Investment in AI infrastructure and model architectures is projected to more than quintuple this year, even as energy, talent, capital and safety constraints raise questions about the pace of development.
Investment in artificial intelligence infrastructure and model architectures is on track to surge to around $769 billion in 2026, more than five times the $145 billion invested in 2025, according to McKinsey’s Technology Trends Outlook 2026, as the global AI buildout accelerates.
The estimate is based on nearly $384 billion invested in the first half of 2026, assuming the current pace continues through the year. AI infrastructure and model architectures are emerging as the most heavily funded technology trend tracked by McKinsey.
The investment boom is being driven by growing demand for the physical infrastructure needed to scale AI, including semiconductors, data centres and power systems.
Major AI companies and investors are also committing substantial capital to the sector. OpenAI is in discussions for additional funding at a potential valuation of around $1.2 trillion, while SoftBank has launched an $11 billion bond offering to support a further $10 billion investment in OpenAI. Anthropic is also in discussions with Nvidia over a potential investment of up to $10 billion, according to Reuters.
AI spending meets growing safety concerns
The rapid flow of capital comes as concerns grow within the industry about whether safety measures and governance frameworks can keep pace with increasingly capable AI systems.
More from BusinessAnthropic CEO Dario Amodei has called for a slower rollout of new AI capabilities to allow more time to address safety risks. OpenAI has also backed mandatory national AI safety requirements, including independent assessments, cybersecurity measures and incident reporting for advanced AI systems.
McKinsey identified five technology trends that are on track to attract more than twice as much investment in 2026 as in 2025: agentic software development, AI infrastructure and model architectures, AI for scientific discovery and engineering, space technologies and robotics.
Energy could become a key constraint
The AI infrastructure boom is also intensifying pressure on energy systems. McKinsey estimates that US data centres running AI workloads could consume as much electricity by 2030 as California uses today.
Globally, more than 2,500 gigawatts of energy projects are waiting for grid connections, highlighting the infrastructure bottlenecks that could constrain further AI expansion.
AI adoption is also spreading across businesses, with 89 per cent of organisations regularly using AI in at least one business function. However, only 37 per cent report a positive impact on EBIT, pointing to a gap between AI adoption and measurable financial returns.
The next phase of AI investment is therefore expected to focus not only on building bigger models and infrastructure but also on integrating AI into business workflows while managing energy, cybersecurity, governance and safety risks.
SEO keywords:
