According to DCD, Microsoft is planning a massive $10 billion investment in AI data centers located in Sines, Portugal. The company’s president Brad Smith revealed the plans during Lisbon’s Web Summit, with Microsoft later confirming the details. This represents a major partnership with Start Campus and Nscale, who will deploy up to 12,600 Nvidia GB300 GPUs at the developing campus. The Sines location already has one operational data center building called SIN01 that launched in January 2025, with SIN02 currently under development offering 180MW of capacity. Microsoft has been aggressively expanding its compute capacity through similar deals worldwide, spending $11.1 billion on leases in just their most recent quarter alone.
The AI capacity gold rush
Here’s the thing about Microsoft‘s Portugal move – it’s not happening in isolation. The company has signed an estimated $60 billion in capacity deals with various “neocloud” providers recently. Just in November, they inked a $9.7 billion deal with Iren and a multi-billion dollar expansion with Lambda. They’re also longtime customers of CoreWeave and have deals with Nebius Group. Basically, Microsoft is throwing money at the AI infrastructure problem from every possible angle. When your CFO mentions “dramatically stepped up data center leasing activity” during earnings calls, you know there’s a serious capacity crunch happening.
Why Portugal makes sense
Sines isn’t exactly the first place you’d think of for hyperscale AI infrastructure, but the location has some strategic advantages. Most Portuguese data centers cluster around Lisbon, which is about 90km away. Sines already hosts the EllaLink subsea cable, with four more cables – Medusa, New CAM Ring, Nuvem, and Olisipo – scheduled to come online in the coming years. That connectivity is crucial for global AI workloads. And let’s be real – when you’re deploying 12,600 of Nvidia’s latest GPUs, you need serious power and connectivity. The Start Campus project aims for 1.2GW of IT capacity across six buildings eventually. That’s industrial-scale computing that demands robust infrastructure support from specialized providers who understand high-performance requirements.
What this means for the AI ecosystem
Microsoft’s spending spree tells us something important about where AI is heading. They’re not just building their own data centers – they’re leasing capacity everywhere they can get it. The partnership model with specialists like Nscale and Start Campus allows them to move faster than building from scratch. But here’s my question: when does this capacity rush start to strain the global supply chain for GPUs, power infrastructure, and specialized components? We’re talking about deployments at a scale that would have been unthinkable just a few years ago. For enterprises looking to deploy AI, this massive infrastructure buildout should mean more available compute down the line. But in the short term? Expect continued competition for GPU capacity and potentially higher costs as everyone scrambles to keep up with demand.
