 today.
The primary reasons for this strategic shift include:
Reducing NVIDIA Reliance: Meta is one of NVIDIA’s largest customers, but it is moving toward a "multi-vendor strategy" to decrease dependence on a single supplier and gain more negotiating leverage.
Cost and Energy Efficiency: TPUs are specialized ASICs (Application-Specific Integrated Circuits) designed specifically for AI workloads. They are reported to deliver lower costs for large-scale training and consume significantly less power than NVIDIA’s high-end GPUs.
Diversifying Workloads: By using different types of silicon, Meta can pair specific AI tasks (like training vs. inference) with the most optimal hardware. While NVIDIA remains the leader for general AI training, TPUs are highly efficient for certain matrix-based neural network tasks.
Software Compatibility (TorchTPU): A major hurdle for using TPUs was their optimization for the JAX framework rather than PyTorch (which Meta created). Google is now working on a "TorchTPU" project to make its chips fully compatible with PyTorch, making it easier for Meta to switch its existing models to Google’s hardware.
Massive Compute Demand: The scale of Meta's "Artificial Superintelligence" (ASI) ambitions requires more compute than any single vendor can currently provide. Meta is also buying $60 billion worth of chips from AMD and continuing its massive spend with NVIDIA simultaneously.
Reddit
Reddit
+8
What are other companies developing AI chips beyond NVIDIA and Google?
What are the technical differences between NVIDIA GPUs and Google TPUs for AI?
Give me some real-world examples of how Torch TPU makes PyTorch models run better on Google's TPUs
Meta might start using Google AI chips - RCRTech
RCRTech
No longer solely reliant on NVIDIA, Meta has spent billions of dollars ...
Moomoo
Meta is considering Google TPUs for their data centers worth billions. - Reddit
Show all
Meta renting AI chips or TPU from google! Diversify AI for LLM and Inferencing
$NVDA..7 months sideways is irrational&ridiculous
$NVDA..P/E 25 totally undervalued after monster er
haha
Have we all been conned again. Or are we just stupid. This isn’t a joke NVDA by all metrics should be at this date $320 !!!!
Shorts: Cover now ! Tomorrow it will rock again and show the valuation calculation
