Google has introduced its latest Tensor Processing Unit (TPU) named Ironwood, which is specifically designed for inference tasks, focusing on reducing the costs associated with AI predictions for millions of users. This shift emphasizes the growing importance of inference in AI applications, as opposed to traditional training-focused chips, and aims to enhance performance and efficiency in AI infrastructure. Ironwood boasts significant technical advancements over its predecessor, Trillium, including higher memory capacity and improved data processing capabilities.

Amazon Web Services is set to unveil an updated Graviton4 chip featuring 600 gigabits per second of network bandwidth, the highest in the public cloud. This advancement positions AWS to compete more effectively against Nvidia in the AI infrastructure market, as the company aims to reduce AI training costs and enhance performance with its upcoming Trainium3 chip. AWS's focus on custom chips illustrates its strategy to dominate the AI infrastructure stack and challenge traditional semiconductor companies like Intel and AMD.

Apple has introduced its custom A19 Pro chip and N1 wireless chip in the new iPhone lineup, enhancing performance and prioritizing artificial intelligence capabilities. With these advancements, Apple aims to gain full control over its core iPhone chips, reducing reliance on third-party suppliers like Qualcomm and Broadcom, while also planning to shift more chip manufacturing to the U.S.

OpenAI has announced a multibillion-dollar partnership with Advanced Micro Devices (AMD) to develop AI data centers powered by AMD processors, marking a significant challenge to Nvidia's dominance in the market. The five-year deal includes a commitment from OpenAI to purchase 6 gigawatts of AMD chips, starting with the MI450 model, as demand for AI computing power continues to surge.

The U.S. government has announced new restrictions on the export of artificial intelligence chips from companies like Nvidia and AMD to China, aiming to hinder the country's advancements in AI technology. This move reflects a broader strategy by the Trump administration to combat China's growing capabilities in the tech sector.

Cerebras Systems has withdrawn its plans for an IPO just days after raising over $1 billion in funding, citing no specific reason for the decision. The company, which aims to compete with Nvidia in the AI chip market, continues to express interest in going public in the future despite the current U.S. government shutdown and its reliance on a single customer, G42.

Nvidia is working on a new AI chip built on its Blackwell architecture, aimed at outperforming its current H20 model available in China. Although U.S. President Trump has hinted at the possibility of allowing the sale of more advanced chips to China, regulatory approval remains uncertain due to security concerns. Samples of the new chip are expected to be delivered to Chinese clients as early as next month.

Annapurna Labs, an Israeli chip-design startup acquired by Amazon a decade ago, is now pivotal in creating a new AI supercomputer in Austin, Texas. Originally operating in secrecy, its innovations in chip technology are crucial to Amazon's competitive edge in the tech industry.

Tech companies and startups are developing innovative microchips aimed at reducing the energy consumption of AI supercomputers. One such startup, Positron, has created chips that are more energy efficient for AI inference, potentially saving companies significant costs and energy as they seek alternatives to Nvidia's dominant products.

Qualcomm is expanding its business beyond smartphone modems by introducing new AI chips aimed at data centers, set to launch next year. This move comes as the company seeks to diversify its revenue streams following the loss of major clients like Huawei and Apple's chip development efforts.