Despite more cloud providers developing custom silicon, none seem ready to replace Nvidia, AMD, or Intel hardware in the near future. GPUs remain the preferred choice for generative AI, with companies like Meta and Microsoft continuing to deploy large quantities of Nvidia GPUs. AWS also announced a large deployment of Nvidia's super chips alongside its own CPUs and AI accelerators.
Key takeaways:
- Nvidia is reportedly creating a business unit to sell its intellectual property and design services to companies like AWS, Microsoft, and Meta, in response to these companies developing their own alternatives to Nvidia's GPUs.
- These companies often rely on intellectual property from other companies, such as Marvell or Broadcom, for their custom chips. Nvidia sees an opportunity to offer its own technologies in a similar manner.
- Nvidia has reportedly approached Amazon, Meta, Microsoft, Google, and OpenAI about the possibility of developing custom chips based on its designs. It has also approached telecom, automotive, and video game customers with similar offers.
- Despite the trend towards custom silicon, GPUs, such as those from Nvidia, AMD, and Intel, remain the preferred choice for generative AI. Companies like Meta and Microsoft continue to deploy large quantities of GPUs alongside their custom silicon.