“The race to build more powerful artificial intelligence is no longer just about better software. Even the world’s biggest technology companies are running into the same problem: there simply isn’t enough computing power to go around.”
Google has placed limits on Meta’s use of its Gemini artificial intelligence models after the Facebook and Instagram owner requested more computing capacity than Google could provide, according to a Financial Times report.
The restrictions have reportedly been in place since March and have disrupted some of Meta’s internal AI work, forcing teams to delay projects and use available computing resources more carefully. Employees have also been encouraged to reduce unnecessary AI usage by managing their token consumption more efficiently.
The development offers a rare glimpse into the growing pressure facing even the largest companies in the AI industry.
Although firms such as Google, Meta, Microsoft and OpenAI continue to spend billions of dollars on data centres and specialised AI chips, demand for computing power has grown even faster. That has left providers struggling to satisfy customers who want to train and run increasingly sophisticated AI models.
Meta has been one of Google’s biggest customers for Gemini’s cloud-based AI services.
According to the report, the company relied on Google’s models for tasks including software development, customer support and safety systems while continuing to build its own in-house AI technology. The shortage has pushed Meta to rely more heavily on its internal Muse Spark model as it looks to reduce dependence on outside providers.
Neither company commented publicly on the report.
SEE ALSO: Mobavenue CEO Ishank Joshi on why Google and Meta are not the enemy
The computing squeeze is affecting more than just Meta.
Google has also had to ration access for other customers, although people familiar with the matter said Meta felt the impact more because of the unusually large amount of computing capacity it had requested.
The shortage comes despite Google’s continued investment in expanding its AI infrastructure.
Earlier this year, Google Chief Executive Sundar Pichai acknowledged that strong demand was limiting the growth of its cloud business, saying the company was working to add more capacity as quickly as possible. More recently, Google agreed a major data-centre capacity deal with SpaceX to help meet rising demand for AI services.
Meta is making similar investments.
The company has committed hundreds of billions of dollars to AI infrastructure over the next few years, betting that artificial intelligence will become central to its future products and business. But the latest report shows that money alone cannot solve every problem overnight.
For now, the biggest challenge is not creating new AI models.
It is finding enough computing power to keep them running.
As the competition between technology giants intensifies, access to chips, servers and data-centre capacity is becoming almost as valuable as the AI models themselves. Companies with the most computing resources may ultimately have the greatest advantage in the next stage of the artificial intelligence race.
