Published: 28 October 2026 | Category: International AI business
OpenAI and Google could deepen their relationship despite competing in AI
OpenAI is reportedly planning to purchase Google’s artificial intelligence chips, according to a Reuters report published on 28 October 2026 that cites The Information. Neither company had provided further details on the reported purchase in the coverage available at the time. Reuters
If confirmed, the move would highlight an important feature of the AI industry: companies that compete directly in software and AI services can still become customers or suppliers within the same technology ecosystem.
OpenAI already relies heavily on computing infrastructure to train and operate its models. Google’s Tensor Processing Units, known as TPUs, offer another option alongside the graphics processing units (GPUs) widely used for AI workloads.
Why AI companies need specialised chips
Modern AI models require enormous amounts of computation, both during training and when responding to users.
Training involves processing large datasets to develop a model’s capabilities. Inference is the process of using a trained model to generate responses, analyse information or perform tasks.
These workloads require powerful processors, memory, networking equipment and reliable data-centre infrastructure.
Nvidia has become a major supplier of AI computing hardware. Its GPUs and software ecosystem are widely used by AI developers and cloud providers.
Google has developed its own TPUs to support its internal AI workloads and cloud services. As demand for computing capacity grows, these chips are increasingly relevant to organisations looking for alternatives to relying entirely on one supplier.
Why Google’s chips could matter to OpenAI
The reported purchase would fit into a wider effort by AI companies to secure enough computing capacity while managing cost and availability.
Different processors have different performance characteristics, software requirements and operating costs. A chip that is well suited to one workload may not be the most efficient choice for another.
Using more than one type of accelerator can potentially give a company greater flexibility. However, supporting different hardware platforms can also introduce engineering complexity, because software must be adapted and optimised for each environment.
The commercial value of a potential deal would therefore depend on factors such as chip performance, price, available capacity and the effort required to integrate the hardware into OpenAI’s systems.
Competitors can also be partners
The AI market is not divided into completely separate camps.
Google develops its own AI models and products, while OpenAI competes in areas such as AI assistants, enterprise services and developer tools. Yet companies in this market also depend on a common infrastructure ecosystem.
Cloud providers sell computing services to customers that may compete with their own products. Hardware developers supply processors to multiple AI companies. Model developers may distribute their technology through platforms operated by other firms.
Reuters also reported that OpenAI had announced plans the previous week to make advanced models available through Google Cloud. The reported chip purchase would deepen the companies’ commercial relationship if it goes ahead. Reuters
This illustrates how business partnerships can be driven by practical infrastructure needs, even when the companies involved compete elsewhere.
The economics of AI are changing
AI infrastructure has become a major strategic expense. Companies must decide how much computing capacity to build, rent or purchase, and how to balance performance against cost.
A more efficient chip can make a difference to the cost of serving large numbers of requests. But the overall economics also depend on electricity, cooling, data-centre construction, networking and software optimisation.
This is one reason the industry is investing in custom silicon. Large technology companies want greater control over their computing platforms rather than relying entirely on external hardware suppliers.
The trade-off is that developing and deploying specialised chips requires substantial investment and long-term planning.
What it means for the wider AI market
If major AI developers use a broader range of processors, the market could become more diverse. Nvidia would remain an important supplier, but alternative accelerators could gain a larger role in particular workloads.
That would not automatically mean that one chip maker is replacing another. AI infrastructure is a complex market in which multiple processor types can coexist.
For customers, more choice could eventually create opportunities to compare price, performance and availability. For developers, it could mean additional work to make applications run efficiently across different platforms.
AINieuws analysis: The reported deal is a reminder that the AI race is not only about who builds the most capable model. Access to affordable computing capacity is becoming a competitive advantage in its own right.
Laat een reactie achter