When we talk about Artificial Intelligence (AI), the image that comes to mind is that of immense data centers full of servers, consuming huge amounts of energy to power calculations. This has been the cradle of great models, but this centralized AI is about to face a radical change.
We are witnessing one of the most significant evolutions in the technological landscape: the installation of AI in portable devices. This means that Artificial Intelligence is beginning to function without constantly depending on connections to the cloud or those massive data centers. This shift toward Edge AI could redefine our technological future.
If the experience with local AI proves positive and, above all, scalable, the economic consequences are hard to ignore.
In the medium term, the need for huge investments in these gigantic data centers could be drastically reduced. Less energy consumption, less infrastructure, and a much more sustainable business model in terms of logistics and the environment. It's a paradigm shift that could shake the foundations of technology giants like never before.
Several software companies are working intensively in this direction, and one of the most prominent is AI21 Labs. The company's website can be found at https://www.ai21.com/
One of their flagship products, which exemplifies this trend, is the model they have dubbed Jamba 3B. This model has the impressive capacity to handle 3 trillion parameters and process up to 250,000 tokens (a token is a unit of text or word). What's truly crucial is that it's designed to work on mobile devices like laptops.
We must be clear about one thing: what works efficiently on a laptop today will soon have optimized versions for our smartphones. Hardware miniaturization and software optimization go hand in hand.
An interesting point about Jamba 3B is that the company claims its model is open source. This statement should always be taken with caution. Although many models are advertised as open, the actual degree of openness can be very limited, restricting the most critical parts of the software or commercial uses.
However, this trend toward small and efficient models, even if not completely open, represents a significant democratization of AI. By reducing dependence on the hardware and APIs of large corporations, it opens the door for smaller companies to play a relevant role. They can innovate, create customized solutions, and compete with giants without needing their infrastructure.
Artificial Intelligence could be freed from the cloud: are small and powerful models about to democratize technology?
This major shift in the evolution of AI, moving from the Cloud to the Edge, will have medium-term consequences that are still difficult to fully imagine. But without a doubt, it seems to be a path that brings us closer to more accessible technology, in everyone's hands. As always, time will reveal the surprises AI has in store for us.
But for now, it makes me somewhat optimistic.
EN 
ES