With a staggering market cap of $4 trillion and the largest weight in the S&P 500 (around 7.5%), Nvidia has become the face of the AI boom – and rightly so. Its GPUs are the computational engines powering most AI models and data centres today. While we’re impressed by its growth trajectory, we’re also mindful of the risks of concentrating exposure in a single name.
That’s why we look beyond Nvidia. In our Rathbone Greenbank Global Sustainability Fund, we’ve built our AI cluster around a multi-layered ecosystem – one that supports not only the digital revolution but also the transformation of physical infrastructure and industrial applications.
We break it down into three critical layers:
Layer 1: the silicon core
This layer is the foundation, these companies design and produce the chips and equipment, they are the starting point of the whole AI system.
Nvidia: GPUs, the brain of AI.
TSMC: The global leader in semiconductor manufacturing, from Nvidia to Apple, all rely on TSMC to deliver the cutting-edge chips.
Cadence: Electronic design software that helps make chips as fast and as small as possible – essential for the most complex chips and systems.
ASML, ASMI: Semiconductor equipment and tools, enabling TSMC and other manufacturers to produce chips. In particular, ASML holds a monopoly in the most advanced ‘extreme ultraviolet light’ equipment.
Amphenol: Cabling and connectors supplier, linking chips with other components and data centres to compute at scale.
Layer 2: the software stack
This layer translates computing to productivity and intelligence. They are the interfaces where AI gets real meaning for enterprises and consumers.
Microsoft: Outlook, Excel and more are revitalised through cloud computing service Azure, AI tool Copilot and enterprise-scale AI.
SAP: Chunky enterprise resource planning systems (which capture and analyse data from all parts of a business) get smarter and smoother with AI assistant SAP Joule, backed by the speed and scale of SAP HANA, a highly efficient analytics system.
Adobe: Gold standard design software unlocks more possibilities with Firefly and other generative AI tools for creativity and marketing.
Layer 3: the physical enablers
AI may live in the cloud, but it needs real-world infrastructure to run. Companies in this layer build and upgrade the physical systems that support AI at scale.
E.On, Schneider Electric, Quanta Services: These are the ‘power plays’ that benefit from surging energy demand driven by hyperscalers: E.On builds the energy networks that supply the electricity; Schneider Electric gets that power to where it’s needed; Quanta Services upgrades the grid network; and Accelleron makes turbochargers that provide back-up power for data centres.
Atlas Copco, Belimo, Veralto: These are the ‘hot and cool’ guys for data centres that handle the thermal and environmental requirements. Atlas Copco’s compressors are used within data centre cooling systems; Belimo’s extensive HVAC portfolio covers wide needs for heating, cooling and ventilation; and Veralto provides control and treatment for the water used in cooling.
Nvidia may be the poster child of AI, but it’s far from the whole story. Our long-term philosophy is to invest throughout the ecosystem. To build a robust and resilient portfolio, we need to understand the full stack, from fabricators to power lines. We believe this layered approach gives more sustainable returns through cycle, rather than chasing short-term hype.
