Battle of computing power heats up: Google open sources strategy + Meta ecosystem turns against each other, aiming to break Nvidia's CUDA ecosystem monopoly.

According to reports, Alphabet Inc. Class C (GOOGL.US) is trying to weaken the advantage established by NVIDIA Corporation (NVDA.US) through the CUDA software platform, and has received some support from Meta (META.US). Sources familiar with the matter stated that the search giant is working on making its own AI chip TPU run smoother with the artificial intelligence framework PyTorch. It is worth noting that since PyTorch was officially released in 2016, the framework has had a deep connection with the NVIDIA Corporation CUDA platform, with Meta being a key creator of this open-source ecosystem. The sources revealed that Alphabet Inc. Class C is also considering whether to open source some of the code to increase customer adoption rates. The company internally named this project TorchTPU and is investing more resources and effort into it. Alphabet Inc. Class C is collaborating with Meta to enable PyTorch to run natively and seamlessly on Alphabet Inc. Class C TPU at a large scale. This means that developers no longer need to rewrite their code to use Alphabet Inc. Class C chips, significantly reducing the barriers to migration from NVIDIA Corporation to Alphabet Inc. Class C. It is worth mentioning that Meta, as one of NVIDIA Corporation's largest customers, is actively seeking alternative solutions to address the high prices and supply bottlenecks of NVIDIA Corporation chips. Last month, reports stated that Meta was negotiating a multi-billion dollar deal with Alphabet Inc. Class C, planning to lease Alphabet Inc. Class C cloud TPU in 2026 and possibly directly purchase their chips for their own data centers in 2027 to reduce dependence on NVIDIA Corporation and its GPUs. From a technical cooperation perspective, Meta is responsible for providing the application ecosystem (PyTorch), while Alphabet Inc. Class C provides the underlying computational power hardware (TPU), working together to build a fast lane around CUDA at the software level. In contrast, NVIDIA Corporation's leading position in the AI field is difficult to shake not only because of its strong GPUs, but also because CUDA has become the "standard language" for AI development. Both Meta and NVIDIA Corporation have not immediately responded to this. A spokesperson for Alphabet Inc. Class C confirmed the plan and stated that the relevant information was publicly announced in October. "Alphabet Inc. Class C Cloud is committed to providing customers with a full range of choices from models, accelerators, frameworks to tools," the spokesperson said via email. "PyTorch is very popular, and our goal is to seamlessly integrate its experience on Alphabet Inc. Class C Cloud TPU. We see a significant and accelerating demand for both TPU and GPU infrastructures. Our focus is to provide developers with the flexibility and scalability they need, regardless of the hardware they choose." As of Wednesday's closing, both Alphabet Inc. Class C and NVIDIA Corporation's stock prices have dropped by over 3%, while Meta's stock price has also dropped by over 1%.