OpenAI reportedly plans to buy Google's AI chips: Nvidia's biggest customer seeks a second supplier
Direct answer
On October 27, 2026, The Information reported that OpenAI plans to buy Google's artificial intelligence chips, known as TPUs (tensor processing units), citing a person familiar with the matter. Reuters picked up the report the same day and stressed what is missing: no financial terms, volume, chip generation or timing were disclosed. OpenAI declined to comment to Reuters, and Google did not immediately respond. Until the companies confirm it, this is a reported plan, not a closed purchase.
From renting capacity to buying the chip
- It already rents TPUs: Reuters reported in June 2025 that OpenAI began renting Google's chips through Google Cloud to lower the cost of running its models.
- Another step last week: OpenAI said it would make its frontier models available through Google Cloud, deepening a commercial relationship with a direct competitor in AI products and search.
- The engineering signal: The Information also reported that OpenAI is hiring engineers to optimize its frontier models for TPUs. Adapting models to a different chip architecture takes serious work, so that hiring points to real workloads, not a superficial test.
- The spending backdrop: Reuters noted that Amazon, Microsoft and Alphabet plan to spend as much as $600 billion combined this year on AI infrastructure, while investors ask when that spending turns into returns.
Why a second supplier changes the negotiation
- Capacity in a tight market: with demand for compute surging, relying on a single supplier leaves any company exposed to waitlists and pricing. A second supplier gives options when allocating workloads.
- Commercial leverage: OpenAI is one of Nvidia's largest GPU buyers. Adding TPUs gives it another price and availability reference at the negotiating table, even as it keeps buying GPUs.
- The nearby precedent: Anthropic, which just made its S-1 prospectus public, describes a multi-chip strategy: Amazon's Trainium, Google's TPUs, Nvidia's GPUs and AMD's GPUs. Diversification is already the norm among the large labs.
- The honest limit: buying the chip is only half the work. Optimizing models, operating the infrastructure and sustaining performance in production take time, and the report gives no timelines.
What it means for your business
- Today, nothing changes on your bill: this is an infrastructure move between suppliers. Its effects, if the plan goes through, would arrive as capacity and pricing over the medium term.
- The lesson you can apply now: in your own business, avoid depending on a single model, a single cloud provider or a single tool for a critical function. The diversification the large labs practice protects small companies too.
- What to watch: a confirmation from OpenAI or Google, the volume and chip generation, and whether TPU-optimized models reach products your company already uses.
Want AI that works for your business?
Message us on WhatsApp and let's talk about automation and AI agents for your business.
WhatsApp inicIARelated articles
- OpenAI launched GPT-6.1 Terra at half price
- Anthropic filed its S-1: revenue, loss and risks
- Meta seeks up to $95B in capital for its AI
Frequently asked questions
Is OpenAI's TPU purchase confirmed?
No. It is a plan reported by The Information on October 27, 2026, picked up by Reuters, citing a person familiar with the matter. OpenAI declined to comment and no terms were disclosed.
Is OpenAI dropping Nvidia chips?
Nothing in the report suggests that. OpenAI remains one of Nvidia's largest GPU buyers; adding TPUs diversifies its supply rather than replacing it overnight.
What is a TPU?
It is Google's AI chip, designed to accelerate machine-learning workloads, especially the large-scale matrix calculations that AI models use.
Does this lower AI prices for my company?
Over the medium term, more competition among compute suppliers helps supply and pricing. In the short term, this report changes no published price.
