Deutsche Bank Weighs in on the $800bn AI and Data Centre Gap

Deutsche Bank responds to new research suggesting that the rapid expansion of AI faces a financial shortfall of US$800bn, driven by the vast infrastructure and data centre requirements needed to sustain projected revenue growth. 

New analysis from the bank reveals that continued growth in AI hinges not on software innovation but on massive capital spending across data centre capacity, computing hardware and power systems.

George Saravelos, Head of FX Research at Deutsche Bank, says: “AI machines – in quite a literal sense – appear to be saving the US economy right now.” He adds: “In the absence of tech-related spending, the US would be close to, or in, recession this year.”

AI growth depends on data centre scale

The analysis highlights that AI capital expenditure now plays a critical role in holding off a downturn in the US economy. 

Much of this investment is flowing directly into large-scale data centre builds and the infrastructure required to power and cool high-performance GPUs (graphics processing units) needed for AI training and inference.

New research from consulting firm Bain & Company warns that the AI sector faces a major challenge in meeting demand for compute capacity by the end of the decade. 

It estimates that the industry will need to generate US$2tn annually in revenue by 2030 to keep up with demand for computing infrastructure. Even after factoring in cost savings across sectors using AI, Bain calculates that the shortfall remains at US$800bn.

Nvidia’s US$100bn commitment to AI leader OpenAI illustrates the capital intensity of AI infrastructure investment. Nvidia supplies the GPUs at the heart of generative AI models, driverless vehicles and enterprise deployments. These chips are deployed in hyperscale data centres operated by cloud providers including Amazon Web Services (AWS), Microsoft Azure and Google Cloud.

Concentration of capital raises concerns

AI-related spending is now so central to the US economy that it has introduced new imbalances in equity markets. 

Deutsche Bank’s Head of Macro and Thematic Research, Jim Reid, draws attention to performance gaps between large-cap tech firms and the rest of the market.

“The S&P 500 is now up 13.81% so far this year, whereas the equal-weighted version is only up 7.65%.” Jim says. 

He adds “It’s been the Magnificent 7 driving the gains,” referring to Apple, Microsoft, Alphabet, Amazon, Meta, Tesla and Nvidia. The rest of the index – the remaining 493 companies – have shown much lower returns.

Torsten Sløk, Partner and Chief Economist at Apollo Management, notes “the upward consensus revision to 2026 earnings for the S&P 500 since Liberation Day comes entirely from the Magnificent 7.” Meanwhile, “earnings expectations for the S&P 493 have remained suppressed and are not moving higher.”

This raises questions about whether current valuation levels are sustainable if infrastructure investment slows down.

Capital spending, not AI software, is fuelling growth

Although AI promises long-term productivity gains, Deutsche Bank identifies infrastructure spending – not the technology itself – as the main source of economic activity at present. Companies are funnelling funds into expanding data centre real estate, procuring specialised hardware and scaling energy provision to support AI workloads.

George says: “growth is not coming from AI itself but from building the factories to generate AI capacity.” He warns that this trend may not last, warning that “in order for the tech cycle to continue contributing to GDP growth, capital investment needs to remain parabolic. This is highly unlikely.”

With capital expenditure on AI already reaching US$368bn through August, the future of AI’s economic role may rest on whether hyperscale investment can continue at this pace. Amazon, Microsoft and Google are leading this charge, but infrastructure projects of this scale demand vast energy input, physical space and hardware procurement.

George suggests that Nvidia now sits at the centre of the AI investment cycle. “It may not be an exaggeration to write that Nvidia – the key supplier of capital goods for the AI investment cycle – is currently carrying the weight of US economic growth,” he says.

While Goldman Sachs offers a more upbeat view, estimating that AI adoption will lift US GDP by 0.4% annually and 1.5% cumulatively, this projection assumes sustained and broad adoption of AI across the economy.

Goldman Sachs economist Manuel Abecasis explains: “We expect productivity gains from AI to boost GDP significantly, by about 0.4% through the next few years and 1.5% cumulatively as adoption rises over the long run.” He argues that “once it is widely adopted, AI is likely to allow workers and firms to produce more output for a given set of inputs, which will raise total factor productivity growth.”

However, the infrastructure question remains unresolved. Bain & Company concludes: “Two trillion dollars in annual revenue is what’s needed to fund computing power needed to meet anticipated AI demand by 2030. However, even with AI-related savings, the world is still US$800bn short to keep pace with demand.”