OpenAI has signed a multi-year agreement of 38,000 million dollars (about 35,500 million euros) with Amazon to run and scale their AI models in AWS. This is a movement of enormous significance that diversifies its dependence beyond Microsoft and that represents a key boost for Amazon in the computing race for AI. The company will be able to start using AWS infrastructure immediately.
Beyond the figure, this pact of seven years implies access to hundreds of thousands of Nvidia GPUs hosted in Amazon data centers, with the aim of deploy all capacity before the end of 2026 and the possibility of expanding it in 2027. The context is clear: increasingly complex models (chatbots, assistants and agents capable of completing tasks on their own) require power that few providers can reliably guarantee.
An agreement to strengthen OpenAI infrastructure
For OpenAI, the immediate benefit is large-scale training and inference capability with guarantees of availability, something critical to continue evolving systems such as ChatGPT and its future versions. The company also gains multi-cloud flexibility to distribute loads and negotiate costs, at a time when computing demand is the main bottleneck in the sector.
On Amazon's side, the agreement represents a boost to AWS in the field of AI. In recent months, the debate over a possible Amazon delay versus Azure and Google Cloud has been recurring; hosting a relevant part of OpenAI payloads is a strong message to the market. It is not surprising that, after the announcement, Amazon shares reacted higherreflecting investor optimism.
What changes for Microsoft and the ecosystem
The agreement does not break with Microsoft, but it does ends de facto exclusivity which had Azure as the primary compute provider for OpenAI. From now on, we will see a more diversified cast between clouds, with impact on prices, chip delivery times and development speed of new functions. For the end user, this can translate into faster throwsbetter service availability and more ambitious AI features on Android apps and devices (from translation and summary in real time even image editing or “always on” assistants on mobile phones and wearables).
At the macro level, the agreement reinforces the trend of mega investments in infrastructure that we are seeing across the AI ecosystem: from clusters with tens of thousands of accelerators to undersea cables and renewable energy to sustain them. The competition for Nvidia chips continues to set the pace, and commitments like this are, at the same time, capacity insurance and a sign that the race is accelerating.
What does it mean for you and for the market?
In the short term, the user will not notice any immediate changes: ChatGPT will continue to function normally. But in the medium term, More computing means more capable models and new integrations on Android and services from Google, Samsung or Amazon. For the market, this pact redefines the balance of forces in the cloud sector, pushes other players to respond with similar agreements and leaves a clear reading: the development of AI, both for consumption and for companies, is played in data centers. It seems that OpenAI has no intention of stopping, remember that recently it also there was a similar alliance with Samsunga technological giant, which could contribute a lot in this area of artificial intelligence. But do you think this movement will accelerate the arrival of really useful assistants on mobile? We read you in the comments.
