--- title: "Under the computing power shortage, \"clear accounting\": Google cuts off supplies to Meta, and Gemini's computing power is limited" type: "News" locale: "en" url: "https://longbridge.com/en/news/291074280.md" description: "Google's computational power bottleneck limits Meta's use of its Gemini AI model, forcing Meta to turn to its self-developed Muse Spark model and optimize resource usage. This move reflects the tight supply of AI infrastructure and the issue of power shortages, as global data center energy consumption surges, making energy supply the primary bottleneck constraining AI development" datetime: "2026-06-29T01:02:03.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/291074280.md) - [en](https://longbridge.com/en/news/291074280.md) - [zh-HK](https://longbridge.com/zh-HK/news/291074280.md) --- # Under the computing power shortage, "clear accounting": Google cuts off supplies to Meta, and Gemini's computing power is limited According to reports from foreign media, Google (GOOGL.US) has imposed restrictions on Meta Platforms Inc. (META.US) regarding the use of its Gemini artificial intelligence model, due to the inability to meet the scale of computing power required by the latter. This move once again reflects the supply bottleneck in AI infrastructure, as Google has implemented quota constraints on multiple clients, with Meta being particularly affected. Three informed sources stated that this restriction has caused a chain reaction within Meta's internal projects and forced the company to inform employees to use AI tokens more efficiently. Reports indicate that due to Gemini's performance surpassing its own Llama open-source model, Meta initially relied on Gemini to automate security review processes, such as removing harmful content and combating fraudulent information. However, Meta is increasingly using its new Muse Spark model in an effort to reduce reliance on external models. The AI boom is testing the global computing power limits at an unprecedented speed, while also consuming vast amounts of energy that support data center operations. According to data from the International Energy Agency (IEA), global data center electricity consumption is expected to reach approximately 485 billion kilowatt-hours in 2025, a year-on-year increase of 17%, with the United States accounting for about 224 billion kilowatt-hours; by 2026, the electricity consumption of global data centers, artificial intelligence, and cryptocurrency is expected to exceed 1,000 TWh. The electricity consumption of a single AI data center is comparable to that of 100,000 households, with some under-construction super-large facilities being 20 times that number. The shortage of electricity supply has become the primary bottleneck for AI development, and in the first half of 2025, 17 states in the U.S. have already implemented "power rationing orders" for data centers due to electricity shortages. Google Cloud CEO Thomas Kurian bluntly stated that energy issues have become "the most challenging matter." In response to this predicament, tech giants are significantly increasing their investments in AI infrastructure. Microsoft, Amazon, Google, and Meta are expected to collectively invest about $400 billion by 2025, with projections to exceed $600 billion to $725 billion in 2026. Google's parent company Alphabet has a capital expenditure plan of up to $175 billion to $185 billion in 2026, focusing on AI computing power and data center infrastructure. Nevertheless, the gap in computing power continues to widen—Google Cloud's backlog of contracts has exceeded $460 billion. Against this backdrop, in early June this year, Google reached a "sky-high" computing power deal with SpaceX. According to regulatory documents submitted by SpaceX to the U.S. Securities and Exchange Commission (SEC), Google has agreed to pay SpaceX $920 million per month from October 2026 to June 2029 to rent approximately 110,000 NVIDIA GPU clusters and related CPU, memory, and other supporting hardware. Based on the agreement period, the total contract amount is approximately $30 billion. The documents also disclose that if SpaceX fails to deliver the agreed number of GPUs by September 30, 2026, Google has the right to terminate the contract after providing a one-month grace period A spokesperson for Google Cloud stated in response that this transaction is "to ensure we have transitional computing power to meet the surge in demand from customers for the AI platform Gemini Enterprise, which has even exceeded our expectations." Google launched the Gemini Enterprise subscription service for large enterprise customers in October 2025. It is worth noting that this is not the first time SpaceX has rented out computing resources. Just last month, AI giant Anthropic reached an agreement with SpaceX to rent all the computing power of its "Colossus" data center located in Memphis, Tennessee, for $1.25 billion per month. The two orders combined generate over $2.1 billion in monthly revenue, with an annual recurring revenue (ARR) of approximately $26 billion. This series of transactions is widely seen by the market as paving the way for SpaceX's planned initial public offering (IPO) — the company is expected to go public on June 12, 2026, with a valuation of up to $1.77 trillion to $1.8 trillion. SpaceX is leading the computing power leasing business through its AI subsidiary xAI, aiming to position itself as a provider of AI computing infrastructure. Although Meta does not sell cloud services, it is prioritizing AI investment as CEO Mark Zuckerberg's top strategy, with the technology becoming central to its corporate vision. Earlier this year, the social media company disclosed plans to lay off 10% of its workforce (about 8,000 positions) to offset high expenses; at the same time, as part of a comprehensive restructuring, it has reassigned 7,000 employees to new AI-related positions ### Related Stocks - [META.US](https://longbridge.com/en/quote/META.US.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [GGLS.US](https://longbridge.com/en/quote/GGLS.US.md) - [METD.US](https://longbridge.com/en/quote/METD.US.md) - [GOOW.US](https://longbridge.com/en/quote/GOOW.US.md) - [GGLL.US](https://longbridge.com/en/quote/GGLL.US.md) - [METU.US](https://longbridge.com/en/quote/METU.US.md) - [FBL.US](https://longbridge.com/en/quote/FBL.US.md) - [METW.US](https://longbridge.com/en/quote/METW.US.md) - [FBY.US](https://longbridge.com/en/quote/FBY.US.md) - [FBYY.US](https://longbridge.com/en/quote/FBYY.US.md) - [MAGX.US](https://longbridge.com/en/quote/MAGX.US.md) - [GOOY.US](https://longbridge.com/en/quote/GOOY.US.md) - [GOU.US](https://longbridge.com/en/quote/GOU.US.md) - [GOOX.US](https://longbridge.com/en/quote/GOOX.US.md) - [GOOP.US](https://longbridge.com/en/quote/GOOP.US.md) - [MSFT.US](https://longbridge.com/en/quote/MSFT.US.md) - [AMZN.US](https://longbridge.com/en/quote/AMZN.US.md) - [SPCX.US](https://longbridge.com/en/quote/SPCX.US.md) ## Related News & Research - [Meta Faces Growing Pressure From Trump Administration To Submit Advanced AI Models For Federal Testing: Report](https://longbridge.com/en/news/290638185.md) - [Why is Google suddenly losing AI talent? The lure of pre-IPO equity is strong.](https://longbridge.com/en/news/290803263.md) - [Meta cut 8,000 jobs after its best quarter ever — months earlier, 6 executives got options worth up to $921 million each](https://longbridge.com/en/news/290591359.md) - [REI got roasted for a nonsensical Instagram ad featuring a bike with 2 sets of handlebars — and blamed a Meta AI tool](https://longbridge.com/en/news/290730677.md) - [Jefferies Says Google's AI Talent Losses Are Just 'Noise,' Here's Why](https://longbridge.com/en/news/290600282.md)