OpenAI Looks Beyond Nvidia for AI Computing Power

OpenAI is reportedly preparing to purchase artificial intelligence chips from Google, marking another step in its effort to diversify the hardware used to power its AI systems. The reported move could reduce its dependence on Nvidia as demand for computing power continues to grow.

The development comes alongside another notable change in the semiconductor industry. Intel CEO Lip-Bu Tan has hinted at possible cooperation with AMD, one of Intel’s long-standing competitors.

Together, these developments highlight how major technology companies are exploring new partnerships and hardware options to meet the growing demands of artificial intelligence.

OpenAI Plans to Purchase Google’s AI Chips

According to reports, OpenAI plans to purchase Google’s Tensor Processing Units (TPUs). These specialised chips are designed to handle AI workloads, including model training and inference.

The reported purchase would give OpenAI access to another major AI chip platform beyond Nvidia’s graphics processing units (GPUs). However, the proposed deal’s financial terms, chip quantities and delivery schedule have not been disclosed. Neither company has publicly confirmed the reported purchase.

OpenAI is already a major Nvidia customer. Its GPUs support the development and operation of AI models behind products such as ChatGPT.

However, the company has been exploring different ways to secure the computing resources required for its growing services. It has also reportedly used Google’s TPUs through cloud infrastructure in the past.

Google’s chips are becoming an increasingly important part of the AI hardware market. The company originally developed TPUs for its own services but has expanded access to external customers.

Companies such as Anthropic have also used Google’s chips to support their AI development work.

Why Google’s TPUs Matter

Nvidia has built a strong position in the AI chip market through its powerful GPUs and software ecosystem. Its CUDA platform is widely used by developers and organisations working on AI applications.

Google’s TPUs offer another approach. They are specialised processors designed for machine learning workloads and are used to train and run Google’s own AI models, including Gemini.

For OpenAI, adding Google chips could provide several potential benefits:

• More computing capacity: Access to another chip supplier could help support growing AI workloads.
• Supplier diversification: Using different hardware platforms can reduce dependence on a single supplier.
• Infrastructure flexibility: Different chips may be better suited to specific training or inference tasks.
• Cost management: Alternative hardware could create more options when evaluating computing costs.

However, moving workloads between different chip architectures is not always straightforward. Developers may need to adapt software, optimise models and improve performance for each platform.

Therefore, Google’s TPUs are not necessarily a direct replacement for Nvidia GPUs. OpenAI could use different processors for different computing requirements.

OpenAI Is Also Developing Its Own AI Chips

Purchasing chips from external suppliers is only one part of OpenAI’s broader hardware strategy.

The company has been working with semiconductor manufacturer Broadcom on custom AI processors. OpenAI has said it plans to deploy its custom-designed chips as part of a wider effort to develop its own computing infrastructure.

This approach could give the company more control over hardware design and performance.

By combining Nvidia GPUs, Google’s TPUs and custom-designed processors, OpenAI could build a more flexible computing environment.

The strategy also reflects a wider trend in the technology industry. Major AI companies are looking beyond traditional hardware suppliers as the cost and scale of AI infrastructure continue to increase.

Intel CEO Hints at Possible Cooperation With AMD

Another interesting development involves Intel and AMD.

During an October 24 fireside chat, Intel CEO Lip-Bu Tan reportedly confirmed that he had discussed possible areas of cooperation with AMD CEO Lisa Su.

The two companies have competed for decades in the processor market. However, competition has not prevented them from working together on shared industry technologies.

Tan did not disclose the details of the discussions or confirm a new partnership. As a result, it remains unclear whether future cooperation could involve processor development, manufacturing, AI hardware or another area.

Intel is also working to strengthen its manufacturing business and regain a stronger position in the semiconductor market. Its foundry operations and advanced manufacturing processes are important parts of that strategy.

Any future cooperation with AMD could be significant, but no specific agreement has been announced.

The AI Chip Market Is Becoming More Competitive

The developments involving OpenAI, Google, Intel and AMD point to a broader change in the semiconductor industry.

For years, Nvidia has been one of the main hardware providers powering the AI boom. Its GPUs are widely used for training large language models and running AI applications.

However, the growing demand for AI computing has encouraged companies to explore alternative processors and develop their own chips.

Google is expanding its TPU business, AMD is competing in the AI accelerator market, and Intel is working to strengthen its position in CPUs, manufacturing and other specialised technologies.

For AI companies, having access to multiple hardware suppliers could improve flexibility and help manage infrastructure requirements.

What Happens Next?

OpenAI’s reported plan to purchase Google AI chips and Intel’s discussions with AMD are still developing stories. The details of any Google purchase and the possible areas of Intel-AMD cooperation remain unclear.

What is becoming clear is that the AI hardware market is no longer centred on a single type of processor or supplier.

As AI models become more demanding, companies will continue looking for hardware that offers the right balance of performance, availability and cost.

Nvidia remains a major player, but Google, AMD, Intel and other semiconductor companies are working to secure a larger role in the next phase of AI infrastructure.