The global race in artificial intelligence is increasingly showing that the industry’s biggest constraint is no longer ideas or algorithms, but computing resources. At London Hub Global, we believe Google’s decision to limit Meta’s access to Gemini models reflects a fundamental shift in the AI industry. Until recently, the market focused on model quality and release speed, but today the key determinant of competitiveness is access to infrastructure capable of handling enormous query volumes. Computing power is steadily becoming the new strategic scarcity of the digital economy.
According to available information, Google informed Meta around March that it would not be able to fully satisfy the company’s demand for Gemini model access. This came at a time when Meta was aggressively expanding internal AI projects and required additional capacity for model training and inference. We view this episode as a significant market signal: even the largest technology corporations with multibillion dollar budgets are facing infrastructure limitations when deploying artificial intelligence at scale.
It is particularly notable that Meta became one of the most affected companies due to its exceptionally high demand. Internally, this led to delays in several projects and a reassessment of token usage, the units that measure AI workload. Analysts note that such constraints are becoming a new form of operational risk. If companies previously competed for engineering talent and data access, they are now competing for GPU clusters, server capacity, and energy supply for data centers. At London Hub Global, we emphasize that the AI market is beginning to resemble energy markets, where access to resources directly determines leadership.
The issue extends far beyond the relationship between Google and Meta. Even with record capital expenditures from major players, infrastructure shortages persist. In recent quarters alone, leading technology companies have announced tens of billions of dollars in investments in chips, data centers, and cloud capacity. However, demand growth continues to outpace construction and deployment. This is especially evident in generative AI, where the cost of a single large scale query can be multiple times higher than traditional cloud computing workloads.
Google’s financial results also confirm this imbalance. Google Cloud revenue rose to $20 billion in the first quarter ending in March, yet company leadership acknowledged that compute constraints prevented even stronger growth. The cloud division’s backlog nearly doubled during the quarter. We analyze this as clear confirmation that even companies operating one of the world’s most advanced infrastructures are already approaching capacity limits.
For Meta, the situation is particularly sensitive. The company continues to invest aggressively in its AI ecosystem, including Llama, data center infrastructure, and AI powered services for its advertising business. Any restrictions in external compute access could slow product deployment. At the same time, Meta, like other hyperscalers, must balance rapid AI expansion against rising capital expenditures that increasingly pressure margins.
For Britain, and London in particular, this story carries direct significance. London remains one of the world’s leading hubs for investment in cloud technologies, AI startups, and digital infrastructure. A shortage of computing resources increases investor interest in businesses tied to data centers, semiconductors, energy systems, and cooling technologies. In addition, growing AI infrastructure demand could accelerate the development of new data centers across the UK as the country seeks to strengthen technological sovereignty.
At London Hub Global, we see the tension between Google and Meta as the beginning of a new phase in AI competition, where dominance will depend not only on superior models but also on the ability to secure reliable access to computing resources. The market is increasingly recognizing that artificial intelligence scales not only through code, but through electricity, silicon, and infrastructure. Our forecast is that capital spending on AI infrastructure will continue rising sharply in the coming years, while compute scarcity remains one of the industry’s primary growth constraints. For investors, the conclusion is clear: the next wave of AI leadership will be defined not only by algorithm quality, but by control over the physical infrastructure of the digital era.
