Nvidia, the leading AI chip provider, is facing potential long-term competition from its largest customers, including Google, Amazon, Microsoft, Meta, and OpenAI, as they develop their own AI chips. However, Nvidia’s CEO, Jensen Huang, has a confident response: ‘We can help you do that.’ While this blog contains speculative information that has not been verified, it is interesting to consider the possibilities.
A Successful Model for Confidentiality
During his time at AMD, the author recalls how the company managed to keep two teams working in isolation to protect client confidentiality. One team designed chips for the Microsoft Xbox, while the other focused on the Sony Playstation. This model allowed AMD to protect each client’s gaming console intellectual property and requirements, leading to their continued success in the market.
The Challenge of Secrecy
While maintaining secrecy can be difficult and expensive, it is also challenging to scale such a business model. Nvidia is now exploring an alternative approach, where the customer takes on more of the design work and contributes their intellectual property to the development of their own chips. By leveraging Nvidia’s relationship with fabrication partners like TSMC and Samsung, clients can lower costs and improve time to market.
Nvidia’s New Business Model
Nvidia has recently announced the formation of a group dedicated to forging this new business model. The company aims to assist clients in building their own solutions using Nvidia intellectual property or chiplets. This strategic move has further boosted Nvidia’s stock, with a 3% increase upon the announcement.
Building an AI Licensing Giant
With this direction, Nvidia may no longer require the acquisition of Arm. Instead, the company is laying the foundation to become an AI licensing giant. Many companies already partner with different solution providers for specific components or intellectual property to integrate into their chips, and Nvidia aims to offer similar services in the AI space. This approach allows companies to lower costs and tailor their solutions while leveraging Nvidia’s expertise.
Potential Partnerships and Differentiation
Companies like Google, Amazon AWS, Meta, and Microsoft Azure have already developed custom chips for in-house AI applications alongside their use of Nvidia GPUs for cloud customers. The question arises: could these clients partner with Nvidia for future designs? Additionally, could they utilize Nvidia’s in-house and AWS supercomputers to accelerate and optimize their design efforts? This collaboration could generate additional revenue for Nvidia and provide these clients with a significant competitive advantage. Nvidia’s decision to host their latest supercomputer, Project Cieba, at AWS data centers suggests a possible avenue for chip design optimization services.
Transforming the Industry once again
While these speculations may push the boundaries, they indicate that Nvidia is aware of the changing dynamics in the industry and is prepared to transform it once again. All technology commoditizes over time, including previous generations of silicon. Nvidia’s potential entrance into licensing deals opens up opportunities for monetizing intellectual property and expertise.
Although this blog presents speculative information, it highlights Nvidia’s response to potential competition from its largest customers. By exploring a new business model that involves clients building their own solutions using Nvidia intellectual property, the company is positioning itself as an AI licensing giant. Additionally, partnerships with major technology companies and the utilization of powerful supercomputers could further accelerate Nvidia’s growth and differentiation in the market.