On September 10, 2026, Amazon Ads announced a pilot integration with OpenAI that allows select American advertisers to place conversational ads in ChatGPT through the Amazon DSP platform. In parallel, at the Goldman Sachs conference, Nvidia CEO Jensen Huang once again confirmed the company's forecast of 70% revenue growth in the next fiscal year, citing unabating demand for AI chips and systems.
The mechanics of Amazon's integration are simple but strategically significant: advertisers use existing campaigns in Amazon DSP to extend reach to ChatGPT inventory, where ads appear beneath organic responses in the form of text and images marked "sponsored." OpenAI retains control over delivery and placement, while Amazon provides management tools, including automatic generation of creatives from product feeds and bidding by CPC or CPM. This is not merely a new channel — it is the transformation of DSP into a gateway to emerging conversational inventory, where users actively compare options and make decisions.
For Nvidia, Huang's forecast rests on the observation that the company "runs every model" — from OpenAI and Anthropic to Google and open-weight solutions. Revenue over the last 12 months has already exceeded 300 billion dollars at a margin of about 75%, and unconstrained demand, according to the CEO, exceeds 100%, although supply limits actual growth. The next generation, Vera Rubin, will raise the system price from 18 thousand to 40 thousand dollars per GPU, which further strengthens revenue per unit.
Methodologically, the statements from Amazon and Nvidia differ in their degree of disclosure. Amazon publishes details of the pilot (limited to the US, a managed service, specific partners such as Delta Vacations), but does not disclose inventory volumes, CPM, or precise performance metrics. Nvidia provides a rare annual forecast for the company, yet does not publish detailed breakdowns by demand segment and acknowledges memory-related constraints. This leaves room for interpretation: how sustainable is 70% growth if supply bottlenecks persist.
In the landscape, these steps contrast with previous approaches. Previously, OpenAI developed Ads Manager on its own and through partners such as Adobe or Criteo; connecting Amazon DSP adds scale and simplifies entry for brands already working with Amazon. For Nvidia, this is a continuation of a strategy in which the company does not merely sell chips but positions itself as infrastructure for the entire AI stack — from training to inference and now physical AI and cybersecurity. Compared with the narrower forecasts of other vendors, Huang's statement looks like a signal to the market of long-term confidence rather than a one-off hype.
The implications for the industry are obvious: conversational advertising is ceasing to be an experiment and is becoming part of the omnichannel funnel, where DSP platforms like Amazon gain additional leverage. For Nvidia, the confirmed growth means continued investment in capacity and new verticals, but it also strengthens the ecosystem's dependence on a single compute supplier. What were previously considered separate markets — advertising in chats and hardware for models — now clearly intersect through shared players.
It remains unclear how quickly OpenAI will scale ChatGPT inventory beyond the pilot and what real lift metrics advertisers will receive compared with traditional channels. For Nvidia, the key question is whether partners and the company itself can remove supply constraints quickly enough to realize unconstrained demand. The next reports and independent audits will show whether 70% is a sustainable trend or a reflection of the current cycle.
Together, these two announcements demonstrate how the infrastructure and advertising layers of the AI economy are synchronizing: access to users and access to compute are becoming interdependent assets.
