Telefónica has deployed a network of 17 mini data centers in Spain that changes the architecture of cloud computing. Instead of sending information via submarine cable to a server in Virginia for processing there, the data stays in your city, possibly in your neighborhood.
Why is it important. This infrastructure edge computing brings computing, storage and AI capacity to the end user. It reduces latency and keeps data under local jurisdiction, two critical requirements for applications that require immediate response or have some regulatory sensitivity.
The panoramic. The telecom has already activated ten nodes in cities such as Madrid, Barcelona, Seville and Bilbao, and will add another seven before June. In a period of five to seven years it could reach a hundred locations, all of them taking advantage of telephone exchanges that are no longer in use. copper quenching after 140 years.
Each node has between 1 and 2 megawatts of power and is equipped with NVIDIA accelerators for AI inference. The network adds 3 MW added starting, expandable according to commercial demand.
The context. European telecoms are desperately seeking new business models in the face of the ‘commoditization’ of connectivity. Telefónica has found a way to turn legacy infrastructure into a new product to sell. After all, their plants already have electricity, fiber and a direct connection to the 5G core.
The project received 93 million euros of European funds. Orange, Deutsche Telekom, Telecom Italia and 4iG already deploy similar networks in their countries: France, Germany, Italy and Hungary respectively.
What is happening. Telefónica does not compete against AWS, Azure or Google Cloud. Their commitment is complementary: solving use cases that large cloud providers do not serve at all well because they require processing alongside the user.
In fact, it negotiates with those same hyperscalers to offer their services over this distributed network. The objective is to combine connectivity, security and computing in a kind of ‘premium offer’ for those who need ultra-low latency.
In detail. Low latency is key in usage applications that do not support delays:
- Real-time video analysis.
- Management of drones or autonomous fleets.
- Digital twins industrial.
- Assisted or autonomous driving.
- Medical image processing.
Yes, but. The question is whether there is enough commercial demand to make one hundred mini-centers profitable. Many applications operate perfectly with the latency of a classic data center.
The success of the model depends on many great use cases that require that ultra-low latency coming to fruition. The autonomous car would be the most obvious, but its real deployment in Spain has not yet gone beyond a green shoot.
And now what. Commercial services will start shortly after the testing phase with clients. The real test will come when we see if companies and institutions are willing to pay a premium for that ultra-low latency.
- If it works, Telefónica will have found an ingenious second life for assets that seemed destined for scrapping or Idealista.
- If not, it will be another failed experiment by telcos to escape the data pipeline business.
In Xataka | Vodafone negotiates with Telefónica and Orange to create a common front: a RANco
Featured image | Telephone
GIPHY App Key not set. Please check settings