Google is quietly stockpiling its custom-designed Tensor Processing Units (TPUs) and purchasing additional third-party compute capacity to support its ambitious push towards Artificial General Intelligence (AGI), according to industry sources. The move underscores the escalating battle among Big Tech firms for the specialised hardware needed to train and run increasingly powerful AI models.
The company's TPUs, which are purpose-built chips for machine learning workloads, are being reserved for internal research projects rather than offered to cloud customers. At the same time, Google Cloud — already Alphabet's fastest-growing business segment, contributing more than a fifth of group revenue — is buying extra compute from external providers to meet surging demand for its G-Cloud infrastructure services.
For UK businesses, the implications are twofold. Smaller firms and startups that rely on Google Cloud's AI services may face higher costs and reduced access to cutting-edge hardware as the tech giant prioritises its own AGI ambitions. 'The risk is that UK companies become dependent on a handful of hyperscalers who are competing for the same scarce chips,' said Dr Eleanor Shaw, a technology policy researcher at the University of Cambridge. 'This could stifle innovation outside of Silicon Valley.'
Regulators are watching closely. The UK's Information Commissioner's Office (ICO) has already signalled concerns about the concentration of AI compute power, while the European Union's AI Act — which came into force in stages from 2025 — imposes strict transparency and risk-management requirements on foundation models. Google's AGI push raises questions about whether existing frameworks can adequately govern systems that may surpass human-level intelligence across multiple domains.
On the consumer front, UK households may eventually benefit from more capable AI assistants and services, but the short-term picture is less rosy. 'We're likely to see a two-tier market where only the largest players can afford the most advanced hardware,' warned tech analyst Ravi Patel of London-based consultancy Digital Futures. 'That could slow the development of homegrown British AI startups and increase our reliance on US tech giants.'
The broader economic impact is uncertain. While Google's investment in AI infrastructure could drive down costs for cloud services over time, the current scramble for chips is already pushing up prices for enterprise customers. The UK government's recent AI Safety Summit and ongoing work on a domestic AI regulatory framework will need to address these competitive dynamics if British businesses are to avoid being locked out of the next wave of AI innovation.