Technology companies AMD and Meta have taken a significant step in the race for artificial intelligence by announcing a multi-year strategic agreement focused on large-scale computing capabilities. The agreement strengthens an existing collaboration between the two companies and marks a long-term commitment to hardware, software, and infrastructure design.
In practical terms, the goal is to deploy up to 6 gigawatts of computing power based on AMD Instinct GPUs, a figure that illustrates the scale of the project and the current pressure to secure sufficient resources to train and run advanced AI models. This deal adds to Meta's other agreements with various vendors, such as AI infrastructure collaborations , and solidifies AMD as one of Mark Zuckerberg's company's key partners.
A massive deployment of GPUs and processors for AI
According to the joint statement, the agreement envisions the deployment of artificial intelligence infrastructure to reach 6 gigawatts of capacity, primarily relying on AMD Instinct GPUs. The collaboration extends beyond the current generation of products, encompassing multiple generations of chips and systems, including both GPUs and EPYC processors for data centers.
The first shipments under this agreement will begin in the second half of 2026 , when the first gigawatt of computing power is expected to be delivered. In this initial phase, Meta will use a custom Instinct GPU based on the MI450 architecture , designed specifically for its internal workloads.
Alongside these GPUs, sixth-generation AMD EPYC processors , codenamed "Venice," will be integrated. These are designed for demanding cloud computing and AI tasks. The ROCm software platform will be the foundation upon which many of the artificial intelligence applications that Meta will run on this infrastructure will be built and optimized.
All these components will be grouped into the AMD Helios rack architecture , an infrastructure design developed jointly within the framework of the Open Compute Project, an initiative that aims to promote open standards for data centers. Helios seeks to facilitate the deployment and maintenance of modular, scalable AI systems with more controlled energy consumption.
With this approach, both companies aim to build large-scale AI platforms that will increase computing capacity without driving up energy costs, an increasingly important aspect in Europe, where regulation and electricity prices influence the design of data centers.
Meta's strategic objectives and AMD's role
Meta founder and CEO Mark Zuckerberg described the partnership as a significant step in the company's strategy to diversify its computing power. The company has long emphasized its need for more powerful and flexible infrastructure to support its AI projects, from generative models to advanced personal assistants.
Meta aims to have massive, scalable capacity to power what the company calls "personal superintelligence," a concept encompassing AI functions for end users, internal tools, and services for businesses. In this context, relying on multiple hardware providers, including AMD, reduces risk and avoids dependence on a single manufacturer.
From AMD's side, President and CEO Lisa Su emphasized that this agreement positions the company as highly relevant within the global AI ecosystem . The semiconductor company aims to strengthen its presence in large European and American data centers, where a significant portion of the demand for computing power for artificial intelligence models is concentrated.
In addition to GPUs, the agreement expands cooperation to include EPYC processors for servers . Meta already has millions of these chips deployed in its global infrastructure and is positioned as a major customer for the next generation, which includes the Venice and Verano models, specifically optimized for data-intensive workloads and AI algorithms.
For the European market, where many large platforms operate data centers serving EU users, the ability to combine high performance and energy efficiency is a key selling point. The systems Meta builds with AMD technology could influence how future AI infrastructures are deployed in the region, both in terms of power consumption and cooling and space requirements.
Financial impact and purchase option for Meta
The agreement extends beyond the technological aspect. AMD has issued a performance-based warrant in favor of Meta , which will allow the social media company to acquire up to 160 million shares of AMD common stock if certain objectives are met.
These rights will be consolidated gradually as specific GPU delivery milestones are reached , starting with the first gigawatt of deployment and extending to the 6 gigawatts stipulated in the agreement. The structure is designed to align the incentives of both parties: the more systems that are deployed, the greater Meta's potential stake.
The possibility of acquiring shares will be contingent upon both technical and commercial parameters , as well as the performance of AMD's stock market value. The instrument will remain in effect until 2031, reinforcing the idea that the agreement is conceived as a very long-term collaboration and not a one-off arrangement.
AMD's Chief Financial Officer, Jean Hu, indicated that substantial revenue growth is expected over several years as a result of this project, along with a positive impact on the company's adjusted earnings per share. Following the announcement, AMD shares reacted with significant gains in pre-market trading, while Meta's shares barely moved.
From the perspective of European investors, these types of agreements are generally seen as a sign of confidence in the future demand for AI infrastructure . As EU companies and public administrations increase their use of natural language models, computer vision, and data analytics, the pressure on data centers will rise, and agreements like this help anticipate the necessary capacity.
A rapidly expanding and highly competitive AI sector
The alliance between AMD and Meta comes amid growing competition to secure high-performance chips for artificial intelligence. Major technology companies, including those with a direct presence in Spain and the rest of Europe, are forging large-scale agreements to guarantee their access to GPUs and processors for years to come.
Meta, for its part, combines collaboration with external suppliers like AMD with the development of its own chips under its MTIA (Meta Training and Inference Accelerator) program . The idea is to build a more diversified, flexible, and resilient infrastructure, capable of adapting quickly to changes in computing needs or regulations.
For AMD, the agreement with Meta adds to other large-scale strategic collaborations related to AI infrastructure, where the company aims to establish itself as a solid alternative to other dominant manufacturers in this market. The goal is to gain a foothold in the most ambitious cloud computing and advanced artificial intelligence projects.
The magnitude of the 6 gigawatts committed places this alliance among the largest AI hardware investment plans announced to date. The companies maintain that this capacity will accelerate the deployment of artificial intelligence services on a global scale, reaching billions of users through social media, messaging, and other digital platforms.
In Europe, where regulatory frameworks such as the EU AI Act are being discussed, the availability of powerful and efficient computing infrastructures will be a determining factor for companies and public bodies to develop AI solutions that meet new regulatory requirements without losing competitiveness compared to other regions.
The agreement between AMD and Meta, with its ambition to rapidly scale artificial intelligence infrastructure , illustrates the extent to which hardware has become a top strategic element. Beyond the gigawatt figures or the number of shares involved, the alliance paints a picture in which the combination of custom GPUs, EPYC processors, and open rack architectures will be fundamental to supporting the next generation of AI services, which will also reach users and businesses in Spain and the rest of Europe.
