By Stuart Kerr, Technology Correspondent, LiveAIWire

The AI arms race Google Amazon Meta are waging is no longer speculative, it is fully underway. In 2025 alone, the three tech giants are on track to spend nearly half a trillion dollars collectively to dominate the AI frontier. These investments go far beyond headline-grabbing models, they extend into the hidden infrastructure, data centres, semiconductors, and green energy solutions, that powers AI’s exponential growth.

According to The Guardian, Amazon is expected to exceed $100 billion in infrastructure spend this year, with Google committing $85 billion and Meta forecasting $64 to $72 billion. These are not vague projections. They reflect real capital deployed across chips, networking, and sovereign cloud expansion. The stakes are ownership of the foundations of next-generation intelligence.

Infrastructure as Power in the AI Arms Race Google Amazon Meta Are Fighting

These expenditures are not simply about faster AI models, they are about control. Vertically integrated infrastructure gives these firms unmatched leverage, both in latency and user lock-in, a dynamic LiveAIWire has also tracked in our comparison of Amazon, Google, and Meta’s AI infrastructure spending strategies in more granular detail.

Investor’s Business Daily highlights the cascading benefits for suppliers like Vertiv and Applied Digital, riding high on massive AI data centre buildouts. The industry’s 2025 capex now exceeds $381 billion, and the majority of it flows through just a handful of hyperscalers.

In parallel, Reuters reports investors are rewarding this spending frenzy with record-breaking stock valuations, proof that Wall Street now sees infrastructure dominance as the key to long-term AI profit.

Energy, Chips, and Sovereignty

These investments are also reactive. Google and Amazon are racing to power their data centres with sustainable energy sources, while Meta explores chip customisation to improve efficiency, a strategy that connects directly to LiveAIWire’s coverage of Google’s nuclear bet on advanced reactors to secure round-the-clock clean power for its own data centre expansion.

A new Intel and IDC report, AI Infrastructure in 2025, adds detail to this strategy: hybrid architectures blending data centre and cloud deployments will become standard, with chip development tied directly to algorithmic demand curves. It is not just about building AI, it is about building the machines that run it better, faster, and cheaper than anyone else.

The Geopolitical Layer

Behind the spending lies a more strategic layer. The competition in the AI arms race Google Amazon Meta are running is not just commercial, it is geopolitical. Nations are quietly treating AI infrastructure as a national security asset.

The arXiv paper Compute at Scale reveals the extent of this shift: governments, especially in Europe and Asia, are pressuring tech firms to develop localised infrastructure for sovereignty reasons. The implication is that whoever controls compute controls capability, and influence, a dynamic LiveAIWire has examined directly in our coverage of AI diplomacy and how nations are using code as a new form of soft power.

What Comes Next

If the AI arms race Google Amazon Meta are running continues at this pace, we may see consolidation not just of platforms, but of knowledge and power. Google’s Gemini 2.5, Amazon’s Bedrock, and Meta’s Llama models are not just products, they are ecosystems growing inside fortified digital infrastructure.

As the global economy restructures itself around artificial intelligence, the AI arms race Google Amazon Meta are locked in will define the winners. Smaller AI labs and national champions will increasingly find themselves priced out of frontier-scale training, left to build on top of infrastructure the three giants control rather than compete with it directly. And it will not be fought in press releases, but in concrete, silicon, and electrons.

About the Author

Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, cybersecurity, and the social impact of emerging technology. He publishes daily at LiveAIWire.com.