An infrastructure for the AI
By Javier Surasky
-AWS.png)
A central part of power in artificial intelligence lies in the infrastructure that enables the training, deployment, integration, and operation of models at scale. That is where Amazon Web Services, better known as AWS, enters the picture.
Launched in 2006, AWS helped create the modern cloud computing market, and its “first-mover advantage” has allowed it to remain the
world’s largest provider of cloud infrastructure, according to data from
Synergy Research, although Microsoft and Google have been narrowing the gap.
AWS’s power comes from the control it exercises over a layer
on which companies, governments, universities, startups, and international
organizations can build data systems, critical applications, and artificial
intelligence solutions.
Its scale helps explain its power: in 2025, AWS reached
$128.7 billion in annual sales, a figure comparable to the nominal GDP of
countries such as the Dominican Republic, Ecuador, or Bulgaria, with
year-on-year growth close to 20%. It also generated $45.6 billion in operating
income, making it one of Amazon’s main sources of profitability.
The AWS cloud spans 123 availability zones across 39
geographic regions, and it is still expanding: AWS has announced plans to
expand to, among other countries, Saudi Arabia and Chile. That territorial
presence matters because AI depends on tangible infrastructure: data centers,
energy, networks, jurisdictions, permits, latency, and regulatory sovereignty.
Whoever controls cloud infrastructure influences entry costs, speed of
adoption, security, data localization, regional availability, and the ability to
experiment with advanced models.
But that is not all: AWS is also developing its own AI
chips, such as Trainium and Inferentia, seeking greater strategic independence
and a foothold in a multibillion-dollar segment of the digital economy.
Basic facts
- AWS was not born as a visible consumer product, but as a way to meet an internal Amazon need: building scalable infrastructure for its own business. That capacity later became a service that could be sold to third parties, with early offerings such as Amazon S3, for storage, and Amazon EC2, for on-demand computing capacity.
- Its main activity is the provision of cloud computing infrastructure and services: storage, servers, databases, networking, cybersecurity, data analytics, artificial intelligence, and machine learning.
- AWS remains the world’s largest provider of cloud infrastructure. Estimates based on Synergy Research Group place it at around 28% of the global cloud infrastructure services market in 2025.
- It provides infrastructure to train, host, and deploy artificial intelligence systems, including machine learning services, foundation models, proprietary chips, and scalable computing capacity.
In #4, the map turns to another key actor, but a very
different one: Stanford HAI.