OpenAI Plans to Procure TPUs from Google, Nvidia's Computing Dominance Faces New Challenge
According to people familiar with the matter, OpenAI intends to buy AI chips from Alphabet-owned Google; the final terms of the deal are not yet clear, and both OpenAI and Google declined to comment. Google began selling TPUs directly to some customers this year, and Oracle also announced this Tuesday that it will for the first time offer computing power based on Google TPUs on its cloud platform. OpenAI previously announced plans to deploy its in-house chip later this year; its internal chip Jalapeno, developed in cooperation with Broadcom, had senior Google employees involved in its development.
Why read this
AI score 60
The original text discloses OpenAI's move to procure Google TPUs, and provides its in-house chip development and long-term partnership with Nvidia as background, to facilitate understanding of changes in the computing supply landscape.
Topics
Body · AI translation
Cailianshe Financial News Agency
Open the APP
OpenAI plans to procure Google TPUs! Nvidia's computing power hegemony faces a new challenge
Cailianshe Xiaoxiang, Editor: Lin Qi 2026-10-28 10:57
① According to people familiar with the matter, OpenAI intends to buy AI chips from Alphabet's Google, the latest move in OpenAI's massive buildout of its AI infrastructure; ② As companies seek cheaper chips suitable for running AI models, Google's tensor processing units (TPUs) are gaining popularity as an alternative to Nvidia's market-leading AI chips.
Cailianshe, October 28 (Editor: Xiaoxiang) — According to people familiar with the matter, OpenAI intends to buy AI chips from Alphabet's Google, the latest move in the ChatGPT developer's massive buildout of its AI infrastructure.
The people said the final terms of any deal are not yet known. OpenAI and Google both declined to comment.
As companies seek cheaper chips suitable for running AI models, Google's tensor processing units (TPUs) are gaining popularity as an alternative to Nvidia's market-leading AI chips. Google began selling TPUs directly to some customers this year.
The move also reflects OpenAI's multi-pronged strategic intent to urgently boost its own AI computing power.
The company previously announced plans to begin deploying its own custom chips later this year to support the operation of its AI models. That means it has joined the "custom chip wave" set off by multiple companies, jointly challenging Nvidia's dominance in the computing power field.
It is worth noting that Jalapeno, the in-house chip OpenAI developed with Broadcom, was developed with the assistance of veteran Google employees. Richard Ho, who heads OpenAI's in-house chip team, previously participated in the development of Google's TPUs.
Google began developing its TPUs more than a decade ago. Initially, these TPUs were used to accelerate the company's web search engine and improve efficiency. But the chips quickly proved extremely valuable for handling AI workloads and in recent years have become an alternative to Nvidia GPUs.
Nvidia's computing power hegemony faces a new challenge
The procurement plan will deepen the ties between OpenAI and Alphabet's Google — even though the two companies are direct competitors in AI products and search.
For a long time, OpenAI's relationship with Nvidia has been quite close. From the training of the first-generation GPT model to ChatGPT sweeping the globe, almost all of OpenAI's core model evolution has been built on clusters of tens of thousands of Nvidia GPUs. Jensen Huang personally delivered the world's first DGX AI supercomputer to OpenAI, and OpenAI's explosive rise directly pushed Nvidia to the top of global market capitalization.
In August this year, Nvidia CEO Jensen Huang said OpenAI had committed to deploying Nvidia AI infrastructure on a large scale by 2030, with existing and planned deployments of about 12 gigawatts and the opportunity to expand to about 16 gigawatts.
However, that has not stopped OpenAI from seeking to diversify its supply chain. Facing increasingly high inference costs, a massive computing power gap that can run into millions of chips, and the bargaining disadvantage and supply-cut risks brought by over-reliance on a single supplier, OpenAI is clearly unwilling to stake its entire fate on Nvidia alone.
It is not alone. Oracle also announced on Tuesday that it will, for the first time, offer customers computing power based on Alphabet's Google TPU chips on its cloud platform.
Regarding this arrangement, Mahesh Thiagarajan, president of Oracle Cloud Infrastructure, said in a statement that different AI workloads have different computing power needs, and this choice allows customers to flexibly select the infrastructure best suited in performance, throughput and economics for each AI workload.
Google executive Amin Vahdat said in a statement announcing the Oracle deal, "The next generation of AI will require diverse computing power."
Global Market Intelligence
Artificial Intelligence
Cailianshe statement: The content of this article is for reference only and does not constitute investment advice. Investors who act on it do so at their own risk.
Business cooperation
Popular unlocks
[Gold-digging industry leader] Robotics + industrial mother machines, the miniature ball screws used in dexterous hands have received customer designation, and the company is customizing joint module products for humanoid robots for domestic enterprises; this company's second-generation machine tool capacity will exceed 40 units/month
Robotics + industrial mother machines, the miniature ball screws used in dexterous hands have received customer designation, and the company is customizing joint module products for humanoid robots for domestic enterprises; second-generation machine tool capacity will exceed 40 units/month, and this company's aircraft electric drive products have completed delivery.
Positioned at the entry point for physical AI tactile data, this company has built a dedicated robot electronic skin R&D platform, completed iterative upgrades of core process equipment, has launched flexible electronic skin products across multiple technology routes, and has sent samples to multiple downstream customers for testing, with individual customers already placing sample orders.
Related reading
SK Hynix announces explosive earnings: revenue tops 100 trillion won for the first time, operating profit soars 632%
Cailianshe Bian Chun 2026-10-28 09:10
① South Korean memory chip giant SK Hynix announced record third-quarter earnings: quarterly revenue exceeded the 100 trillion won mark for the first time, and operating profit soared 632.1%; ② SK Hynix expects memory demand to continue exceeding supply through 2030; ③ SK Hynix also said it plans to return about 100 trillion won (about $75.5 billion) to shareholders this year.
No rate hike tonight? Fed decision day arrives: Wall Street closely watches Warsh's remarks!
Cailianshe Xiaoxiang 2026-10-28 09:02
① According to the schedule, the Federal Reserve will announce its October interest rate decision at 2 a.m. Beijing time on Thursday, and Fed Chair Warsh will hold a regular press conference at 2:30. ② The market currently widely expects that, after raising rates by 25 basis points in September, the Fed will likely keep rates unchanged at this meeting. The CME FedWatch tool shows only about a 7% probability that the Fed will raise rates tonight.
① Melania said that the ceasefire agreement between Russia and Ukraine in Aleshki took effect early at 4 a.m. Moscow time on the 27th, and the ceasefire scope has been expanded to areas beyond Aleshki; ② Both the Russian Defense Ministry and senior Ukrainian officials confirmed that the two sides had ceased fire in the Aleshki area that day to carry out humanitarian operations.