Two numbers tell the story of AI infrastructure — as explored in the economics of AI compute infrastructure — in 2026. BlackRock’s consortium just agreed to acquire Aligned Data Centers for $40 billion — one of the largest private infrastructure deals in history. Meanwhile, Meta raised its 2026 capital expenditure guidance to $125-145 billion, nearly double what it spent in 2025.
Combined, these two moves represent roughly $185 billion flowing into AI infrastructure in a single year. The question is not whether AI needs this much compute. It is who will own it — and what that ownership means for the competitive landscape.
The BlackRock-led consortium — which includes Abu Dhabi’s MGX, Nvidia, and Microsoft through the AI Infrastructure Partnership (AIP) — is acquiring Aligned Data Centers as a platform play. Aligned operates 5 GW-plus of capacity across multiple markets, making it one of the largest AI-first data center platforms in the world.
The strategic logic is classic infrastructure investing: buy scarce physical assets, improve operations, and lease capacity to hyperscalers and AI lab — as explored in the intelligence factory race between AI labs — s who need compute but do not want to build it themselves. Macquarie Asset Management built Aligned from a niche U.S. operator into a global platform. BlackRock is buying at the inflection point where AI demand outpaces grid and supply-chain capacity.
The seller consortium includes Nvidia and Microsoft as strategic participants — not just financial investors. This means Aligned’s data centers will likely be optimized for Nvidia’s GPU architectures and integrated with Microsoft’s Azure cloud, creating a vertically integrated infrastructure stack.
Meta’s Approach: Build Everything In-House
Meta is taking the opposite approach. Rather than buying existing infrastructure, Zuckerberg is building from scratch — and spending more than any single company in history to do it. The $125-145 billion capex guidance for 2026 alone exceeds what Meta spent in 2024 and 2025 combined.
This spending is driven by Meta Superintelligence Labs, the research division tasked with building artificial general intelligence. Meta’s thesis is that AGI development requires proprietary infrastructure that cannot be rented or shared. If you are training models at the frontier, you need dedicated data centers with custom cooling, custom networking, and guaranteed power — none of which are available on the open market at the scaleMeta requires.
The Power Bottleneck
Both strategies converge on the same constraint: electricity. AI data centers consume orders of magnitude more power than traditional cloud infrastructure. A single AI training cluster can draw as much electricity as a small city. The companies that secure long-term power purchase agreements will have a structural advantage over those that do not.
BlackRock’s advantage is financial engineering. As the world’s largest asset manager, it can structure power deals at scale, partner with utilities, and deploy patient capital across decades-long infrastructure projects. Meta’s advantage is operational control. By owning its data centers, it can co-locate with power sources and design facilities around energy efficiency from day one.
What This Means for the AI Industry
Infrastructure is the new moat. Model architectures can be replicated. Training data can be collected. But physical data centers with guaranteed power take years to build and cannot be copied.
The buy-vs-build split will define the next decade. Companies that own infrastructure will have lower marginal costs. Companies that rent will have more flexibility but higher variable costs.
Nvidia is the arms dealer. Its participation in both the BlackRock consortium and its sales to Meta mean it profits regardless of which strategy wins.
AI startups face a squeeze. As hyperscalers lock up compute capacity through ownership and long-term leases, independent AI labs will find it harder and more expensive to access the GPUs they need.
The Bottom Line
The AI infrastructure race has entered its industrial phase. The era of renting cloud GPUs and training models on commodity hardware is ending. What replaces it is a capital-intensive, vertically integrated infrastructure buildout that resembles the early days of the electrical grid or the telecommunications backbone. The companies that control this infrastructure will control the AI economy.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
