SAN FRANCISCO (Reuters) - Alphabet, on Wednesday, April 10, 2025, unveiled its seventh-generation artificial intelligence chip, named Ironwood, marking a significant advancement in AI processing technology. According to the company, Ironwood is meticulously designed to accelerate the performance of AI applications, addressing the escalating demands of modern AI workloads.

The Ironwood processor is specifically geared towards the intensive data crunching required when users interact with AI software, such as OpenAI's ChatGPT. In the tech industry, this is commonly referred to as "inference" computing. These chips are engineered to perform rapid calculations, essential for generating answers in chatbots or producing various other types of responses with minimal latency.

Google's multi-billion dollar investment, spanning roughly a decade, underscores its commitment to developing viable alternatives to Nvidia's powerful AI processors. Google's tensor processing units (TPUs), while primarily accessible to its internal engineers or through its cloud service, have already provided a competitive edge in its AI endeavors.

Notably, Google has strategically bifurcated its TPU family for at least one generation. One version is meticulously tuned for constructing large AI models from the ground up. Concurrently, its engineers have developed a secondary line of chips, streamlining certain model-building features to curtail the costs associated with running AI applications.

Amin Vahdat, a Google vice president, explained that the Ironwood chip is explicitly designed for running AI applications, focusing on inference. It is engineered to operate seamlessly in clusters of up to 9,216 chips, facilitating large-scale AI deployments. Unveiled at a cloud conference, the new chip amalgamates functionalities from earlier split designs and augments the available memory, thereby optimizing it for serving AI applications. Vahdat emphasized the escalating importance of inference in contemporary AI applications.

The Ironwood chips boast double the performance per unit of energy compared to Google's Trillium chip, which was announced the previous year. This leap in energy efficiency marks a substantial improvement in computational capabilities. Google utilizes its own chips to build and deploy its Gemini AI models, showcasing a vertically integrated approach. However, the company has not disclosed the specific chip manufacturer responsible for producing the Google design.

Alphabet shares experienced a surge during the regular session, closing up by 9.7% following President Donald Trump's abrupt reversal on tariffs. This market reaction underscores the significance of technological advancements in the AI sector and their potential impact on market dynamics.

(Reporting by Max A. Cherney in San Francisco; Editing by Sonali Paul) By Max A. Cherney