Even after releasing its own AI chip… Google: "Nvidia and we are in love with each other"

Even after releasing its own AI chip… Google: "Nvidia and we are in love with each other"

"We love Nvidia, and Nvidia loves us."

Google emphasized that it maintains a strong relationship with Nvidia, even as it fully unveiled its 8th-generation new product of its own artificial intelligence (AI) semiconductor, the Tensor Processing Unit (TPU).

Mark Lohmeyer, VP of Google Cloud AI and Compute Infrastructure, answered this question about Google's relationship with Nvidia at a press conference for 'Google Cloud Next' held on the 23rd (local time) at the Mandalay Bay Convention Center in Las Vegas, USA.

VP Lohmeyer emphasized, "Many of our customers use (Nvidia's) Graphics Processing Units (GPUs), and we are deeply collaborating with them."

He added, "The core of Google Cloud's strategy is customer choice and openness, so we are collaborating more and more deeply with Nvidia," and "We plan to introduce Nvidia's new AI accelerator 'Vera Rubin' on Google Cloud later this year."

He also introduced that 'Thinking Machines Lab', a startup founded by Mira Murati, former Chief Technology Officer (CTO) of OpenAI, is using Nvidia GPUs through Google Cloud.

However, he added that Google, with its 8th-generation TPUs, is ahead of cloud competitors like Amazon Web Services (AWS) or Microsoft (MS) 'Azure', which still only have 1st to 3rd generation self-developed chips.

Google primarily uses Nvidia GPUs even in on-premise cloud environments that operate on customers' internal servers.

Muninder Sambi, VP of Google Distributed Cloud (GDC) Product Management, explained, "We are collaborating with Nvidia to provide top-performing GPUs in on-premise environments."

He further explained that GDC servers built on customer premises cannot use TPUs due to the difficulty of installing water-cooling systems, and therefore are mainly configured with GPUs.

Google's careful attention to its relationship with Nvidia, even while showcasing the technical prowess of its own chips, is interpreted as being due to the high customer demand for Nvidia GPUs.

Therefore, this is a strategy to seize leadership in the AI infrastructure market through a dual approach, equipped with both TPUs and GPUs.

Nvidia also highlighted the two companies' collaborative relationship in conjunction with Google Cloud's event.

On the first day of the 'Google Cloud Next' event, the 22nd, Nvidia emphasized through its official blog, "Nvidia and Google Cloud have been collaborating for over 10 years," and "We have jointly developed a comprehensive AI platform spanning all technology layers."

When Google officially launched its 7th-generation TPU, 'Ironwood', in November last year, Nvidia, through its official X (formerly Twitter) account, expressed, "We are pleased with Google's success," but also took a cautious stance, stating, "We are a generation ahead of the industry."