Artificial intelligence runs on hardware. Behind every large language model, every agentic workflow, every enterprise AI deployment is a physical facility filled with GPUs, cooling systems, fiber networks, and dedicated power infrastructure. The scale at which that infrastructure is being built right now is historically unusual — and the money fueling it is coming from a set of overlapping investors whose motivations, risk profiles, and incentive structures vary considerably.
Hassan Taher, whose consulting firm Taher AI Solutions has advised organizations on AI strategy since 2019, has written about AI’s societal and environmental implications in several contexts. His upcoming book on AI and climate change examines how the technology can address environmental challenges — which makes the energy appetite of AI data centers a subject of particular relevance to his work. The infrastructure boom now underway is producing both the compute capacity that makes AI possible and a set of financial and environmental questions that organizations building on top of that infrastructure cannot avoid.
The Scale of the Buildout
The numbers being spent on AI infrastructure are large enough that standard frames of reference don’t hold. Tech giants spent roughly $580 billion in 2025 alone converting empty fields, deserts, and former industrial sites into GPU-dense facilities. Looking ahead, hyperscalers are planning to spend nearly $700 billion on data center projects in 2026: Amazon projects $200 billion in capital expenditures, Alphabet between $175 billion and $185 billion, and Meta between $115 billion and $135 billion. All told, McKinsey estimates that $7 trillion in capital will flow into global data center infrastructure by 2030, with over 40% of that spending expected in the United States.
The single largest announced commitment is the Stargate Project, a joint venture formed in January 2025 by OpenAI, SoftBank, and Oracle, with a $500 billion, multi-year buildout target across multiple U.S. states. At full build, Stargate is projected to deliver up to 10 gigawatts of AI-ready power — equivalent to the combined power consumption of New York City and San Diego. OpenAI’s own announcement confirmed that five new Stargate sites were selected from more than 300 proposals across 30 states, with a five-year agreement between OpenAI and Oracle exceeding $300 billion.
The Hyperscalers
Amazon, Microsoft, Google, and Meta collectively represent the largest single category of data center investment. These hyperscalers are spending at levels that have unsettled their own investors. Amazon’s $200 billion capex projection for 2026 was well above consensus expectations; Amazon’s stock dropped roughly 8-10% on the announcement, reflecting investor concern about the payback timeline. Alphabet’s $175-185 billion figure has been revised upward three times from an initial estimate of $71-73 billion for 2025.
Microsoft’s approach has been distinctive. Rather than building exclusively, the company has committed more than $60 billion to leases with neocloud providers that build data centers and stock them with advanced AI chips. Nscale alone secured $23 billion in business from Microsoft, with planned facilities in the UK, Norway, Portugal, and Texas. Microsoft CEO Satya Nadella has acknowledged the risk publicly, predicting a potential “overbuild” of computing capacity — a statement that did nothing to slow the company’s commitments.
Nvidia and the Circular Financing Dynamic
GPU manufacturer Nvidia sits at the center of the infrastructure buildout in a way that extends well beyond selling chips. As AI labs have scrambled to secure compute, Nvidia has been investing its cash back into the industry in increasingly unconventional ways. In September 2025, the company received a $100 billion investment from OpenAI, giving it capacity to buy more of its own GPUs. Nvidia also bought a 4% stake in rival Intel for $5 billion.
This has contributed to what analysts now call circular financing — a structural pattern where investors pour capital into AI startups, which then spend that money on the investors’ own chips, cloud services, and infrastructure. Microsoft invested billions in OpenAI, which funneled much of that back into Microsoft’s Azure cloud, creating guaranteed demand for the services funding the investment. OpenAI contracted Oracle to build Stargate data centers while using AMD and Nvidia GPUs purchased with investor-backed capital. The concern among some analysts is that when a company’s customers are also its investors, it becomes harder to distinguish genuine market demand from manufactured revenue.
Asset Managers and Institutional Capital
Beyond the technology companies, a growing pool of institutional and sovereign capital is flowing into AI infrastructure as an asset class. Brookfield Asset Management launched its Brookfield Artificial Intelligence Infrastructure Fund in November 2025, targeting $10 billion in equity commitments to acquire up to $100 billion in AI infrastructure assets across the full value chain, from energy and land to data centers and compute. The fund already had $5 billion in commitments from Brookfield itself, Nvidia, and the Kuwait Investment Authority at launch.
At least $178.5 billion in data center credit deals were struck in the United States in 2025 alone. Oracle, Meta, and Alphabet contributed to pushing global bond issuance above $6.57 trillion for the year. JPMorgan Chase expects issuers to tap essentially all major debt markets to finance the buildout going forward. In October 2025, Meta closed a $30 billion deal structured by Morgan Stanley — the largest private capital transaction on record — where the financing sits in a special purpose vehicle tied to Blue Owl Capital, keeping the debt off Meta’s balance sheet.
Geographic Distribution and Sovereign Participation
The capital flowing into AI data centers is not exclusively American. McKinsey’s framework for categorizing investors in the data center buildout identifies five archetypes: Builders (real estate developers and construction firms), Energizers (utilities and equipment manufacturers), Technology developers (semiconductor firms), Operators (hyperscalers and colocation providers), and AI Architects (model developers and enterprises) . Sovereign wealth funds and national governments are increasingly active participants across several of these categories.
Saudi Arabia announced more than $15 billion in new AI investments at LEAP 2025, including a $10 billion partnership between its Public Investment Fund and Google Cloud. The UAE is developing what it describes as the largest AI campus outside the United States — a 26 square kilometer facility in Abu Dhabi with 5 gigawatts of planned capacity. The European Union has launched a €200 billion AI Continent Action Plan, split between €50 billion in public funding and €150 billion from private sources. France alone announced $112 billion in AI-related private sector commitments .
Questions the Capital Doesn’t Answer
Hassan Taher has described AI as capable of addressing some of the most significant societal challenges, including climate change and sustainable resource management. That framing sits in some tension with the energy demands of the infrastructure being built. AI data centers pushed the U.S. power grid to notable limits in 2025. The facilities being built now, at the scale currently planned, will require substantial additions to energy generation capacity. Whether that energy comes from renewable sources will shape the environmental ledger of the AI buildout considerably.
The financial questions are also unresolved. Amazon CEO Andy Jassy has argued that AI capacity is being monetized as quickly as it is installed, pointing to AWS reaching a $142 billion annualized revenue run rate with growth accelerating to 24% year-over-year. But analyst concern persists. The overbuilt infrastructure risk that Satya Nadella flagged, the circular financing structures that blur real demand from manufactured revenue, and the debt levels accumulating across the sector represent risks that the current pace of investment has not yet tested under adverse conditions. Whether the $7 trillion projected to flow into AI infrastructure by 2030 produces returns commensurate with the commitment is the defining financial question of the decade — and one that organizations building AI strategy on top of this infrastructure have to take seriously.
#Funding #Trillion #Infrastructure