Google chip manufacturing takes aim at Nvidia's generative A.I.
Google's AI Chip
In this video, Deirdre Bosa discusses Google's in-house chip for artificial intelligence (AI) and how it compares to other chips in the market.
Google's AI Chip
- Google has developed an in-house chip called the Processing Unit (TPU), which is faster and more efficient than comparable systems from NVIDIA.
- The TPU is being used for 90% of Google's work on AI training, and can be strung together to create better generative AI.
- More big tech companies are designing their own chips, such as Apple with its M1 series that displaced Intel, and Amazon's custom computing chips that lower costs for customers.
- This is likely the start of the AI battle at the chip or compute level that will drive consumer-facing products.
Manufacturing of Google's AI Chip
- Google designs the chip with the help of ARM, a chip company owned by Softbank that is expected to IPO soon.
- The chips are typically made by one of the big manufacturers like TSMC.
Implications for NVIDIA
- NVIDIA is seen as the leader at the compute level for artificial intelligence, but there could be less reliance on them in the longer term if more companies design their own chips.
- The battle in AI is taking place at both the top stock for consumers and at the bottom of the chip level.