A reported five hundred billion dollar Nvidia partnership deal highlights a critical twenty year financing gap for large scale data center development.
According to Jonathan Mauck, senior managing director at Digital Bridge Holdings, speaking during a fireside interview at Datacloud USA x Metro Connect Fall 2026 in Austin, data centres are constructed as 20-year industrial assets while the GPUs housed within them typically have a useful lifespan of only five to seven years.
“You’re effectively a 20-year creditor,” Mauck said, outlining the risk operators face when committing capital to a facility under the assumption that a tenant will continue paying rent over two decades, despite the underlying hardware requiring replacement several times over.
In August, Nvidia announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to establish independent compute financing platforms. These structures aim to mobilise capital, with Nvidia potentially backing up to $125bn. The goal is to transform Nvidia compute into a distinct, investable asset class, separate from the civil and electrical infrastructure that houses it.
On the mechanics of the deal, Mauck remained measured. “That looks like a version of round tripping my capital,” he observed, framing the arrangement as Nvidia leveraging financial partners to effectively backstop GPU sales rather than guaranteeing buybacks directly. He argued these structures are best suited for neo-clouds and other operators lacking investment-grade credit ratings, rather than hyperscalers like Microsoft and Amazon, which already enjoy ready access to capital. A parallel example of this funding gap is evident in Nvidia’s reported discussions to guarantee up to $250bn in financing for OpenAI’s approximately $500bn Ohio data centre project.
Mauck also noted a structural shift as AI workloads transition from training to inference. While large training campuses continue to be developed, growth is increasingly concentrated in smaller 20 to 40MW facilities located in tier-two and tier-three markets closer to end users—a model he compared to the enterprise colocation sites of previous years.
This trend aligns with a broader globalisation of compute demand, driven by expansion across Latin America, Asia, and Europe. Much of this shift stems from capacity constraints and local opposition to new builds in North America, an issue Capacity has consistently tracked regarding community backlash, costs, power limitations, and regulatory pushback.
When asked whether AI infrastructure investment constitutes a bubble, Mauck contrasted the current landscape with the 2001 downturn, noting that today’s capital is underpinned by cash-generative businesses rather than speculative valuations alone. However, he cautioned that revenue must eventually catch up to the scale of deployed capital, warning that any hyperscaler scaling back on capital expenditure “would result in a pretty volatile year” across the sector.
Looking ahead, Mauck predicted the next major growth driver would be the “industrialisation of AI” through robotics and real-world applications, pointing to substantial investments from the United States, China, and Japan.