The Trillion-Dollar Build-Out
The numbers are staggering. According to recent analysis from McKinsey, the race to scale artificial intelligence has kicked off one of the largest infrastructure projects in modern history, with global spending on data centres potentially reaching $7
trillion by 2030. Another projection from the firm suggests that companies across the value chain will need to invest $5.2 trillion in AI-ready data centres by 2030 to meet demand. This spending isn't just about software; it's about the tangible, power-hungry hardware that makes large-scale AI possible. The technology story of 2026 has decisively moved from the abstract world of code into the physical realm of power grids, construction, and advanced manufacturing. In 2026 alone, investment in AI infrastructure and model architectures is on track to hit approximately $769 billion, a more than fivefold increase from the $145 billion spent in 2025. This massive capital flow underscores a fundamental truth: AI is an industrial-scale endeavour.
Data Centres: The New AI Factories
Forget the old image of data centres as quiet digital warehouses. The facilities required for AI are more like high-performance factories. AI workloads, particularly for training advanced models, demand immense power and specialized cooling systems that far exceed traditional computing needs. A standard data centre rack from a decade ago might have consumed 2-4 kilowatts of power; a modern AI rack can demand as much as 140 kilowatts, with future generations pushing toward a megawatt. This leap in power density is forcing a complete redesign of data centre architecture, with a focus on liquid cooling and ultra-efficient power distribution. Consequently, the cost of building these facilities has soared. While a standard data centre might cost $10-13 million per megawatt, an AI-optimized build can run from $15 million to over $20 million per megawatt. The sheer demand for this new type of infrastructure is projected to nearly triple global data centre capacity between 2025 and 2030, with AI-related demand growing 3.5 times to account for around 70% of the total.
Chips: The Brains Behind the Boom
At the heart of every AI data centre are the specialized chips that perform the complex calculations. Graphics Processing Units (GPUs), dominated by companies like Nvidia, have become the default hardware for training and running AI models. This has created an insatiable demand, turning the AI chip market into a foundational pillar of the infrastructure boom. The market for AI chips used in data centres is projected to explode from $124 billion in 2024 to $860 billion by 2030, representing an annual growth rate of 38%. This surge is so significant that McKinsey recently revised its forecast for the entire semiconductor market, attributing the upgrade largely to AI data centre demand. While GPUs are the current leaders, the industry is diversifying. Major tech companies like Google, Amazon, and Microsoft are developing their own custom chips, known as Application-Specific Integrated Circuits (ASICs), to optimize performance and reduce reliance on single suppliers. These chips are designed for specific tasks, offering greater efficiency for AI inference—the process of running a trained model—which is rapidly becoming the dominant workload and cost driver in the AI ecosystem.
The Ripple Effect on the Economy
This massive infrastructure investment creates a powerful ripple effect across the global economy. The primary beneficiaries are not just the chip designers and tech giants. The build-out requires enormous capital, benefiting the financial sector. It also drives demand for industrial players who manufacture everything from electrical switchgear to advanced cooling systems. Furthermore, the construction of these massive facilities creates thousands of jobs, and their operation requires a new workforce of skilled technicians to manage the complex hardware. The availability of affordable, reliable power has become the single most critical factor determining where these multi-billion dollar investments are made, elevating the importance of energy providers and utility infrastructure. As nations and corporations race to secure their place in the AI future, the development of this physical backbone—the data centres and the chips inside them—is proving to be the main event, shaping economic competitiveness for decades to come.
