Nuway Capital’s latest strategic pivot offers a telling snapshot of how the artificial-intelligence boom is maturing. The UK-based alternative investment and infrastructure advisory firm is expanding its focus on graphics processing units (GPUs) and data-centre infrastructure, following the joint publication of new research with KPMG.
The move reflects a broader reassessment among investors: in the race to profit from AI, the scarce resource is no longer clever code, but the physical machinery that makes computation possible.
For much of the past decade, capital chasing digital megatrends, from cloud computing and social media to gaming and blockchain, has gravitated towards software champions and platform economics. Artificial intelligence initially followed the same pattern. Yet as models have grown larger and more power-hungry, attention has shifted down the value chain. The latest report from Nuway and KPMG highlights a widening structural gap between AI-driven demand for computing capacity and the availability of the underlying infrastructure required to supply it.
How fast is global data-centre capacity expanding?
According to the UK House of Commons Library, global data-centre capacity is forecast to expand by around 15% a year through to 2027. That sounds impressive, until it is set against demand growth driven by AI training, inference and cloud adoption, which is rising far faster. At the same time,
GPU supply remains constrained by limited semiconductor manufacturing capacity, geopolitical fragility in global supply chains, and acute bottlenecks in power, land and grid connectivity. The result is an imbalance that is increasingly shaping investment behaviour.
- Permutable AI provides insights into how markets consume news
- NVIDIA, China and the great AI convergence
- How AI shortages will shape financial markets in 2026
“Demand for GPUs is no longer episodic, it’s structural,” says Colin Bosher, founder of Nuway Capital. “The economics are becoming more infrastructure-like: sustained demand, physical bottlenecks, and a growing emphasis on operational efficiency and asset selection.”
In other words, GPUs and data centres are beginning to look less like speculative technology plays and more like regulated utilities with attractive scarcity premiums.
Survey data from KPMG supports this shift. More than 70% of high-net-worth individuals, and over 60% of family offices and wealth managers, now rank GPUs as a more attractive investment opportunity than other emerging technologies, including blockchain, quantum computing and even renewable energy. GPUs offer something those sectors often lack: indispensable, real-world utility tied directly to the digital transformation of the global economy.
The financial performance of leading producers helps explain the enthusiasm. Sustained GPU scarcity has boosted margins and valuations across the sector, with NVIDIA NASDAQ:NVDA now boasting a market capitalisation in the trillions, becoming a bellwether for broader equity markets.
Yet the competitive landscape is far from settled. Rivalry is intensifying as alternative architectures, such as Google’s AI-native tensor processing units (TPUs), edge closer to wider commercial availability. Speculation that Meta may lease TPUs from Google has already been enough to unsettle investors in incumbent suppliers, underscoring how sensitive the market remains to shifts in supply.
Hardware does not tell the full story
Still, hardware alone does not tell the full story. Even where chips are available, deploying them at scale is increasingly difficult. Suitable real estate for hyperscale data centres is scarce; grid capacity is constrained in many technology hubs; and environmental concerns are prompting tighter regulation of energy and water usage, particularly in Europe and America.
Add long development cycles, complex cooling requirements, skills shortages and rising financing costs, and it becomes clear why new capacity struggles to keep pace.
It is against this backdrop that Nuway Capital is expanding its advisory and platform capabilities around GPU capacity and high-performance computing infrastructure, including GPU-as-a-Service models. These structures lower the barrier for companies unable to finance dedicated infrastructure, while offering investors exposure to recurring, usage-based revenues, an appealing blend of growth and predictability.
Chris Brown, partner and head of KPMG Strategy in Ireland, captures the broader implication. “The next phase of artificial intelligence will be shaped less by software innovation alone and more by access to physical infrastructure,” he argues. As GPUs, data centres and power availability become the binding constraints, investors are recalibrating how they assess risk, return and long-term value across the AI ecosystem.
The conclusion is a sobering one for software purists. In the AI gold rush, it is not just the prospectors who profit. Increasingly, it is the owners of the picks, shovels (and power stations) who hold the upper hand.
Comments (0)