Google to Restrict Gemini Supply, Notifies Major Companies Including Meta Platforms [Global AI Briefing]

[Financial News] Google has declared restrictions on the supply of its artificial intelligence model Gemini. The company has been pouring huge sums into securing AI computing resources, but with demand still outpacing supply, it has decided to limit access.
As AI usage surges, shortages in AI infrastructure are becoming a reality, and a fierce global race for AI computing resources is expected to intensify.
On the 29th local time, major foreign media outlets including the Financial Times and Engadget reported that Google informed Meta Platforms in March that it could not provide all of the computing capacity requested. The move delayed some of Meta Platforms' internal AI projects, and employees were also told to conserve AI token usage. Google is also said to have notified other customers of limits on Gemini capacity.

Meta Platforms does not operate its own cloud infrastructure. As a result, it has relied on external AI models for key tasks across Facebook and Instagram, including customer service chatbots, AI services for advertisers, software coding, harmful-content monitoring, and fraud detection. It has made Gemini its main model because it outperforms its own AI model Llama, while also using Anthropic's Claude in parallel.
After its major projects were disrupted by Google's restrictions on Gemini use, Meta Platforms is reportedly working on ways to reduce its dependence on external AI, including expanding the use of its in-house AI model Muse Spark.
Experts said Google's limits on AI usage are a sign that shortages in AI infrastructure are becoming a reality.
Big tech companies are pouring massive amounts of money into semiconductors, data centers, and power supplies, but they still have not secured enough computing capacity to handle the rapidly growing demand for AI model training and inference.
To meet the computing needs of Google Gemini Enterprise services, Google is said to be nearly doubling its planned spending this year to a total of $180 billion to $190 billion, or about 292 trillion won. It is also making every effort to secure infrastructure, including leasing AI data centers from SpaceX and xAI, but has still been unable to keep up with demand and has resorted to restricting supply.
One industry expert explained that Google's restrictions on Gemini use clearly show how competition in the AI industry is expanding beyond model development and into a race to secure infrastructure.
cafe9@fnnews.com Lee Gu-soon Reporter