Multiple media outlets cited reports from the Financial Times stating that against the backdrop of the global generative AI boom around 2023 increasing demand for high-performance GPUs and related infrastructure, and the supply of computing power being insufficient, Google has imposed restrictions on Meta's use of its Gemini AI model. The reports pointed out that the demand scale for computing power and services related to Gemini proposed by Meta exceeds what Google can currently provide, which directly triggered this "traffic limit," also exposing the tension in resource allocation between the two companies in their AI collaboration relationship. Given that the cloud computing market is highly concentrated among a few large tech companies, Google and Meta are both competitors in the AI field while also being interdependent in certain aspects. This incident is seen as a typical case: when infrastructure is scarce and demand continues to expand, the party that controls computing power will prioritize resources for its own business and key ecosystem partners. From an investment perspective, this news reinforces two key signals: first, the shortage of AI computing power remains a medium-to-long-term mainline issue, and the bargaining power and value re-evaluation logic of infrastructure assets have not weakened; second, under the framework of relying on competitors for key computing power, the competition among major companies accelerates, and the risk premium concentrated on leading firms will continue to be the core pricing variable for AI and computing power-related assets in both traditional capital markets and crypto markets, and all of this is still primarily based on media disclosures, and neither party has provided more detailed public statements.
Power Shortage Intensifies: Even Meta Faces Throttling
From the supply side, even a tech giant like Meta cannot bypass the "power allocation system" in the current cycle. Since 2023, reports of shortages in high-performance GPUs (such as NVIDIA's high-end chips) have been prevalent throughout this round of the AI boom, with hardware delivery, data center transformation, electricity, and cooling all facing cyclical and supply chain constraints, leading to the expansion rate of AI infrastructure being naturally slower than the explosive growth in demand for large models. In this structural contradiction, the cloud computing market is highly concentrated among a few large tech companies, and large cloud vendors will prioritize the occupation of core GPUs and network resources when training and deploying their own large models, leaving only the "remaining capacity" framework for external customers.
Multiple media outlets, citing the Financial Times, reported that Google's restrictions on Meta's use of Gemini are directly due to the supply-demand imbalance: Meta's proposed demand for computing power and services related to Gemini has been internally assessed by Google to exceed what it can provide. For cloud service providers like Google, when high-performance GPUs become scarce resources, they can only manage the allocation among multiple large customers through quotas, scheduling, and prioritized queues: on the one hand, they must ensure the launch pace of their own models and products, and on the other hand, they need to balance resources and business relationships among partners. In such a power shortage environment, even Meta has to accept the reality of being included in a unified scheduling and throttling management system, which itself indicates that the constraints of AI infrastructure remain the toughest boundary in the current competition among large models.
Cooperation and Competition: The Subtle Race Between Google and Meta
In the AI landscape, the relationship between Google and Meta is essentially a long-term race of "fighting while buying." On one hand is Meta, heavily betting on the open-source large model route and continuously investing in the Llama series; on the other hand is Google, fully building the Gemini product and ecosystem. The two companies directly collide in advertising, content distribution, and AI foundational models, but at the level of computing power and infrastructure, it has been reported that Meta has actively sought Gemini-related computing power and services from Google, which means that competitors must share the most critical production resources on the same cloud pipeline.
In light of the continuing tight supply of computing power, Google's prioritization of the Gemini ecosystem is essentially written at the top of resource allocation: first ensuring the training of its own models, product iterations, and internal business, then opening capabilities for external customers—including potential direct competitors like Meta—within the remaining capacity. Media reports have mentioned that Meta's demand for Gemini computing power and services has exceeded what Google can supply, and the throttling in this structure is more like a result of prioritization rather than a mere technical failure. For Meta and even other large companies, entrusting critical AI infrastructure to competitors means that once market conditions or the competitors' strategies change, they may face tightening quotas, delayed schedules, or even systematic risks regarding prices and bargaining power; these latent constraints will, in turn, shape their long-term strategic choices regarding building their own infrastructure, diversifying cloud vendors, and pushing forward with self-research large models.
Cloud Bottlenecks Overflow: On-Chain AI Computing Power Narrative Heats Up Again
When Google sets limits on Meta's use of Gemini computing power against the backdrop of supply shortages, what the market interprets is not just business friction between the two companies but the structural tension at the level of AI infrastructure where "resources are locked up by a few cloud giants." Industry analysts generally point out that global cloud computing power is highly concentrated among a few large tech companies. When these companies reallocate resources among their own large models, internal customers, and external partners, external demand-side players are easily forced to accept throttling, queuing, and weaker bargaining positions. This reality is quickly extrapolated into the macro narrative that "cloud computing power is a scarce public good."
Within this narrative, projects like distributed GPU networks, AI inference networks, and data labeling networks that have emerged in the crypto market over the past two years are being repackaged as alternative paths "to bypass the centralized cloud resource bottlenecks." They generally claim that by incentivizing tokens to mobilize idle computing power and scheduling resources on open networks, they can alleviate the tightness and overly strong bargaining power of centralized cloud computing to some extent; whenever news of "power shortages" or throttling by major firms like Google comes up, AI and computing power-related crypto assets are often revisited and hyped by the market, leading on-chain decentralized computing power to continually be integrated into the logical chain of "long-term scarcity of computing power, value re-evaluation of infrastructure." How far this on-chain narrative, starting from cloud bottlenecks, can go depends on their ability to scale and support real AI workloads.
How Capital Interprets This AI Computing Power Squeeze
From an investor's perspective, the statement "Meta's demand exceeds what Google can provide" sends the most direct signal: following a continuous increase in capital expenditure since 2023, computing power is still one of the most tightly constrained resources in the AI era. The traditional capital market has already treated cloud vendors' AI capital expenditure and computing power utilization rates in their financial reports as core indicators. Such throttling events will be interpreted as—new GPUs and infrastructure can still be quickly filled, and the long-term supply of AI infrastructure is still tight, rather than being seen as an isolated dispute with a specific customer. At the same time, Google's limits on competitor clients during periods of computing power shortages will also reinforce the perception that "cloud computing power is highly concentrated, and infrastructure providers have bargaining and resource allocation power," leading capital to reassess the safety margins and competitive positions of different large model camps on the infrastructure side.
In the crypto market, macro news like "computing power shortages" and "large model expansions" often synchronously accompany price fluctuations in the AI and computing power sectors, seen as emotional catalysts validating the "long-term scarcity of infrastructure." For investors betting on the AI + crypto narrative, key variables entail not only assessing the pace of large models and AI application expansions on the demand side but also examining whether projects can provide real, sustainable computing power supply on the supply side, and whether their economic models can support long-term operations. Against a backdrop where computing power has been repeatedly proven to be a scarce resource, whoever can turn distributed computing power from narrative into a sustainable infrastructure supporting real AI workloads is likely to obtain a more stable premium in capital terms.
From Gemini Throttling to AI Infrastructure Reevaluation
Looking back at the controversy surrounding Google's throttling of Meta's Gemini, the core conflict does not lie in a single contract or individual product, but in the fact that after the global generative AI boom around 2023 significantly increased demand, the supply of high-performance GPUs and related infrastructure has consistently lagged behind the rising computing power demands of large companies. Multiple media outlets cited the Financial Times reporting that Meta's demand for Gemini-related computing power and services exceeded what Google could provide, with this specific conflict placed within the broader context of "insufficient power supply," highlighting that when cloud computing power is highly concentrated among a few large tech companies, the interdependence and interactions among competitors regarding key infrastructure are almost inevitable. Google and Meta both cooperate and compete in the AI field, compounded by the scarcity of infrastructure, making such cooperation more interchangeable and tactical. Future similar computing power conflicts will not disappear simply because of a throttling episode. For the crypto market, AI and computing power-related projects will continue to frame these incidents with narratives like "alleviating centralized cloud computing power bottlenecks" and "supporting traditional clouds, sharing pressures," and market sentiment will maintain high sensitivity to macro news like "computing power shortages" and "large model expansions." However, investment and industry analysis need to place single events back within the long-term supply-demand structure, acknowledging that the foundational logic of the AI + computing power narrative remains valid while compressing the emotional amplification effect of single news events on price and valuation.
Join our community, let's discuss and become stronger together!
AiCoin exclusive Hyperliquid benefits: https://app.hyperliquid.xyz/join/AICOIN88
AiCoin exclusive Aster benefits: https://www.asterdex.com/zh-CN/referral/9C50e2
On-chain Telegram community: https://t.me/AiCoinWhaleData
On-chain community: https://www.aicoin.com/link/chat?cid=N6OVMor5g
AiCoin on-chain Twitter: https://x.com/aicoinwhaledata
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
