PayPal has started to build on its LLM-powered agentic commerce services through the use of NVIDIA Nemotron open models.
Following this announcement, the strategy represents an important development towards gaining more control over its AI stack, from model selection and training to the process of deployment and optimisation. This is set to enable the company to deliver AI experiences with improved precision, lower latency, and more reliability to its collaborators and customers.
Both companies share the belief that the use of open LLMs has the possibility to play an important role in the process of democratising AI, as well as enabling organisations to build production-grade systems without the need to rely entirely on proprietary models.
More information on PayPal’s process of building LLM-powered services with NVIDIA Nemotron open models
According to the official press release, open models were developed in order to enable organisations to fine-tune AI systems for their specific domains, which represents an important capability in enterprise environments. At the same time, this strategy aims to give companies like PayPal optimised control over how they deploy AI in order to address real-world business challenges.
This announcement comes around the same time PayPal announced its adoption of ACP in order to broaden payments and commerce in ChatGPT. As part of its collaboration with OpenAI, PayPal was set to support payment processing for merchants utilising OpenAI Instant Checkout, while also connecting its global merchant network to the company and developing a platform for small businesses and brands to sell within ChatGPT.
Furthermore, the partnership was expected to facilitate the capabilities of PayPal’s wallet in Instant Checkout, including multiple funding options such as bank, balance, and cards, as well as the company’s buyer and seller safeguards, and post-purchase services like tracking and dispute resolution.
Earlier in September 2025, NVIDIA invested USD 100 billion in an OpenAI partnership, with both companies confirming the plans in a joint announcement, describing the arrangement as two connected transactions: NVIDIA focused on the process of providing non-voting shares funding, which OpenAI then could use to acquire the hardware.
