Enterprise AI solutions, built from Granada
GPU Solutions runs its cluster in Madrid, but the company was born and thinks from Granada: headquarters at the PTS, a chair with the University of Granada, servers inside iQuantum, and an Andalusian ecosystem that no longer needs permission to do serious AI.
When a company looks for artificial intelligence solutions in Granada, it usually ends up talking to a generalist consultancy in Madrid or to an integrator reselling American cloud. We are the third option, and it deserves a proper explanation: GPU Solutions is the brand of BIAI Technology Project S.L., a Granada company headquartered on Avenida del Conocimiento — in the heart of the Health Sciences Technology Park (PTS) — operating its own NVIDIA GPU infrastructure in Madrid's only Tier III datacenter. The iron sits where it must; the head, the team and the R&D are here.
This article explains what we do from Granada, who we do it with — the University of Granada, the city council's iQuantum centre — and what an Andalusian company can ask of us if it wants to use AI without its data leaving Spain.
Why Granada can do serious AI
Granada needs no introduction: the University of Granada's ETSIIT has spent decades producing some of Spain's best computer engineers, and the city combines talent, reasonable costs and a research density few mid-sized European capitals can match. When El Confidencial titled its feature on us 'AI made in Granada', it wasn't poetic licence: it was the thesis. The Andalusian bet on AI is explicit public policy too — the regional government of Andalusia has placed artificial intelligence and the digital ecosystem among the priorities of its regional strategy, and Granada concentrates much of that ambition.
On that foundation we have built three pieces that reinforce each other: a university chair, a collaboration with Spain's first urban AI demonstration centre, and a commercial sovereign AI platform. One at a time.
The GPU Solutions-UGR Chair: accelerated computing and sustainable AI
On 10 November 2025, the rector of the University of Granada, Pedro Mercado, and our CEO, Julio Sola, signed the creation of the GPU Solutions-UGR Chair of Accelerated Computing Infrastructures and Sustainable AI, directed by Pablo García Sánchez, tenured professor at the Department of Computer Engineering, Automation and Robotics, based at the ETSIIT.
GPU-Lab-UGR
A demonstration laboratory with real accelerated-computing infrastructure, so researchers and students work on the same hardware industry uses.
Sustainable AI
Research into the energy efficiency of AI models — how much training and inference consume, and how to reduce it. The same obsession we apply to our liquid-cooled cluster with a PUE below 1.2.
Training
Programmes with university micro-credentials in accelerated computing, designed for working professionals, not just undergraduates.
Transfer
Carrying research results into companies and public administrations — the link that almost always fails in Spain, and the chair's reason to exist.
For a company, the chair means something very concrete: the infrastructure provider you talk to actively researches how to make AI consume less and perform more, and trains in Granada the talent that can later maintain your systems.
iQuantum: AI applied to running the city
iQuantum is the urban AI demonstration centre driven by Granada's city council — the first of its kind in Spain. In December 2025 it announced its equipment: three Lenovo ThinkStation PGX servers built on the NVIDIA Grace Blackwell (GB10) superchip, able to run models of up to 200 billion parameters locally to develop and test solutions for mobility, security, energy efficiency and urban resource management.
Our role: GPU Solutions provides its NVIDIA servers to take the developments born inside iQuantum to production, guaranteeing regulatory compliance and data residency on national territory. The demonstrator tests; our platform scales. It is exactly the circuit we advocate for all Spanish public-sector AI: experiment locally, produce on sovereign infrastructure, and let no citizen's data cross a border along the way.
What an Andalusian company can ask of us
The same as any of our clients — with the difference that the team serving you is a short walk from the PTS rather than the far side of a ticket queue. Our platform covers the full cycle:
- Private coding assistants for development teams — your code never leaves the cluster, no rate limits, predictable cost.
- Inference on open-source models (Qwen, Llama, DeepSeek, Mistral) by tokens or on dedicated GPU, served from Madrid with European latencies.
- Corporate RAG over your documentation — the corpus, the vector index and the model live inside the same perimeter.
- Development environments (pods) with B200 GPUs to train and fine-tune your own models, with Exascaler parallel storage.
- Compliance built in: ISO 27001, ENS Categoría Media, NIS2 and GDPR — with no US sub-processors in the chain.
For regulated Andalusian sectors — healthcare, public administration, agri-food with sensitive data, cooperative banking — that last line is not a footnote: it is the difference between using AI seriously and staying stuck in pilots.
From Granada, with the door open
We sign our R&D products with a 'From Granada with ♥' that is not irony: OdiModel, Odiverse and the rest of our lab are built here. If your company is in Granada, in Andalusia or anywhere else and wants to see how sovereign AI is actually built, write to us — and if you want to touch the hardware, we'll show you the cluster. The architecture conversation is free, and you leave it with a concrete proposal.