- Nvidia and OpenAI inked a $100B non-binding letter of intent for OpenAI to deploy up to 10GW of GPU infrastructure to train next-gen AI models
- The deal could benefit telcos by increasing demand for fiber and network capacity and creating new partnership opportunities for AI as a service
- But telcos could also find themselves competing for AIaaS business
Nvidia's $100 billion agreement with OpenAI to provide investment, GPUs and infrastructure potentially opens both opportunities and threats for telcos looking to use AI for their own operations and to provide AI as a service to customers.
Nvidia and OpenAI signed a "letter of intent for a strategic partnership to deploy at least 10 gigawatts of Nvidia systems for OpenAI's next-generation AI infrastructure to train and run its next generation of models on the path to deploying superintelligence," according to a statement from the two companies on Monday.
Nvidia plans to invest $100 billion in OpenAI as new systems are deployed, with the first phase planned to come online in the second half of 2026 using the Nvidia Vera Rubin platform.
Scott Raynovich, founder and principal analyst at Futuriom, noted that the deal is still tentative. "This announcement is a ‘letter of intent,’ as labeled by Nvidia itself, not a contract. So it’s a loose partnership," he said.
"It can also be read as a circular vendor financing deal, because Nvidia is pledging money to OpenAI in exchange for OpenAI buying Nvidia systems. It’s mutually beneficial,” he said.
Sign of a bubble?
The agreement is potentially a sign of a data center bubble. “Lately, we’ve been reading about people asking if the acceleration of data center capacity construction is a 2000-like bubble. There are many similarities. Announcing a vendor financing arrangement doesn’t exactly dissuade that notion,” Raynovich said.
Additional data center capacity would step up the need for fiber and network connections – good news for telcos. "But right now it appears to me that the capacity is primarily gated by power constraints, and we aren’t sure if all the capacity will be needed," Raynovich said.
For Nvidia, the deal locks up a key GPU customer, said Mario Morales, IDC group VP, enabling technologies and semiconductors. If OpenAI is using Nvidia processors, it's not using AMD or Qualcomm. However, the OpenAI deal is not exclusive to Nvidia; OpenAI is free to deploy competing chips from AMD or others, according to a blistering and insightful analysis on The Register.
"The announcement has enough wiggle room to drive an AI-powered self-driving semi through – it's a letter of intent for a strategic partnership, which is a non-binding kinda-sorta contract, and the deal calls only for Nvidia to invest 'up to' $100 billion for as long as OpenAI keeps buying its chips," the Reg notes.
The deal expands OpenAI's capacity, potentially bringing in new customers, for which Nvidia would likely share revenue, Morales said – although specifics of the deal were not disclosed.
Hedging against the hyperscalers
This would not be Nvidia's first investment in a customer. Nvidia owns about 7% of CoreWeave's Class A shares as of June 30 and made an order with CoreWeave worth a minimum $6.3 billion. "Nvidia is likely using CoreWeave as a hedge against the hyperscalers, who are all large Nvidia customers, but also potential competitors in that all three have some version of their own AI silicon," AvidThink analyst Roy Chua said last year.
Telcos are among the customers that the partnership would bring in. "Telcos will be a vertical that OpenAI is going to go after," Morales said. Telcos can partner with OpenAI to serve enterprises. But telcos will also compete for enterprise business with OpenAI and other AI providers. And telcos will want to preserve choice, and not become locked in to Nivdia as a platform provider, Morales said.
The Nvidia deal potentially provides funding to help OpenAI meet other financial commitments, including a $10 billion deal with Broadcom to jointly develop AI chips and a whopping $300 billion to Oracle for cloud infrastructure. OpenAI is only expecting to bring in about $13 billion in revenue in 2025 and not turn a profit until 2029, as noted by my colleague Diana Goovaerts.
For telcos and service providers, Nvidia’s massive investment in OpenAI signals both promise and peril, offering new avenues for AI partnerships and infrastructure growth, while also highlighting the risks of platform dependency and speculative overbuild in the datacenter market.