In the first half of 2026, venture capital firms invested $127 billion in AI-related startups globally — more than the total for all of 2023 and 2024 combined. The funding winter that followed the 2021-2022 boom has given way to something that looks, to veterans of previous cycles, uncomfortably like a new bubble.
The deals are getting larger, the valuations more stretched, and the due diligence — by some accounts — more cursory. A cohort of AI infrastructure companies that were valued at $500 million 18 months ago are now raising at $5 billion. The justification, in most cases, is the same: the total addressable market is so large that even a small share of it justifies almost any valuation.
The Infrastructure Layer
The most intense competition for capital is at the infrastructure layer: the companies building the tools, platforms, and services that other AI companies depend on. Vector databases, model fine-tuning platforms, AI observability tools, and inference optimisation services are attracting capital at a pace that suggests investors believe the infrastructure layer will be as valuable as the application layer — or more so.
Alexandra Whitfield
Technology Editor
Alexandra Whitfield is The Business Magazin's Technology Editor, covering semiconductors, AI infrastructure, and the geopolitics of tech.
